msvcrt: Add _ismbslead_l implementation.
[wine.git] / dlls / msvcrt / mbcs.c
blob0e6e0bdcc7f90cb80fbe14518d11938aa408fcd5
1 /*
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
21 * FIXME
22 * Not currently binary compatible with win32. MSVCRT_mbctype must be
23 * populated correctly and the ismb* functions should reference it.
26 #include <stdio.h>
27 #include <limits.h>
28 #include <mbctype.h>
29 #include <mbstring.h>
31 #include "msvcrt.h"
32 #include "mtdll.h"
33 #include "winnls.h"
34 #include "wine/debug.h"
36 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
38 unsigned char MSVCRT_mbctype[257] = { 0 };
40 /* It seems that the data about valid trail bytes is not available from kernel32
41 * so we have to store is here. The format is the same as for lead bytes in CPINFO */
42 struct cp_extra_info_t
44 int cp;
45 BYTE TrailBytes[MAX_LEADBYTES];
48 static struct cp_extra_info_t g_cpextrainfo[] =
50 {932, {0x40, 0x7e, 0x80, 0xfc, 0, 0}},
51 {936, {0x40, 0xfe, 0, 0}},
52 {949, {0x41, 0xfe, 0, 0}},
53 {950, {0x40, 0x7e, 0xa1, 0xfe, 0, 0}},
54 {1361, {0x31, 0x7e, 0x81, 0xfe, 0, 0}},
55 {20932, {1, 255, 0, 0}}, /* seems to give different results on different systems */
56 {0, {1, 255, 0, 0}} /* match all with FIXME */
59 /* Maps cp932 single byte character to multi byte character */
60 static const unsigned char mbbtombc_932[] = {
61 0x40,0x49,0x68,0x94,0x90,0x93,0x95,0x66,0x69,0x6a,0x96,0x7b,0x43,0x7c,0x44,0x5e,
62 0x4f,0x50,0x51,0x52,0x53,0x54,0x55,0x56,0x57,0x58,0x46,0x47,0x83,0x81,0x84,0x48,
63 0x97,0x60,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,
64 0x6f,0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x6d,0x8f,0x6e,0x4f,0x76,
65 0x77,0x78,0x79,0x6d,0x8f,0x6e,0x4f,0x51,0x65,0x81,0x82,0x83,0x84,0x85,0x86,0x87,
66 0x88,0x89,0x8a,0x8b,0x8c,0x8d,0x8e,0x8f,0x90,0x91,0x92,0x93,0x94,0x95,0x50,
67 0x42,0x75,0x76,0x41,0x45,0x92,0x40,0x42,0x44,0x46,0x48,0x83,0x85,0x87,0x62,
68 0x5b,0x41,0x43,0x45,0x47,0x49,0x4a,0x4c,0x4e,0x50,0x52,0x54,0x56,0x58,0x5a,0x5c,
69 0x5e,0x60,0x63,0x65,0x67,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,0x71,0x74,0x77,0x7a,0x7d,
70 0x7e,0x80,0x81,0x82,0x84,0x86,0x88,0x89,0x8a,0x8b,0x8c,0x8d,0x8f,0x93,0x4a,0x4b };
72 /* Maps multibyte cp932 punctuation marks to single byte equivalents */
73 static const unsigned char mbctombb_932_punct[] = {
74 0x20,0xa4,0xa1,0x2c,0x2e,0xa5,0x3a,0x3b,0x3f,0x21,0xde,0xdf,0x00,0x00,0x00,0x5e,
75 0x7e,0x5f,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0xb0,0x00,0x00,0x2f,0x00,
76 0x00,0x00,0x7c,0x00,0x00,0x60,0x27,0x00,0x22,0x28,0x29,0x00,0x00,0x5b,0x5d,0x7b,
77 0x7d,0x00,0x00,0x00,0x00,0xa2,0xa3,0x00,0x00,0x00,0x00,0x2b,0x2d,0x00,0x00,0x00,
78 0x00,0x3d,0x00,0x3c,0x3e,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x5c,
79 0x24,0x00,0x00,0x25,0x23,0x26,0x2a,0x40};
81 /* Maps multibyte cp932 hiragana/katakana to single-byte equivalents */
82 static const unsigned char mbctombb_932_kana[] = {
83 0xa7,0xb1,0xa8,0xb2,0xa9,0xb3,0xaa,0xb4,0xab,0xb5,0xb6,0xb6,0xb7,0xb7,0xb8,0xb8,
84 0xb9,0xb9,0xba,0xba,0xbb,0xbb,0xbc,0xbc,0xbd,0xbd,0xbe,0xbe,0xbf,0xbf,0xc0,0xc0,
85 0xc1,0xc1,0xaf,0xc2,0xc2,0xc3,0xc3,0xc4,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xca,
86 0xca,0xcb,0xcb,0xcb,0xcc,0xcc,0xcc,0xcd,0xcd,0xcd,0xce,0xce,0xce,0xcf,0xd0,0xd1,
87 0xd2,0xd3,0xac,0xd4,0xad,0xd5,0xae,0xd6,0xd7,0xd8,0xd9,0xda,0xdb,0xdc,0xdc,0xb2,
88 0xb4,0xa6,0xdd,0xb3,0xb6,0xb9};
90 static wchar_t msvcrt_mbc_to_wc_l(unsigned int ch, _locale_t locale)
92 pthreadmbcinfo mbcinfo;
93 wchar_t chW;
94 char mbch[2];
95 int n_chars;
97 if(locale)
98 mbcinfo = locale->mbcinfo;
99 else
100 mbcinfo = get_mbcinfo();
102 if (ch <= 0xff) {
103 mbch[0] = ch;
104 n_chars = 1;
105 } else {
106 mbch[0] = (ch >> 8) & 0xff;
107 mbch[1] = ch & 0xff;
108 n_chars = 2;
110 if (!MultiByteToWideChar(mbcinfo->mbcodepage, 0, mbch, n_chars, &chW, 1))
112 WARN("MultiByteToWideChar failed on %x\n", ch);
113 return 0;
115 return chW;
118 static wchar_t msvcrt_mbc_to_wc(unsigned int ch)
120 return msvcrt_mbc_to_wc_l(ch, NULL);
123 static inline size_t u_strlen( const unsigned char *str )
125 return strlen( (const char*) str );
128 static inline unsigned char* u_strncat( unsigned char* dst, const unsigned char* src, size_t len )
130 return (unsigned char*)strncat( (char*)dst, (const char*)src, len);
133 static inline int u_strcmp( const unsigned char *s1, const unsigned char *s2 )
135 return strcmp( (const char*)s1, (const char*)s2 );
138 static inline int u_strcasecmp( const unsigned char *s1, const unsigned char *s2 )
140 return _stricmp( (const char*)s1, (const char*)s2 );
143 static inline int u_strncmp( const unsigned char *s1, const unsigned char *s2, size_t len )
145 return strncmp( (const char*)s1, (const char*)s2, len );
148 static inline int u_strncasecmp( const unsigned char *s1, const unsigned char *s2, size_t len )
150 return _strnicmp( (const char*)s1, (const char*)s2, len );
153 static inline unsigned char *u_strchr( const unsigned char *s, unsigned char x )
155 return (unsigned char*) strchr( (const char*)s, x );
158 static inline unsigned char *u_strrchr( const unsigned char *s, unsigned char x )
160 return (unsigned char*) strrchr( (const char*)s, x );
163 static inline unsigned char *u__strset( unsigned char *s, unsigned char c )
165 return (unsigned char*) _strset( (char*)s, c);
168 static inline unsigned char *u__strnset( unsigned char *s, unsigned char c, size_t len )
170 return (unsigned char*) _strnset( (char*)s, c, len );
173 /*********************************************************************
174 * __p__mbctype (MSVCRT.@)
176 unsigned char* CDECL __p__mbctype(void)
178 return get_mbcinfo()->mbctype;
181 /*********************************************************************
182 * __p___mb_cur_max(MSVCRT.@)
184 int* CDECL __p___mb_cur_max(void)
186 return &get_locinfo()->mb_cur_max;
189 /*********************************************************************
190 * ___mb_cur_max_func(MSVCRT.@)
192 int CDECL ___mb_cur_max_func(void)
194 return get_locinfo()->mb_cur_max;
197 #if _MSVCR_VER>=80
198 /*********************************************************************
199 * ___mb_cur_max_l_func (MSVCR80.@)
201 int CDECL ___mb_cur_max_l_func(_locale_t locale)
203 pthreadlocinfo locinfo;
205 if(!locale)
206 locinfo = get_locinfo();
207 else
208 locinfo = locale->locinfo;
210 return locinfo->mb_cur_max;
212 #endif
214 threadmbcinfo* create_mbcinfo(int cp, LCID lcid, threadmbcinfo *old_mbcinfo)
216 threadmbcinfo *mbcinfo;
217 int newcp;
218 CPINFO cpi;
219 BYTE *bytes;
220 WORD chartypes[256];
221 char bufA[256];
222 WCHAR bufW[256], lowW[256], upW[256];
223 int charcount, maxchar;
224 int ret;
225 int i;
227 if(old_mbcinfo && cp==old_mbcinfo->mbcodepage
228 && (lcid==-1 || lcid==old_mbcinfo->mblcid)) {
229 InterlockedIncrement(&old_mbcinfo->refcount);
230 return old_mbcinfo;
233 mbcinfo = malloc(sizeof(threadmbcinfo));
234 if(!mbcinfo)
235 return NULL;
236 mbcinfo->refcount = 1;
238 switch (cp)
240 case _MB_CP_ANSI:
241 newcp = GetACP();
242 break;
243 case _MB_CP_OEM:
244 newcp = GetOEMCP();
245 break;
246 case _MB_CP_LOCALE:
247 newcp = get_locinfo()->lc_codepage;
248 if(newcp)
249 break;
250 /* fall through (C locale) */
251 case _MB_CP_SBCS:
252 newcp = 20127; /* ASCII */
253 break;
254 default:
255 newcp = cp;
256 break;
259 if(lcid == -1) {
260 sprintf(bufA, ".%d", newcp);
261 mbcinfo->mblcid = locale_to_LCID(bufA, NULL, NULL);
262 } else {
263 mbcinfo->mblcid = lcid;
266 if(mbcinfo->mblcid == -1)
268 WARN("Can't assign LCID to codepage (%d)\n", mbcinfo->mblcid);
269 mbcinfo->mblcid = 0;
272 if (!GetCPInfo(newcp, &cpi))
274 WARN("Codepage %d not found\n", newcp);
275 free(mbcinfo);
276 return NULL;
279 /* setup the _mbctype */
280 memset(mbcinfo->mbctype, 0, sizeof(unsigned char[257]));
281 memset(mbcinfo->mbcasemap, 0, sizeof(unsigned char[256]));
283 bytes = cpi.LeadByte;
284 while (bytes[0] || bytes[1])
286 for (i = bytes[0]; i <= bytes[1]; i++)
287 mbcinfo->mbctype[i + 1] |= _M1;
288 bytes += 2;
291 if (cpi.MaxCharSize == 2)
293 /* trail bytes not available through kernel32 but stored in a structure in msvcrt */
294 struct cp_extra_info_t *cpextra = g_cpextrainfo;
296 mbcinfo->ismbcodepage = 1;
297 while (TRUE)
299 if (cpextra->cp == 0 || cpextra->cp == newcp)
301 if (cpextra->cp == 0)
302 FIXME("trail bytes data not available for DBCS codepage %d - assuming all bytes\n", newcp);
304 bytes = cpextra->TrailBytes;
305 while (bytes[0] || bytes[1])
307 for (i = bytes[0]; i <= bytes[1]; i++)
308 mbcinfo->mbctype[i + 1] |= _M2;
309 bytes += 2;
311 break;
313 cpextra++;
316 else
317 mbcinfo->ismbcodepage = 0;
319 maxchar = (newcp == CP_UTF8) ? 128 : 256;
321 /* we can't use GetStringTypeA directly because we don't have a locale - only a code page
323 charcount = 0;
324 for (i = 0; i < maxchar; i++)
325 if (!(mbcinfo->mbctype[i + 1] & _M1))
326 bufA[charcount++] = i;
328 ret = MultiByteToWideChar(newcp, 0, bufA, charcount, bufW, charcount);
329 if (ret != charcount)
331 ERR("MultiByteToWideChar of chars failed for cp %d, ret=%d (exp %d), error=%ld\n",
332 newcp, ret, charcount, GetLastError());
335 GetStringTypeW(CT_CTYPE1, bufW, charcount, chartypes);
336 LCMapStringW(mbcinfo->mblcid, LCMAP_LOWERCASE, bufW, charcount, lowW, charcount);
337 LCMapStringW(mbcinfo->mblcid, LCMAP_UPPERCASE, bufW, charcount, upW, charcount);
339 charcount = 0;
340 for (i = 0; i < maxchar; i++)
341 if (!(mbcinfo->mbctype[i + 1] & _M1))
343 if (chartypes[charcount] & C1_UPPER)
345 mbcinfo->mbctype[i + 1] |= _SBUP;
346 bufW[charcount] = lowW[charcount];
348 else if (chartypes[charcount] & C1_LOWER)
350 mbcinfo->mbctype[i + 1] |= _SBLOW;
351 bufW[charcount] = upW[charcount];
353 charcount++;
356 ret = WideCharToMultiByte(newcp, 0, bufW, charcount, bufA, charcount, NULL, NULL);
357 if (ret != charcount)
359 ERR("WideCharToMultiByte failed for cp %d, ret=%d (exp %d), error=%ld\n",
360 newcp, ret, charcount, GetLastError());
363 charcount = 0;
364 for (i = 0; i < maxchar; i++)
366 if(!(mbcinfo->mbctype[i + 1] & _M1))
368 if(mbcinfo->mbctype[i + 1] & (_SBUP | _SBLOW))
369 mbcinfo->mbcasemap[i] = bufA[charcount];
370 charcount++;
374 if (newcp == 932) /* CP932 only - set _MP and _MS */
376 /* On Windows it's possible to calculate the _MP and _MS from CT_CTYPE1
377 * and CT_CTYPE3. But as of Wine 0.9.43 we return wrong values what makes
378 * it hard. As this is set only for codepage 932 we hardcode it what gives
379 * also faster execution.
381 for (i = 161; i <= 165; i++)
382 mbcinfo->mbctype[i + 1] |= _MP;
383 for (i = 166; i <= 223; i++)
384 mbcinfo->mbctype[i + 1] |= _MS;
387 mbcinfo->mbcodepage = newcp;
388 return mbcinfo;
391 /*********************************************************************
392 * _setmbcp (MSVCRT.@)
394 int CDECL _setmbcp(int cp)
396 thread_data_t *data = msvcrt_get_thread_data();
397 threadmbcinfo *mbcinfo;
399 mbcinfo = create_mbcinfo(cp, -1, get_mbcinfo());
400 if(!mbcinfo)
402 *_errno() = EINVAL;
403 return -1;
406 if(data->locale_flags & LOCALE_THREAD)
408 if(data->locale_flags & LOCALE_FREE)
409 free_mbcinfo(data->mbcinfo);
410 data->mbcinfo = mbcinfo;
412 else
414 _lock(_MB_CP_LOCK);
415 free_mbcinfo(MSVCRT_locale->mbcinfo);
416 MSVCRT_locale->mbcinfo = mbcinfo;
417 memcpy(MSVCRT_mbctype, MSVCRT_locale->mbcinfo->mbctype, sizeof(MSVCRT_mbctype));
418 _unlock(_MB_CP_LOCK);
420 return 0;
423 /*********************************************************************
424 * _getmbcp (MSVCRT.@)
426 int CDECL _getmbcp(void)
428 return get_mbcinfo()->mbcodepage;
431 /*********************************************************************
432 * _mbsnextc_l(MSVCRT.@)
434 unsigned int CDECL _mbsnextc_l(const unsigned char* str, _locale_t locale)
436 if(_ismbblead_l(*str, locale))
437 return *str << 8 | str[1];
438 return *str;
441 /*********************************************************************
442 * _mbsnextc(MSVCRT.@)
444 unsigned int CDECL _mbsnextc(const unsigned char* str)
446 return _mbsnextc_l(str, NULL);
449 /*********************************************************************
450 * _mbctolower_l(MSVCRT.@)
452 unsigned int CDECL _mbctolower_l(unsigned int c, _locale_t locale)
454 unsigned char str[2], ret[2];
455 pthreadmbcinfo mbcinfo;
457 if(!locale)
458 mbcinfo = get_mbcinfo();
459 else
460 mbcinfo = locale->mbcinfo;
462 if (c > 0xff)
464 if (!_ismbblead_l((c >> 8) & 0xff, locale))
465 return c;
467 str[0] = c >> 8;
468 str[1] = c;
469 switch(__crtLCMapStringA(mbcinfo->mblcid, LCMAP_LOWERCASE,
470 (char*)str, 2, (char*)ret, 2, mbcinfo->mbcodepage, 0))
472 case 0:
473 return c;
474 case 1:
475 return ret[0];
476 default:
477 return ret[1] + (ret[0] << 8);
481 return mbcinfo->mbctype[c + 1] & _SBUP ? mbcinfo->mbcasemap[c] : c;
484 /*********************************************************************
485 * _mbctolower(MSVCRT.@)
487 unsigned int CDECL _mbctolower(unsigned int c)
489 return _mbctolower_l(c, NULL);
492 /*********************************************************************
493 * _mbctoupper_l(MSVCRT.@)
495 unsigned int CDECL _mbctoupper_l(unsigned int c, _locale_t locale)
497 unsigned char str[2], ret[2];
498 pthreadmbcinfo mbcinfo;
500 if(!locale)
501 mbcinfo = get_mbcinfo();
502 else
503 mbcinfo = locale->mbcinfo;
505 if (c > 0xff)
507 if (!_ismbblead_l((c >> 8) & 0xff, locale))
508 return c;
510 str[0] = c >> 8;
511 str[1] = c;
512 switch(__crtLCMapStringA(mbcinfo->mblcid, LCMAP_UPPERCASE,
513 (char*)str, 2, (char*)ret, 2, mbcinfo->mbcodepage, 0))
515 case 0:
516 return c;
517 case 1:
518 return ret[0];
519 default:
520 return ret[1] + (ret[0] << 8);
524 return mbcinfo->mbctype[c + 1] & _SBLOW ? mbcinfo->mbcasemap[c] : c;
527 /*********************************************************************
528 * _mbctoupper(MSVCRT.@)
530 unsigned int CDECL _mbctoupper(unsigned int c)
532 return _mbctoupper_l(c, NULL);
535 /*********************************************************************
536 * _mbctombb (MSVCRT.@)
538 unsigned int CDECL _mbctombb(unsigned int c)
540 unsigned int value;
542 if(get_mbcinfo()->mbcodepage == 932)
544 if(c >= 0x829f && c <= 0x82f1) /* Hiragana */
545 return mbctombb_932_kana[c - 0x829f];
546 if(c >= 0x8340 && c <= 0x8396 && c != 0x837f) /* Katakana */
547 return mbctombb_932_kana[c - 0x8340 - (c >= 0x837f ? 1 : 0)];
548 if(c >= 0x8140 && c <= 0x8197) /* Punctuation */
550 value = mbctombb_932_punct[c - 0x8140];
551 return value ? value : c;
553 if((c >= 0x824f && c <= 0x8258) || /* Fullwidth digits */
554 (c >= 0x8260 && c <= 0x8279)) /* Fullwidth capitals letters */
555 return c - 0x821f;
556 if(c >= 0x8281 && c <= 0x829a) /* Fullwidth small letters */
557 return c - 0x8220;
558 /* all other cases return c */
560 return c;
563 /*********************************************************************
564 * _mbcjistojms(MSVCRT.@)
566 * Converts a jis character to sjis.
567 * Based on description from
568 * http://www.slayers.ne.jp/~oouchi/code/jistosjis.html
570 unsigned int CDECL _mbcjistojms(unsigned int c)
572 /* Conversion takes place only when codepage is 932.
573 In all other cases, c is returned unchanged */
574 if(get_mbcinfo()->mbcodepage == 932)
576 if(HIBYTE(c) >= 0x21 && HIBYTE(c) <= 0x7e &&
577 LOBYTE(c) >= 0x21 && LOBYTE(c) <= 0x7e)
579 if(HIBYTE(c) % 2)
580 c += 0x1f;
581 else
582 c += 0x7d;
584 if(LOBYTE(c) >= 0x7F)
585 c += 0x1;
587 c = (((HIBYTE(c) - 0x21)/2 + 0x81) << 8) | LOBYTE(c);
589 if(HIBYTE(c) > 0x9f)
590 c += 0x4000;
592 else
593 return 0; /* Codepage is 932, but c can't be converted */
596 return c;
599 /*********************************************************************
600 * _mbcjmstojis(MSVCRT.@)
602 * Converts a sjis character to jis.
604 unsigned int CDECL _mbcjmstojis(unsigned int c)
606 /* Conversion takes place only when codepage is 932.
607 In all other cases, c is returned unchanged */
608 if(get_mbcinfo()->mbcodepage == 932)
610 if(_ismbclegal(c) && HIBYTE(c) < 0xf0)
612 if(HIBYTE(c) >= 0xe0)
613 c -= 0x4000;
615 c = (((HIBYTE(c) - 0x81)*2 + 0x21) << 8) | LOBYTE(c);
617 if(LOBYTE(c) > 0x7f)
618 c -= 0x1;
620 if(LOBYTE(c) > 0x9d)
621 c += 0x83;
622 else
623 c -= 0x1f;
625 else
626 return 0; /* Codepage is 932, but c can't be converted */
629 return c;
632 /*********************************************************************
633 * _mbsdec(MSVCRT.@)
635 unsigned char* CDECL _mbsdec(const unsigned char* start, const unsigned char* cur)
637 if(start >= cur)
638 return NULL;
640 if(get_mbcinfo()->ismbcodepage)
641 return (unsigned char *)(_ismbstrail(start,cur-1) ? cur - 2 : cur -1);
643 return (unsigned char *)cur - 1; /* ASCII CP or SB char */
646 /*********************************************************************
647 * _mbclen_l(MSVCRT.@)
649 size_t CDECL _mbclen_l(const unsigned char* str, _locale_t locale)
651 return _ismbblead_l(*str, locale) && str[1] ? 2 : 1;
654 /*********************************************************************
655 * _mbclen(MSVCRT.@)
657 size_t CDECL _mbclen(const unsigned char* str)
659 return _mbclen_l(str, NULL);
662 /*********************************************************************
663 * _mbsinc(MSVCRT.@)
665 unsigned char* CDECL _mbsinc(const unsigned char* str)
667 return (unsigned char *)(str + _mbclen(str));
670 /*********************************************************************
671 * _mbsninc(MSVCRT.@)
673 unsigned char* CDECL _mbsninc(const unsigned char* str, size_t num)
675 if(!str)
676 return NULL;
678 while (num > 0 && *str)
680 if (_ismbblead(*str))
682 if (!*(str+1))
683 break;
684 str++;
686 str++;
687 num--;
690 return (unsigned char*)str;
693 /*********************************************************************
694 * _mbsnlen_l(MSVCRT.@)
696 size_t CDECL _mbsnlen_l(const unsigned char *str,
697 size_t maxsize, _locale_t locale)
699 pthreadmbcinfo mbcinfo;
700 size_t i = 0, len = 0;
702 if(!locale)
703 mbcinfo = get_mbcinfo();
704 else
705 mbcinfo = locale->mbcinfo;
707 if(!mbcinfo->ismbcodepage)
708 return strnlen((const char*)str, maxsize);
710 while(i<maxsize && str[i])
712 if (_ismbblead_l(str[i], locale))
714 i++;
715 if (!str[i]) /* count only full chars */
716 break;
718 i++;
719 len++;
721 return i < maxsize ? len : maxsize;
724 /*********************************************************************
725 * _mbslen(MSVCRT.@)
727 size_t CDECL _mbslen(const unsigned char* str)
729 return _mbsnlen_l(str, -1, NULL);
732 /*********************************************************************
733 * _mbslen_l(MSVCRT.@)
735 size_t CDECL _mbslen_l(const unsigned char* str, _locale_t locale)
737 return _mbsnlen_l(str, -1, locale);
740 /*********************************************************************
741 * _mbsnlen(MSVCRT.@)
743 size_t CDECL _mbsnlen(const unsigned char* str, size_t maxsize)
745 return _mbsnlen_l(str, maxsize, NULL);
748 /*********************************************************************
749 * _mbccpy_s_l(MSVCRT.@)
751 int CDECL _mbccpy_s_l(unsigned char* dest, size_t maxsize,
752 int *copied, const unsigned char* src, _locale_t locale)
754 if(copied) *copied = 0;
755 if(!MSVCRT_CHECK_PMT(dest != NULL && maxsize >= 1)) return EINVAL;
756 dest[0] = 0;
757 if(!MSVCRT_CHECK_PMT(src != NULL)) return EINVAL;
759 if(_ismbblead_l(*src, locale)) {
760 if(!src[1]) {
761 if(copied) *copied = 1;
762 *_errno() = EILSEQ;
763 return EILSEQ;
766 if(maxsize < 2) {
767 MSVCRT_INVALID_PMT("dst buffer is too small", ERANGE);
768 return ERANGE;
771 *dest++ = *src++;
772 *dest = *src;
773 if(copied) *copied = 2;
774 }else {
775 *dest = *src;
776 if(copied) *copied = 1;
779 return 0;
782 /*********************************************************************
783 * _mbccpy(MSVCRT.@)
785 void CDECL _mbccpy(unsigned char* dest, const unsigned char* src)
787 _mbccpy_s_l(dest, 2, NULL, src, NULL);
790 /*********************************************************************
791 * _mbccpy_l(MSVCRT.@)
793 void CDECL _mbccpy_l(unsigned char* dest, const unsigned char* src,
794 _locale_t locale)
796 _mbccpy_s_l(dest, 2, NULL, src, locale);
799 /*********************************************************************
800 * _mbccpy_s(MSVCRT.@)
802 int CDECL _mbccpy_s(unsigned char* dest, size_t maxsize,
803 int *copied, const unsigned char* src)
805 return _mbccpy_s_l(dest, maxsize, copied, src, NULL);
808 /*********************************************************************
809 * _mbsncpy(MSVCRT.@)
810 * REMARKS
811 * The parameter n is the number or characters to copy, not the size of
812 * the buffer. Use _mbsnbcpy for a function analogical to strncpy
814 unsigned char* CDECL _mbsncpy(unsigned char* dst, const unsigned char* src, size_t n)
816 unsigned char* ret = dst;
817 if(!n)
818 return dst;
819 if (get_mbcinfo()->ismbcodepage)
821 while (*src && n)
823 n--;
824 if (_ismbblead(*src))
826 if (!*(src+1))
828 *dst++ = 0;
829 *dst++ = 0;
830 break;
833 *dst++ = *src++;
836 *dst++ = *src++;
839 else
841 while (n)
843 n--;
844 if (!(*dst++ = *src++)) break;
847 while (n--) *dst++ = 0;
848 return ret;
851 /*********************************************************************
852 * _mbsnbcpy_s_l(MSVCRT.@)
853 * REMARKS
854 * Unlike _mbsnbcpy this function does not pad the rest of the dest
855 * string with 0
857 int CDECL _mbsnbcpy_s_l(unsigned char* dst, size_t size,
858 const unsigned char* src, size_t n, _locale_t locale)
860 size_t pos = 0;
862 if(!dst || size == 0)
863 return EINVAL;
864 if(!src)
866 dst[0] = '\0';
867 return EINVAL;
869 if(!n)
870 return 0;
872 if((locale ? locale->mbcinfo : get_mbcinfo())->ismbcodepage)
874 BOOL is_lead = FALSE;
875 while (*src && n)
877 if(pos == size)
879 dst[0] = '\0';
880 return ERANGE;
882 is_lead = (!is_lead && _ismbblead(*src));
883 n--;
884 dst[pos++] = *src++;
887 if (is_lead) /* if string ends with a lead, remove it */
888 dst[pos - 1] = 0;
890 else
892 while (n)
894 n--;
895 if(pos == size)
897 dst[0] = '\0';
898 return ERANGE;
901 if(!(*src)) break;
902 dst[pos++] = *src++;
906 if(pos < size)
907 dst[pos] = '\0';
908 else
910 dst[0] = '\0';
911 return ERANGE;
914 return 0;
917 /*********************************************************************
918 * _mbsnbcpy_s(MSVCRT.@)
920 int CDECL _mbsnbcpy_s(unsigned char* dst, size_t size, const unsigned char* src, size_t n)
922 return _mbsnbcpy_s_l(dst, size, src, n, NULL);
925 /*********************************************************************
926 * _mbscpy_s_l(MSVCRT.@)
928 int CDECL _mbscpy_s_l(unsigned char *dst, size_t size,
929 const unsigned char *src, _locale_t locale)
931 return _mbsnbcpy_s_l(dst, size, src, -1, locale);
934 /*********************************************************************
935 * _mbscpy_s(MSVCRT.@)
937 int CDECL _mbscpy_s(unsigned char *dst, size_t size, const unsigned char *src)
939 return _mbscpy_s_l(dst, size, src, NULL);
942 /*********************************************************************
943 * _mbsnbcpy(MSVCRT.@)
944 * REMARKS
945 * Like strncpy this function doesn't enforce the string to be
946 * NUL-terminated
948 unsigned char* CDECL _mbsnbcpy(unsigned char* dst, const unsigned char* src, size_t n)
950 unsigned char* ret = dst;
951 if(!n)
952 return dst;
953 if(get_mbcinfo()->ismbcodepage)
955 BOOL is_lead = FALSE;
956 while (*src && n)
958 is_lead = (!is_lead && _ismbblead(*src));
959 n--;
960 *dst++ = *src++;
963 if (is_lead) /* if string ends with a lead, remove it */
964 *(dst - 1) = 0;
966 else
968 while (n)
970 n--;
971 if (!(*dst++ = *src++)) break;
974 while (n--) *dst++ = 0;
975 return ret;
978 /*********************************************************************
979 * _mbscmp_l(MSVCRT.@)
981 int CDECL _mbscmp_l(const unsigned char* str, const unsigned char* cmp, _locale_t locale)
983 pthreadmbcinfo mbcinfo;
985 if (!MSVCRT_CHECK_PMT(str && cmp))
986 return _NLSCMPERROR;
988 mbcinfo = locale ? locale->mbcinfo : get_mbcinfo();
990 if(mbcinfo->ismbcodepage)
992 unsigned int strc, cmpc;
993 do {
994 if(!*str)
995 return *cmp ? -1 : 0;
996 if(!*cmp)
997 return 1;
998 strc = _mbsnextc_l(str, locale);
999 cmpc = _mbsnextc_l(cmp, locale);
1000 if(strc != cmpc)
1001 return strc < cmpc ? -1 : 1;
1002 str +=(strc > 255) ? 2 : 1;
1003 cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
1004 } while(1);
1006 return u_strcmp(str, cmp); /* ASCII CP */
1009 /*********************************************************************
1010 * _mbscmp(MSVCRT.@)
1012 int CDECL _mbscmp(const unsigned char* str, const unsigned char* cmp)
1014 return _mbscmp_l(str, cmp, NULL);
1017 /*********************************************************************
1018 * _mbsnbicoll_l(MSVCRT.@)
1020 int CDECL _mbsnbicoll_l(const unsigned char *str1, const unsigned char *str2, size_t len, _locale_t locale)
1022 pthreadmbcinfo mbcinfo;
1024 if(!locale)
1025 mbcinfo = get_mbcinfo();
1026 else
1027 mbcinfo = locale->mbcinfo;
1029 if(!mbcinfo->ismbcodepage)
1030 return _strnicoll_l((const char*)str1, (const char*)str2, len, locale);
1031 return CompareStringA(mbcinfo->mblcid, NORM_IGNORECASE, (const char*)str1, len, (const char*)str2, len)-CSTR_EQUAL;
1034 /*********************************************************************
1035 * _mbsicoll_l(MSVCRT.@)
1037 int CDECL _mbsicoll_l(const unsigned char *str1, const unsigned char *str2, _locale_t locale)
1039 return _mbsnbicoll_l(str1, str2, -1, locale);
1042 /*********************************************************************
1043 * _mbsnbicoll(MSVCRT.@)
1045 int CDECL _mbsnbicoll(const unsigned char *str1, const unsigned char *str2, size_t len)
1047 return _mbsnbicoll_l(str1, str2, len, NULL);
1050 /*********************************************************************
1051 * _mbsicoll(MSVCRT.@)
1053 int CDECL _mbsicoll(const unsigned char* str, const unsigned char* cmp)
1055 #if _MSVCR_VER>=60 && _MSVCR_VER<=71
1056 return CompareStringA(get_mbcinfo()->mblcid, NORM_IGNORECASE,
1057 (const char*)str, -1, (const char*)cmp, -1)-CSTR_EQUAL;
1058 #else
1059 return _mbsnbicoll_l(str, cmp, -1, NULL);
1060 #endif
1063 /*********************************************************************
1064 * _mbsnbcoll_l(MSVCRT.@)
1066 int CDECL _mbsnbcoll_l(const unsigned char *str1, const unsigned char *str2, size_t len, _locale_t locale)
1068 pthreadmbcinfo mbcinfo;
1070 if(!locale)
1071 mbcinfo = get_mbcinfo();
1072 else
1073 mbcinfo = locale->mbcinfo;
1075 if(!mbcinfo->ismbcodepage)
1076 return _strncoll_l((const char*)str1, (const char*)str2, len, locale);
1077 return CompareStringA(mbcinfo->mblcid, 0, (const char*)str1, len, (const char*)str2, len)-CSTR_EQUAL;
1080 /*********************************************************************
1081 * _mbscoll_l(MSVCRT.@)
1083 int CDECL _mbscoll_l(const unsigned char *str1, const unsigned char *str2, _locale_t locale)
1085 return _mbsnbcoll_l(str1, str2, -1, locale);
1088 /*********************************************************************
1089 * _mbsnbcoll(MSVCRT.@)
1091 int CDECL _mbsnbcoll(const unsigned char *str1, const unsigned char *str2, size_t len)
1093 return _mbsnbcoll_l(str1, str2, len, NULL);
1096 /*********************************************************************
1097 * _mbscoll(MSVCRT.@)
1099 int CDECL _mbscoll(const unsigned char* str, const unsigned char* cmp)
1101 #if _MSVCR_VER>=60 && _MSVCR_VER<=71
1102 return CompareStringA(get_mbcinfo()->mblcid, 0,
1103 (const char*)str, -1, (const char*)cmp, -1)-CSTR_EQUAL;
1104 #else
1105 return _mbsnbcoll_l(str, cmp, -1, NULL);
1106 #endif
1109 /*********************************************************************
1110 * _mbsicmp_l(MSVCRT.@)
1112 int CDECL _mbsicmp_l(const unsigned char* str, const unsigned char* cmp, _locale_t locale)
1114 pthreadmbcinfo mbcinfo;
1116 if(!MSVCRT_CHECK_PMT(str && cmp))
1117 return _NLSCMPERROR;
1119 if(!locale)
1120 mbcinfo = get_mbcinfo();
1121 else
1122 mbcinfo = locale->mbcinfo;
1123 if(mbcinfo->ismbcodepage)
1125 unsigned int strc, cmpc;
1126 do {
1127 if(!*str)
1128 return *cmp ? -1 : 0;
1129 if(!*cmp)
1130 return 1;
1131 strc = _mbctolower_l(_mbsnextc_l(str, locale), locale);
1132 cmpc = _mbctolower_l(_mbsnextc_l(cmp, locale), locale);
1133 if(strc != cmpc)
1134 return strc < cmpc ? -1 : 1;
1135 str +=(strc > 255) ? 2 : 1;
1136 cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
1137 } while(1);
1139 return u_strcasecmp(str, cmp); /* ASCII CP */
1142 /*********************************************************************
1143 * _mbsicmp(MSVCRT.@)
1145 int CDECL _mbsicmp(const unsigned char* str, const unsigned char* cmp)
1147 return _mbsicmp_l(str, cmp, NULL);
1150 /*********************************************************************
1151 * _mbsncmp(MSVCRT.@)
1153 int CDECL _mbsncmp(const unsigned char* str, const unsigned char* cmp, size_t len)
1155 if(!len)
1156 return 0;
1158 if(get_mbcinfo()->ismbcodepage)
1160 unsigned int strc, cmpc;
1161 while(len--)
1163 int inc;
1164 if(!*str)
1165 return *cmp ? -1 : 0;
1166 if(!*cmp)
1167 return 1;
1168 strc = _mbsnextc(str);
1169 cmpc = _mbsnextc(cmp);
1170 if(strc != cmpc)
1171 return strc < cmpc ? -1 : 1;
1172 inc=(strc > 255) ? 2 : 1; /* Equal, use same increment */
1173 str += inc;
1174 cmp += inc;
1176 return 0; /* Matched len chars */
1178 return u_strncmp(str, cmp, len); /* ASCII CP */
1181 /*********************************************************************
1182 * _mbsnbcmp(MSVCRT.@)
1184 int CDECL _mbsnbcmp(const unsigned char* str, const unsigned char* cmp, size_t len)
1186 if (!len)
1187 return 0;
1188 if(get_mbcinfo()->ismbcodepage)
1190 unsigned int strc, cmpc;
1191 while (len)
1193 int clen;
1194 if(!*str)
1195 return *cmp ? -1 : 0;
1196 if(!*cmp)
1197 return 1;
1198 if (_ismbblead(*str))
1200 strc=(len>=2)?_mbsnextc(str):0;
1201 clen=2;
1203 else
1205 strc=*str;
1206 clen=1;
1208 if (_ismbblead(*cmp))
1209 cmpc=(len>=2)?_mbsnextc(cmp):0;
1210 else
1211 cmpc=*cmp;
1212 if(strc != cmpc)
1213 return strc < cmpc ? -1 : 1;
1214 len -= clen;
1215 str += clen;
1216 cmp += clen;
1218 return 0; /* Matched len chars */
1220 return u_strncmp(str,cmp,len);
1223 /*********************************************************************
1224 * _mbsnicmp(MSVCRT.@)
1226 * Compare two multibyte strings case insensitively to 'len' characters.
1228 int CDECL _mbsnicmp(const unsigned char* str, const unsigned char* cmp, size_t len)
1230 /* FIXME: No tolower() for mb strings yet */
1231 if(get_mbcinfo()->ismbcodepage)
1233 unsigned int strc, cmpc;
1234 while(len--)
1236 if(!*str)
1237 return *cmp ? -1 : 0;
1238 if(!*cmp)
1239 return 1;
1240 strc = _mbctolower(_mbsnextc(str));
1241 cmpc = _mbctolower(_mbsnextc(cmp));
1242 if(strc != cmpc)
1243 return strc < cmpc ? -1 : 1;
1244 str +=(strc > 255) ? 2 : 1;
1245 cmp +=(strc > 255) ? 2 : 1; /* Equal, use same increment */
1247 return 0; /* Matched len chars */
1249 return u_strncasecmp(str, cmp, len); /* ASCII CP */
1252 /*********************************************************************
1253 * _mbsnbicmp(MSVCRT.@)
1255 int CDECL _mbsnbicmp(const unsigned char* str, const unsigned char* cmp, size_t len)
1257 if (!len)
1258 return 0;
1259 if(get_mbcinfo()->ismbcodepage)
1261 unsigned int strc, cmpc;
1262 while (len)
1264 int clen;
1265 if(!*str)
1266 return *cmp ? -1 : 0;
1267 if(!*cmp)
1268 return 1;
1269 if (_ismbblead(*str))
1271 strc=(len>=2)?_mbsnextc(str):0;
1272 clen=2;
1274 else
1276 strc=*str;
1277 clen=1;
1279 if (_ismbblead(*cmp))
1280 cmpc=(len>=2)?_mbsnextc(cmp):0;
1281 else
1282 cmpc=*cmp;
1283 strc = _mbctolower(strc);
1284 cmpc = _mbctolower(cmpc);
1285 if(strc != cmpc)
1286 return strc < cmpc ? -1 : 1;
1287 len -= clen;
1288 str += clen;
1289 cmp += clen;
1291 return 0; /* Matched len bytes */
1293 return u_strncasecmp(str,cmp,len);
1296 /*********************************************************************
1297 * _mbscat (MSVCRT.@)
1299 unsigned char * CDECL _mbscat( unsigned char *dst, const unsigned char *src )
1301 strcat( (char *)dst, (const char *)src );
1302 return dst;
1305 /*********************************************************************
1306 * _mbscat_s_l (MSVCRT.@)
1308 int CDECL _mbscat_s_l( unsigned char *dst, size_t size,
1309 const unsigned char *src, _locale_t locale )
1311 size_t i, j;
1312 int ret = 0;
1314 if(!MSVCRT_CHECK_PMT(dst != NULL)) return EINVAL;
1315 if(!MSVCRT_CHECK_PMT(src != NULL)) return EINVAL;
1317 for(i=0; i<size; i++)
1318 if(!dst[i]) break;
1319 if(i == size) {
1320 MSVCRT_INVALID_PMT("dst is not NULL-terminated", EINVAL);
1321 if(size) dst[0] = 0;
1322 return EINVAL;
1325 if(i && _ismbblead_l(dst[i-1], locale)) {
1326 ret = EILSEQ;
1327 i--;
1330 for(j=0; src[j] && i+j<size; j++)
1331 dst[i+j] = src[j];
1332 if(i+j == size) {
1333 MSVCRT_INVALID_PMT("dst buffer is too small", ERANGE);
1334 dst[0] = 0;
1335 return ERANGE;
1338 if(j && _ismbblead_l(src[j-1], locale)) {
1339 ret = EILSEQ;
1340 j--;
1343 dst[i+j] = 0;
1344 return ret;
1347 /*********************************************************************
1348 * _mbscat_s (MSVCRT.@)
1350 int CDECL _mbscat_s( unsigned char *dst, size_t size, const unsigned char *src )
1352 return _mbscat_s_l(dst, size, src, NULL);
1355 /*********************************************************************
1356 * _mbscpy (MSVCRT.@)
1358 unsigned char* CDECL _mbscpy( unsigned char *dst, const unsigned char *src )
1360 strcpy( (char *)dst, (const char *)src );
1361 return dst;
1364 /*********************************************************************
1365 * _mbsstr (MSVCRT.@)
1367 unsigned char * CDECL _mbsstr(const unsigned char *haystack, const unsigned char *needle)
1369 return (unsigned char *)strstr( (const char *)haystack, (const char *)needle );
1372 /*********************************************************************
1373 * _mbschr_l(MSVCRT.@)
1375 unsigned char* CDECL _mbschr_l(const unsigned char* s, unsigned int x, _locale_t locale)
1377 pthreadmbcinfo mbcinfo;
1379 if(!MSVCRT_CHECK_PMT(s))
1380 return NULL;
1382 if(locale)
1383 mbcinfo = locale->mbcinfo;
1384 else
1385 mbcinfo = get_mbcinfo();
1386 if(mbcinfo->ismbcodepage)
1388 unsigned int c;
1389 while (1)
1391 c = _mbsnextc_l(s, locale);
1392 if (c == x)
1393 return (unsigned char*)s;
1394 if (!c)
1395 return NULL;
1396 s += c > 255 ? 2 : 1;
1399 return u_strchr(s, x); /* ASCII CP */
1402 /*********************************************************************
1403 * _mbschr(MSVCRT.@)
1405 unsigned char* CDECL _mbschr(const unsigned char* s, unsigned int x)
1407 return _mbschr_l(s, x, NULL);
1410 /*********************************************************************
1411 * _mbsrchr(MSVCRT.@)
1413 unsigned char* CDECL _mbsrchr(const unsigned char* s, unsigned int x)
1415 if(get_mbcinfo()->ismbcodepage)
1417 unsigned int c;
1418 unsigned char* match=NULL;
1419 if(!s)
1420 return NULL;
1421 while (1) {
1422 c = _mbsnextc(s);
1423 if (c == x)
1424 match=(unsigned char*)s;
1425 if (!c)
1426 return match;
1427 s +=(c > 255) ? 2 : 1;
1430 return u_strrchr(s, x);
1433 /*********************************************************************
1434 * _mbstok_s_l(MSVCRT.@)
1436 unsigned char* CDECL _mbstok_s_l(unsigned char *str, const unsigned char *delim,
1437 unsigned char **ctx, _locale_t locale)
1439 pthreadmbcinfo mbcinfo;
1440 unsigned int c;
1442 if(!MSVCRT_CHECK_PMT(delim != NULL)) return NULL;
1443 if(!MSVCRT_CHECK_PMT(ctx != NULL)) return NULL;
1444 if(!MSVCRT_CHECK_PMT(str || *ctx)) return NULL;
1446 if(locale)
1447 mbcinfo = locale->mbcinfo;
1448 else
1449 mbcinfo = get_mbcinfo();
1451 if(!mbcinfo->ismbcodepage)
1452 return (unsigned char*)strtok_s((char*)str, (const char*)delim, (char**)ctx);
1454 if(!str)
1455 str = *ctx;
1457 while((c=_mbsnextc(str)) && _mbschr(delim, c))
1458 str += c>255 ? 2 : 1;
1459 if(!*str)
1461 *ctx = str;
1462 return NULL;
1465 *ctx = str + (c>255 ? 2 : 1);
1466 while((c=_mbsnextc(*ctx)) && !_mbschr(delim, c))
1467 *ctx += c>255 ? 2 : 1;
1468 if (**ctx) {
1469 *(*ctx)++ = 0;
1470 if(c > 255)
1471 *(*ctx)++ = 0;
1474 return str;
1478 /*********************************************************************
1479 * _mbstok_s(MSVCRT.@)
1481 unsigned char* CDECL _mbstok_s(unsigned char *str,
1482 const unsigned char *delim, unsigned char **ctx)
1484 return _mbstok_s_l(str, delim, ctx, NULL);
1487 /*********************************************************************
1488 * _mbstok_l(MSVCRT.@)
1490 unsigned char* CDECL _mbstok_l(unsigned char *str,
1491 const unsigned char *delim, _locale_t locale)
1493 return _mbstok_s_l(str, delim, &msvcrt_get_thread_data()->mbstok_next, locale);
1496 /*********************************************************************
1497 * _mbstok(MSVCRT.@)
1499 unsigned char* CDECL _mbstok(unsigned char *str, const unsigned char *delim)
1501 thread_data_t *data = msvcrt_get_thread_data();
1503 #if _MSVCR_VER == 0
1504 if(!str && !data->mbstok_next)
1505 return NULL;
1506 #endif
1508 return _mbstok_s_l(str, delim, &data->mbstok_next, NULL);
1511 /*********************************************************************
1512 * _mbbtombc(MSVCRT.@)
1514 unsigned int CDECL _mbbtombc(unsigned int c)
1516 if(get_mbcinfo()->mbcodepage == 932)
1518 if(c >= 0x20 && c <= 0x7e) {
1519 if((c >= 0x41 && c <= 0x5a) || (c >= 0x61 && c <= 0x7a) || (c >= 0x30 && c <= 0x39))
1520 return mbbtombc_932[c - 0x20] | 0x8200;
1521 else
1522 return mbbtombc_932[c - 0x20] | 0x8100;
1524 else if(c >= 0xa1 && c <= 0xdf) {
1525 if(c >= 0xa6 && c <= 0xdd && c != 0xb0)
1526 return mbbtombc_932[c - 0xa1 + 0x5f] | 0x8300;
1527 else
1528 return mbbtombc_932[c - 0xa1 + 0x5f] | 0x8100;
1531 return c; /* not Japanese or no MB char */
1534 /*********************************************************************
1535 * _ismbbkana_l(MSVCRT.@)
1537 int CDECL _ismbbkana_l(unsigned int c, _locale_t locale)
1539 pthreadmbcinfo mbcinfo;
1541 if(locale)
1542 mbcinfo = locale->mbcinfo;
1543 else
1544 mbcinfo = get_mbcinfo();
1546 if(mbcinfo->mbcodepage == 932)
1548 /* Japanese/Katakana, CP 932 */
1549 return (c >= 0xa1 && c <= 0xdf);
1551 return 0;
1554 /*********************************************************************
1555 * _ismbbkana(MSVCRT.@)
1557 int CDECL _ismbbkana(unsigned int c)
1559 return _ismbbkana_l( c, NULL );
1562 /*********************************************************************
1563 * _ismbcdigit_l(MSVCRT.@)
1565 int CDECL _ismbcdigit_l(unsigned int ch, _locale_t locale)
1567 return _iswdigit_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1570 /*********************************************************************
1571 * _ismbcdigit(MSVCRT.@)
1573 int CDECL _ismbcdigit(unsigned int ch)
1575 return _ismbcdigit_l( ch, NULL );
1578 /*********************************************************************
1579 * _ismbcgraph_l(MSVCRT.@)
1581 int CDECL _ismbcgraph_l(unsigned int ch, _locale_t locale)
1583 return _iswgraph_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1586 /*********************************************************************
1587 * _ismbcgraph(MSVCRT.@)
1589 int CDECL _ismbcgraph(unsigned int ch)
1591 return _ismbcgraph_l( ch, NULL );
1594 /*********************************************************************
1595 * _ismbcalpha_l (MSVCRT.@)
1597 int CDECL _ismbcalpha_l(unsigned int ch, _locale_t locale)
1599 return _iswalpha_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1602 /*********************************************************************
1603 * _ismbcalpha (MSVCRT.@)
1605 int CDECL _ismbcalpha(unsigned int ch)
1607 return _ismbcalpha_l( ch, NULL );
1610 /*********************************************************************
1611 * _ismbclower_l (MSVCRT.@)
1613 int CDECL _ismbclower_l(unsigned int ch, _locale_t locale)
1615 return _iswlower_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1618 /*********************************************************************
1619 * _ismbclower (MSVCRT.@)
1621 int CDECL _ismbclower(unsigned int ch)
1623 return _ismbclower_l( ch, NULL );
1626 /*********************************************************************
1627 * _ismbcupper_l (MSVCRT.@)
1629 int CDECL _ismbcupper_l(unsigned int ch, _locale_t locale)
1631 return _iswupper_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1634 /*********************************************************************
1635 * _ismbcupper (MSVCRT.@)
1637 int CDECL _ismbcupper(unsigned int ch)
1639 return _ismbcupper_l( ch, NULL );
1642 /*********************************************************************
1643 * _ismbcsymbol(MSVCRT.@)
1645 int CDECL _ismbcsymbol(unsigned int ch)
1647 wchar_t wch = msvcrt_mbc_to_wc( ch );
1648 WORD ctype;
1649 if (!GetStringTypeW(CT_CTYPE3, &wch, 1, &ctype))
1651 WARN("GetStringTypeW failed on %x\n", ch);
1652 return 0;
1654 return ((ctype & C3_SYMBOL) != 0);
1657 /*********************************************************************
1658 * _ismbcalnum_l (MSVCRT.@)
1660 int CDECL _ismbcalnum_l(unsigned int ch, _locale_t locale)
1662 return _iswalnum_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1665 /*********************************************************************
1666 * _ismbcalnum (MSVCRT.@)
1668 int CDECL _ismbcalnum(unsigned int ch)
1670 return _ismbcalnum_l( ch, NULL );
1673 /*********************************************************************
1674 * _ismbcspace_l (MSVCRT.@)
1676 int CDECL _ismbcspace_l(unsigned int ch, _locale_t locale)
1678 return _iswspace_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1681 /*********************************************************************
1682 * _ismbcspace (MSVCRT.@)
1684 int CDECL _ismbcspace(unsigned int ch)
1686 return _ismbcspace_l( ch, NULL );
1689 /*********************************************************************
1690 * _ismbcprint_l (MSVCRT.@)
1692 int CDECL _ismbcprint_l(unsigned int ch, _locale_t locale)
1694 return _iswprint_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1697 /*********************************************************************
1698 * _ismbcprint (MSVCRT.@)
1700 int CDECL _ismbcprint(unsigned int ch)
1702 return _ismbcprint_l( ch, NULL );
1705 /*********************************************************************
1706 * _ismbcpunct_l (MSVCRT.@)
1708 int CDECL _ismbcpunct_l(unsigned int ch, _locale_t locale)
1710 return _iswpunct_l( msvcrt_mbc_to_wc_l(ch, locale), locale );
1713 /*********************************************************************
1714 * _ismbcpunct(MSVCRT.@)
1716 int CDECL _ismbcpunct(unsigned int ch)
1718 return _ismbcpunct_l( ch, NULL );
1721 /*********************************************************************
1722 * _ismbchira(MSVCRT.@)
1724 int CDECL _ismbchira(unsigned int c)
1726 if(get_mbcinfo()->mbcodepage == 932)
1728 /* Japanese/Hiragana, CP 932 */
1729 return (c >= 0x829f && c <= 0x82f1);
1731 return 0;
1734 /*********************************************************************
1735 * _ismbckata(MSVCRT.@)
1737 int CDECL _ismbckata(unsigned int c)
1739 if(get_mbcinfo()->mbcodepage == 932)
1741 /* Japanese/Katakana, CP 932 */
1742 return (c >= 0x8340 && c <= 0x8396 && c != 0x837f);
1744 return 0;
1748 /*********************************************************************
1749 * _ismbblead_l(MSVCRT.@)
1751 int CDECL _ismbblead_l(unsigned int c, _locale_t locale)
1753 pthreadmbcinfo mbcinfo;
1755 if(!locale)
1756 mbcinfo = get_mbcinfo();
1757 else
1758 mbcinfo = locale->mbcinfo;
1760 return (mbcinfo->mbctype[(c&0xff) + 1] & _M1) != 0;
1763 /*********************************************************************
1764 * _ismbblead(MSVCRT.@)
1766 int CDECL _ismbblead(unsigned int c)
1768 return _ismbblead_l(c, NULL);
1771 /*********************************************************************
1772 * _ismbbtrail_l(MSVCRT.@)
1774 int CDECL _ismbbtrail_l(unsigned int c, _locale_t locale)
1776 pthreadmbcinfo mbcinfo;
1778 if(!locale)
1779 mbcinfo = get_mbcinfo();
1780 else
1781 mbcinfo = locale->mbcinfo;
1783 return (mbcinfo->mbctype[(c&0xff) + 1] & _M2) != 0;
1786 /*********************************************************************
1787 * _ismbbtrail(MSVCRT.@)
1789 int CDECL _ismbbtrail(unsigned int c)
1791 return _ismbbtrail_l(c, NULL);
1794 /*********************************************************************
1795 * _ismbclegal_l(MSVCRT.@)
1797 int CDECL _ismbclegal_l(unsigned int c, _locale_t locale)
1799 return _ismbblead_l(HIBYTE(c), locale) && _ismbbtrail_l(LOBYTE(c), locale);
1802 /*********************************************************************
1803 * _ismbclegal(MSVCRT.@)
1805 int CDECL _ismbclegal(unsigned int c)
1807 return _ismbclegal_l(c, NULL);
1810 /*********************************************************************
1811 * _ismbslead_l(MSVCRT.@)
1813 int CDECL _ismbslead_l(const unsigned char* start, const unsigned char* str, _locale_t locale)
1815 pthreadmbcinfo mbcinfo;
1816 int lead = 0;
1818 if (!MSVCRT_CHECK_PMT(start && str))
1819 return 0;
1821 if(locale)
1822 mbcinfo = locale->mbcinfo;
1823 else
1824 mbcinfo = get_mbcinfo();
1826 if(!mbcinfo->ismbcodepage)
1827 return 0;
1829 /* Lead bytes can also be trail bytes so we need to analyse the string
1831 while (start <= str)
1833 if (!*start)
1834 return 0;
1835 lead = !lead && _ismbblead_l(*start, locale);
1836 start++;
1839 return lead ? -1 : 0;
1842 /*********************************************************************
1843 * _ismbslead(MSVCRT.@)
1845 int CDECL _ismbslead(const unsigned char* start, const unsigned char* str)
1847 return _ismbslead_l(start, str, NULL);
1850 /*********************************************************************
1851 * _ismbstrail(MSVCRT.@)
1853 int CDECL _ismbstrail(const unsigned char* start, const unsigned char* str)
1855 /* Note: this function doesn't check _ismbbtrail */
1856 if ((str > start) && _ismbslead(start, str-1))
1857 return -1;
1858 else
1859 return 0;
1862 /*********************************************************************
1863 * _mbbtype_l(MSVCRT.@)
1865 int CDECL _mbbtype_l(unsigned char c, int type, _locale_t locale)
1867 if (type == 1)
1868 return _ismbbtrail_l(c, locale) ? _MBC_TRAIL : _MBC_ILLEGAL;
1869 else
1870 return _ismbblead_l(c, locale) ? _MBC_LEAD
1871 : _isprint_l(c, locale) ? _MBC_SINGLE : _MBC_ILLEGAL;
1874 /*********************************************************************
1875 * _mbbtype(MSVCRT.@)
1877 int CDECL _mbbtype(unsigned char c, int type)
1879 return _mbbtype_l(c, type, NULL);
1882 /*********************************************************************
1883 * _mbsbtype (MSVCRT.@)
1885 int CDECL _mbsbtype(const unsigned char *str, size_t count)
1887 int lead = 0;
1888 const unsigned char *end = str + count;
1890 /* Lead bytes can also be trail bytes so we need to analyse the string.
1891 * Also we must return _MBC_ILLEGAL for chars past the end of the string
1893 while (str < end) /* Note: we skip the last byte - will check after the loop */
1895 if (!*str)
1896 return _MBC_ILLEGAL;
1897 lead = get_mbcinfo()->ismbcodepage && !lead && _ismbblead(*str);
1898 str++;
1901 if (lead)
1902 if (_ismbbtrail(*str))
1903 return _MBC_TRAIL;
1904 else
1905 return _MBC_ILLEGAL;
1906 else
1907 if (_ismbblead(*str))
1908 return _MBC_LEAD;
1909 else
1910 return _MBC_SINGLE;
1913 /*********************************************************************
1914 * _mbsset(MSVCRT.@)
1916 unsigned char* CDECL _mbsset(unsigned char* str, unsigned int c)
1918 unsigned char* ret = str;
1920 if(!get_mbcinfo()->ismbcodepage || c < 256)
1921 return u__strset(str, c); /* ASCII CP or SB char */
1923 c &= 0xffff; /* Strip high bits */
1925 while(str[0] && str[1])
1927 *str++ = c >> 8;
1928 *str++ = c & 0xff;
1930 if(str[0])
1931 str[0] = '\0'; /* FIXME: OK to shorten? */
1933 return ret;
1936 /*********************************************************************
1937 * _mbsnbset(MSVCRT.@)
1939 unsigned char* CDECL _mbsnbset(unsigned char *str, unsigned int c, size_t len)
1941 unsigned char *ret = str;
1943 if(!len)
1944 return ret;
1946 if(!get_mbcinfo()->ismbcodepage || c < 256)
1947 return u__strnset(str, c, len); /* ASCII CP or SB char */
1949 c &= 0xffff; /* Strip high bits */
1951 while(str[0] && str[1] && (len > 1))
1953 *str++ = c >> 8;
1954 len--;
1955 *str++ = c & 0xff;
1956 len--;
1958 if(len && str[0]) {
1959 /* as per msdn pad with a blank character */
1960 str[0] = ' ';
1963 return ret;
1966 /*********************************************************************
1967 * _mbsnset(MSVCRT.@)
1969 unsigned char* CDECL _mbsnset(unsigned char* str, unsigned int c, size_t len)
1971 unsigned char *ret = str;
1973 if(!len)
1974 return ret;
1976 if(!get_mbcinfo()->ismbcodepage || c < 256)
1977 return u__strnset(str, c, len); /* ASCII CP or SB char */
1979 c &= 0xffff; /* Strip high bits */
1981 while(str[0] && str[1] && len--)
1983 *str++ = c >> 8;
1984 *str++ = c & 0xff;
1986 if(len && str[0])
1987 str[0] = '\0'; /* FIXME: OK to shorten? */
1989 return ret;
1992 /*********************************************************************
1993 * _mbsnccnt(MSVCRT.@)
1994 * 'c' is for 'character'.
1996 size_t CDECL _mbsnccnt(const unsigned char* str, size_t len)
1998 size_t ret;
1999 if(get_mbcinfo()->ismbcodepage)
2001 ret=0;
2002 while(*str && len-- > 0)
2004 if(_ismbblead(*str))
2006 if (!len)
2007 break;
2008 len--;
2009 str++;
2011 str++;
2012 ret++;
2014 return ret;
2016 ret=u_strlen(str);
2017 return min(ret, len); /* ASCII CP */
2020 /*********************************************************************
2021 * _mbsnbcnt(MSVCRT.@)
2022 * 'b' is for byte count.
2024 size_t CDECL _mbsnbcnt(const unsigned char* str, size_t len)
2026 size_t ret;
2027 if(get_mbcinfo()->ismbcodepage)
2029 const unsigned char* xstr = str;
2030 while(*xstr && len-- > 0)
2032 if (_ismbblead(*xstr++))
2033 xstr++;
2035 return xstr-str;
2037 ret=u_strlen(str);
2038 return min(ret, len); /* ASCII CP */
2041 /*********************************************************************
2042 * _mbsnbcat(MSVCRT.@)
2044 unsigned char* CDECL _mbsnbcat(unsigned char* dst, const unsigned char* src, size_t len)
2046 if(get_mbcinfo()->ismbcodepage)
2048 unsigned char *res = dst;
2049 while (*dst) {
2050 if (_ismbblead(*dst++)) {
2051 if (*dst) {
2052 dst++;
2053 } else {
2054 /* as per msdn overwrite the lead byte in front of '\0' */
2055 dst--;
2056 break;
2060 while (*src && len--) *dst++ = *src++;
2061 *dst = '\0';
2062 return res;
2064 return u_strncat(dst, src, len); /* ASCII CP */
2067 int CDECL _mbsnbcat_s(unsigned char *dst, size_t size, const unsigned char *src, size_t len)
2069 unsigned char *ptr = dst;
2070 size_t i;
2072 if (!dst && !size && !src && !len)
2073 return 0;
2075 if (!dst || !size || !src)
2077 if (dst && size)
2078 *dst = '\0';
2080 *_errno() = EINVAL;
2081 return EINVAL;
2084 /* Find the null terminator of the destination buffer. */
2085 while (size && *ptr)
2086 size--, ptr++;
2088 if (!size)
2090 *dst = '\0';
2091 *_errno() = EINVAL;
2092 return EINVAL;
2095 /* If necessary, check that the character preceding the null terminator is
2096 * a lead byte and move the pointer back by one for later overwrite. */
2097 if (ptr != dst && get_mbcinfo()->ismbcodepage && _ismbblead(*(ptr - 1)))
2098 size++, ptr--;
2100 for (i = 0; *src && i < len; i++)
2102 *ptr++ = *src++;
2103 size--;
2105 if (!size)
2107 *dst = '\0';
2108 *_errno() = ERANGE;
2109 return ERANGE;
2113 *ptr = '\0';
2114 return 0;
2117 /*********************************************************************
2118 * _mbsncat(MSVCRT.@)
2120 unsigned char* CDECL _mbsncat(unsigned char* dst, const unsigned char* src, size_t len)
2122 if(get_mbcinfo()->ismbcodepage)
2124 unsigned char *res = dst;
2125 while (*dst)
2127 if (_ismbblead(*dst++))
2128 dst++;
2130 while (*src && len--)
2132 *dst++ = *src;
2133 if(_ismbblead(*src++))
2134 *dst++ = *src++;
2136 *dst = '\0';
2137 return res;
2139 return u_strncat(dst, src, len); /* ASCII CP */
2143 /*********************************************************************
2144 * _mbslwr(MSVCRT.@)
2146 unsigned char* CDECL _mbslwr(unsigned char* s)
2148 unsigned char *ret = s;
2149 if (!s)
2150 return NULL;
2151 if (get_mbcinfo()->ismbcodepage)
2153 unsigned int c;
2154 while (*s)
2156 c = _mbctolower(_mbsnextc(s));
2157 /* Note that I assume that the size of the character is unchanged */
2158 if (c > 255)
2160 *s++=(c>>8);
2161 c=c & 0xff;
2163 *s++=c;
2166 else for ( ; *s; s++) *s = _tolower_l(*s, NULL);
2167 return ret;
2170 /*********************************************************************
2171 * _mbslwr_s_l(MSVCRT.@)
2173 int CDECL _mbslwr_s_l(unsigned char* s, size_t len, _locale_t locale)
2175 unsigned char *p = s;
2177 if (!s && !len)
2178 return 0;
2179 if (!MSVCRT_CHECK_PMT(s && len))
2180 return EINVAL;
2182 if (get_mbcinfo()->ismbcodepage)
2184 unsigned int c;
2185 for ( ; *s && len > 0; len--)
2187 c = _mbctolower_l(_mbsnextc_l(s, locale), locale);
2188 /* Note that I assume that the size of the character is unchanged */
2189 if (c > 255)
2191 *s++=(c>>8);
2192 c=c & 0xff;
2194 *s++=c;
2197 else
2199 for ( ; *s && len > 0; s++, len--)
2200 *s = _tolower_l(*s, locale);
2203 if (!MSVCRT_CHECK_PMT(len))
2205 *p = 0;
2206 return EINVAL;
2208 *s = 0;
2209 return 0;
2212 /*********************************************************************
2213 * _mbslwr_s(MSVCRT.@)
2215 int CDECL _mbslwr_s(unsigned char* str, size_t len)
2217 return _mbslwr_s_l(str, len, NULL);
2220 /*********************************************************************
2221 * _mbsupr(MSVCRT.@)
2223 unsigned char* CDECL _mbsupr(unsigned char* s)
2225 unsigned char *ret = s;
2226 if (!s)
2227 return NULL;
2228 if (get_mbcinfo()->ismbcodepage)
2230 unsigned int c;
2231 while (*s)
2233 c = _mbctoupper(_mbsnextc(s));
2234 /* Note that I assume that the size of the character is unchanged */
2235 if (c > 255)
2237 *s++=(c>>8);
2238 c=c & 0xff;
2240 *s++=c;
2243 else for ( ; *s; s++) *s = _toupper_l(*s, NULL);
2244 return ret;
2247 /*********************************************************************
2248 * _mbsupr_s_l(MSVCRT.@)
2250 int CDECL _mbsupr_s_l(unsigned char* s, size_t len, _locale_t locale)
2252 unsigned char *p = s;
2254 if (!s && !len)
2255 return 0;
2256 if (!MSVCRT_CHECK_PMT(s && len))
2257 return EINVAL;
2259 if (get_mbcinfo()->ismbcodepage)
2261 unsigned int c;
2262 for ( ; *s && len > 0; len--)
2264 c = _mbctoupper_l(_mbsnextc_l(s, locale), locale);
2265 /* Note that I assume that the size of the character is unchanged */
2266 if (c > 255)
2268 *s++=(c>>8);
2269 c=c & 0xff;
2271 *s++=c;
2274 else
2276 for ( ; *s && len > 0; s++, len--)
2277 *s = _toupper_l(*s, locale);
2280 if (!MSVCRT_CHECK_PMT(len))
2282 *p = 0;
2283 return EINVAL;
2285 *s = 0;
2286 return 0;
2289 /*********************************************************************
2290 * _mbsupr_s(MSVCRT.@)
2292 int CDECL _mbsupr_s(unsigned char* s, size_t len)
2294 return _mbsupr_s_l(s, len, NULL);
2297 /*********************************************************************
2298 * _mbsspn_l (MSVCRT.@)
2300 size_t CDECL _mbsspn_l(const unsigned char* string,
2301 const unsigned char* set, _locale_t locale)
2303 const unsigned char *p, *q;
2305 for (p = string; *p; p++)
2307 for (q = set; *q; q++)
2309 if (_ismbblead_l(*q, locale))
2311 /* duplicate a bug in native implementation */
2312 if (!q[1]) break;
2314 if (p[0] == q[0] && p[1] == q[1])
2316 p++;
2317 break;
2319 q++;
2321 else
2323 if (p[0] == q[0]) break;
2326 if (!*q) break;
2328 return p - string;
2331 /*********************************************************************
2332 * _mbsspn (MSVCRT.@)
2334 size_t CDECL _mbsspn(const unsigned char* string, const unsigned char* set)
2336 return _mbsspn_l(string, set, NULL);
2339 /*********************************************************************
2340 * _mbsspnp (MSVCRT.@)
2342 unsigned char* CDECL _mbsspnp(const unsigned char* string, const unsigned char* set)
2344 string += _mbsspn( string, set );
2345 return *string ? (unsigned char*)string : NULL;
2348 /*********************************************************************
2349 * _mbscspn_l (MSVCRT.@)
2351 size_t CDECL _mbscspn_l(const unsigned char* str,
2352 const unsigned char* cmp, _locale_t locale)
2354 const unsigned char *p, *q;
2356 for (p = str; *p; p++)
2358 for (q = cmp; *q; q++)
2360 if (_ismbblead_l(*q, locale))
2362 /* duplicate a bug in native implementation */
2363 if (!q[1]) return 0;
2365 if (p[0] == q[0] && p[1] == q[1])
2366 return p - str;
2367 q++;
2369 else if (p[0] == q[0])
2370 return p - str;
2373 return p - str;
2376 /*********************************************************************
2377 * _mbscspn (MSVCRT.@)
2379 size_t CDECL _mbscspn(const unsigned char* str, const unsigned char* cmp)
2381 return _mbscspn_l(str, cmp, NULL);
2384 /*********************************************************************
2385 * _mbsrev (MSVCRT.@)
2387 unsigned char* CDECL _mbsrev(unsigned char* str)
2389 int i, len = _mbslen(str);
2390 unsigned char *p, *temp=malloc(len*2);
2392 if(!temp)
2393 return str;
2395 /* unpack multibyte string to temp buffer */
2396 p=str;
2397 for(i=0; i<len; i++)
2399 if (_ismbblead(*p))
2401 temp[i*2]=*p++;
2402 temp[i*2+1]=*p++;
2404 else
2406 temp[i*2]=*p++;
2407 temp[i*2+1]=0;
2411 /* repack it in the reverse order */
2412 p=str;
2413 for(i=len-1; i>=0; i--)
2415 if(_ismbblead(temp[i*2]))
2417 *p++=temp[i*2];
2418 *p++=temp[i*2+1];
2420 else
2422 *p++=temp[i*2];
2426 free(temp);
2428 return str;
2431 /*********************************************************************
2432 * _mbspbrk (MSVCRT.@)
2434 unsigned char* CDECL _mbspbrk(const unsigned char* str, const unsigned char* accept)
2436 const unsigned char* p;
2438 while(*str)
2440 for(p = accept; *p; p += (_ismbblead(*p)?2:1) )
2442 if (*p == *str)
2443 if( !_ismbblead(*p) || ( *(p+1) == *(str+1) ) )
2444 return (unsigned char*)str;
2446 str += (_ismbblead(*str)?2:1);
2448 return NULL;
2453 * Functions depending on locale codepage
2456 /*********************************************************************
2457 * mblen(MSVCRT.@)
2458 * REMARKS
2459 * Unlike most of the multibyte string functions this function uses
2460 * the locale codepage, not the codepage set by _setmbcp
2462 int CDECL mblen(const char* str, size_t size)
2464 if (str && *str && size)
2466 if(get_locinfo()->mb_cur_max == 1)
2467 return 1; /* ASCII CP */
2469 return !isleadbyte((unsigned char)*str) ? 1 : (size>1 ? 2 : -1);
2471 return 0;
2474 /*********************************************************************
2475 * mbrlen(MSVCRT.@)
2477 size_t CDECL mbrlen(const char *str, size_t len, mbstate_t *state)
2479 mbstate_t s = (state ? *state : 0);
2480 size_t ret;
2482 if(!len || !str || !*str)
2483 return 0;
2485 if(get_locinfo()->mb_cur_max == 1) {
2486 return 1;
2487 }else if(!s && isleadbyte((unsigned char)*str)) {
2488 if(len == 1) {
2489 s = (unsigned char)*str;
2490 ret = -2;
2491 }else {
2492 ret = 2;
2494 }else if(!s) {
2495 ret = 1;
2496 }else {
2497 s = 0;
2498 ret = 2;
2501 if(state)
2502 *state = s;
2503 return ret;
2506 /*********************************************************************
2507 * _mbstrlen_l(MSVCRT.@)
2509 size_t CDECL _mbstrlen_l(const char* str, _locale_t locale)
2511 pthreadlocinfo locinfo;
2513 if(!locale)
2514 locinfo = get_locinfo();
2515 else
2516 locinfo = locale->locinfo;
2518 if(locinfo->mb_cur_max > 1) {
2519 size_t len;
2520 len = MultiByteToWideChar(locinfo->lc_codepage, MB_ERR_INVALID_CHARS,
2521 str, -1, NULL, 0);
2522 if (!len) {
2523 *_errno() = EILSEQ;
2524 return -1;
2526 return len - 1;
2529 return strlen(str);
2532 /*********************************************************************
2533 * _mbstrlen(MSVCRT.@)
2535 size_t CDECL _mbstrlen(const char* str)
2537 return _mbstrlen_l(str, NULL);
2540 /*********************************************************************
2541 * _mbtowc_l(MSVCRT.@)
2543 int CDECL _mbtowc_l(wchar_t *dst, const char* str, size_t n, _locale_t locale)
2545 pthreadlocinfo locinfo;
2546 wchar_t tmpdst;
2548 if(!locale)
2549 locinfo = get_locinfo();
2550 else
2551 locinfo = locale->locinfo;
2553 if(n <= 0 || !str)
2554 return 0;
2556 if(!*str) {
2557 if(dst) *dst = 0;
2558 return 0;
2561 if(!locinfo->lc_codepage) {
2562 if(dst) *dst = (unsigned char)*str;
2563 return 1;
2565 if(n>=2 && _isleadbyte_l((unsigned char)*str, locale)) {
2566 if(!MultiByteToWideChar(locinfo->lc_codepage, 0, str, 2, &tmpdst, 1))
2567 return -1;
2568 if(dst) *dst = tmpdst;
2569 return 2;
2571 if(!MultiByteToWideChar(locinfo->lc_codepage, 0, str, 1, &tmpdst, 1))
2572 return -1;
2573 if(dst) *dst = tmpdst;
2574 return 1;
2577 /*********************************************************************
2578 * mbtowc(MSVCRT.@)
2580 int CDECL mbtowc(wchar_t *dst, const char* str, size_t n)
2582 return _mbtowc_l(dst, str, n, NULL);
2585 /*********************************************************************
2586 * btowc(MSVCRT.@)
2588 wint_t CDECL btowc(int c)
2590 unsigned char letter = c;
2591 wchar_t ret;
2593 if(c == EOF)
2594 return WEOF;
2595 if(!get_locinfo()->lc_codepage)
2596 return c & 255;
2597 if(!MultiByteToWideChar(get_locinfo()->lc_codepage,
2598 MB_ERR_INVALID_CHARS, (LPCSTR)&letter, 1, &ret, 1))
2599 return WEOF;
2601 return ret;
2604 /*********************************************************************
2605 * mbrtowc(MSVCRT.@)
2607 size_t CDECL mbrtowc(wchar_t *dst, const char *str,
2608 size_t n, mbstate_t *state)
2610 pthreadlocinfo locinfo = get_locinfo();
2611 mbstate_t s = (state ? *state : 0);
2612 char tmpstr[2];
2613 int len = 0;
2615 if(dst)
2616 *dst = 0;
2618 if(!n || !str || !*str)
2619 return 0;
2621 if(locinfo->mb_cur_max == 1) {
2622 tmpstr[len++] = *str;
2623 }else if(!s && isleadbyte((unsigned char)*str)) {
2624 if(n == 1) {
2625 s = (unsigned char)*str;
2626 len = -2;
2627 }else {
2628 tmpstr[0] = str[0];
2629 tmpstr[1] = str[1];
2630 len = 2;
2632 }else if(!s) {
2633 tmpstr[len++] = *str;
2634 }else {
2635 tmpstr[0] = s;
2636 tmpstr[1] = *str;
2637 len = 2;
2638 s = 0;
2641 if(len > 0) {
2642 if(!MultiByteToWideChar(locinfo->lc_codepage, 0, tmpstr, len, dst, dst ? 1 : 0))
2643 len = -1;
2646 if(state)
2647 *state = s;
2648 return len;
2651 /*********************************************************************
2652 * _mbstowcs_l(MSVCRT.@)
2654 size_t CDECL _mbstowcs_l(wchar_t *wcstr, const char *mbstr,
2655 size_t count, _locale_t locale)
2657 pthreadlocinfo locinfo;
2658 size_t i, size;
2660 if(!mbstr) {
2661 *_errno() = EINVAL;
2662 return -1;
2665 if(!locale)
2666 locinfo = get_locinfo();
2667 else
2668 locinfo = locale->locinfo;
2670 if(!locinfo->lc_codepage) {
2671 if(!wcstr)
2672 return strlen(mbstr);
2674 for(i=0; i<count; i++) {
2675 wcstr[i] = (unsigned char)mbstr[i];
2676 if(!wcstr[i]) break;
2678 return i;
2681 /* Ignore count parameter */
2682 if(!wcstr) {
2683 size = MultiByteToWideChar(locinfo->lc_codepage,
2684 MB_ERR_INVALID_CHARS, mbstr, -1, NULL, 0);
2685 if(!size) {
2686 *_errno() = EILSEQ;
2687 return -1;
2689 return size - 1;
2692 for(i=0, size=0; i<count; i++) {
2693 if(mbstr[size] == '\0')
2694 break;
2696 size += (_isleadbyte_l((unsigned char)mbstr[size], locale) ? 2 : 1);
2699 if(size) {
2700 size = MultiByteToWideChar(locinfo->lc_codepage,
2701 MB_ERR_INVALID_CHARS, mbstr, size, wcstr, count);
2702 if(!size) {
2703 if(count) wcstr[0] = '\0';
2704 *_errno() = EILSEQ;
2705 return -1;
2709 if(size<count)
2710 wcstr[size] = '\0';
2712 return size;
2715 /*********************************************************************
2716 * mbstowcs(MSVCRT.@)
2718 size_t CDECL mbstowcs(wchar_t *wcstr,
2719 const char *mbstr, size_t count)
2721 return _mbstowcs_l(wcstr, mbstr, count, NULL);
2724 /*********************************************************************
2725 * _mbstowcs_s_l(MSVCRT.@)
2727 int CDECL _mbstowcs_s_l(size_t *ret, wchar_t *wcstr,
2728 size_t size, const char *mbstr, size_t count, _locale_t locale)
2730 size_t conv;
2731 int err = 0;
2733 if(!wcstr && !size) {
2734 conv = _mbstowcs_l(NULL, mbstr, 0, locale);
2735 if(ret)
2736 *ret = conv+1;
2737 return 0;
2740 if (!MSVCRT_CHECK_PMT(wcstr != NULL)) return EINVAL;
2741 if (!MSVCRT_CHECK_PMT(mbstr != NULL)) {
2742 if(size) wcstr[0] = '\0';
2743 return EINVAL;
2746 if(count==_TRUNCATE || size<count)
2747 conv = size;
2748 else
2749 conv = count;
2751 conv = _mbstowcs_l(wcstr, mbstr, conv, locale);
2752 if(conv<size)
2753 wcstr[conv++] = '\0';
2754 else if(conv==size && count==_TRUNCATE && wcstr[conv-1]!='\0') {
2755 wcstr[conv-1] = '\0';
2756 err = STRUNCATE;
2757 }else if(conv==size && wcstr[conv-1]!='\0') {
2758 MSVCRT_INVALID_PMT("wcstr[size] is too small", ERANGE);
2759 if(size)
2760 wcstr[0] = '\0';
2761 return ERANGE;
2764 if(ret)
2765 *ret = conv;
2766 return err;
2769 /*********************************************************************
2770 * mbstowcs_s(MSVCRT.@)
2772 int CDECL _mbstowcs_s(size_t *ret, wchar_t *wcstr,
2773 size_t size, const char *mbstr, size_t count)
2775 return _mbstowcs_s_l(ret, wcstr, size, mbstr, count, NULL);
2778 /*********************************************************************
2779 * mbsrtowcs(MSVCRT.@)
2781 size_t CDECL mbsrtowcs(wchar_t *wcstr,
2782 const char **pmbstr, size_t count, mbstate_t *state)
2784 mbstate_t s = (state ? *state : 0);
2785 wchar_t tmpdst;
2786 size_t ret = 0;
2787 const char *p;
2789 if(!MSVCRT_CHECK_PMT(pmbstr != NULL))
2790 return -1;
2792 p = *pmbstr;
2793 while(!wcstr || count>ret) {
2794 int ch_len = mbrtowc(&tmpdst, p, 2, &s);
2795 if(wcstr)
2796 wcstr[ret] = tmpdst;
2798 if(ch_len < 0) {
2799 return -1;
2800 }else if(ch_len == 0) {
2801 if(wcstr) *pmbstr = NULL;
2802 return ret;
2805 p += ch_len;
2806 ret++;
2809 if(wcstr) *pmbstr = p;
2810 return ret;
2813 /*********************************************************************
2814 * mbsrtowcs_s(MSVCRT.@)
2816 int CDECL mbsrtowcs_s(size_t *ret, wchar_t *wcstr, size_t len,
2817 const char **mbstr, size_t count, mbstate_t *state)
2819 size_t tmp;
2821 if(!ret) ret = &tmp;
2822 if(!MSVCRT_CHECK_PMT(!!wcstr == !!len)) {
2823 *ret = -1;
2824 return EINVAL;
2827 *ret = mbsrtowcs(wcstr, mbstr, count>len ? len : count, state);
2828 if(*ret == -1) {
2829 if(wcstr) *wcstr = 0;
2830 return *_errno();
2832 (*ret)++;
2833 if(*ret > len) {
2834 /* no place for terminating '\0' */
2835 if(wcstr) *wcstr = 0;
2836 return 0;
2838 if(wcstr) wcstr[(*ret)-1] = 0;
2839 return 0;
2842 /*********************************************************************
2843 * _mbctohira (MSVCRT.@)
2845 * Converts a sjis katakana character to hiragana.
2847 unsigned int CDECL _mbctohira(unsigned int c)
2849 if(_ismbckata(c) && c <= 0x8393)
2850 return (c - 0x8340 - (c >= 0x837f ? 1 : 0)) + 0x829f;
2851 return c;
2854 /*********************************************************************
2855 * _mbctokata (MSVCRT.@)
2857 * Converts a sjis hiragana character to katakana.
2859 unsigned int CDECL _mbctokata(unsigned int c)
2861 if(_ismbchira(c))
2862 return (c - 0x829f) + 0x8340 + (c >= 0x82de ? 1 : 0);
2863 return c;
2866 /*********************************************************************
2867 * _ismbcl0_l (MSVCRT.@)
2869 int CDECL _ismbcl0_l(unsigned int c, _locale_t locale)
2871 pthreadmbcinfo mbcinfo;
2873 if(!locale)
2874 mbcinfo = get_mbcinfo();
2875 else
2876 mbcinfo = locale->mbcinfo;
2878 if(mbcinfo->mbcodepage == 932)
2880 /* JIS non-Kanji */
2881 return _ismbclegal_l(c, locale) && c >= 0x8140 && c <= 0x889e;
2884 return 0;
2887 /*********************************************************************
2888 * _ismbcl0 (MSVCRT.@)
2890 int CDECL _ismbcl0(unsigned int c)
2892 return _ismbcl0_l(c, NULL);
2895 /*********************************************************************
2896 * _ismbcl1_l (MSVCRT.@)
2898 int CDECL _ismbcl1_l(unsigned int c, _locale_t locale)
2900 pthreadmbcinfo mbcinfo;
2902 if(!locale)
2903 mbcinfo = get_mbcinfo();
2904 else
2905 mbcinfo = locale->mbcinfo;
2907 if(mbcinfo->mbcodepage == 932)
2909 /* JIS level-1 */
2910 return _ismbclegal_l(c, locale) && c >= 0x889f && c <= 0x9872;
2913 return 0;
2916 /*********************************************************************
2917 * _ismbcl1 (MSVCRT.@)
2919 int CDECL _ismbcl1(unsigned int c)
2921 return _ismbcl1_l(c, NULL);
2924 /*********************************************************************
2925 * _ismbcl2_l (MSVCRT.@)
2927 int CDECL _ismbcl2_l(unsigned int c, _locale_t locale)
2929 pthreadmbcinfo mbcinfo;
2931 if(!locale)
2932 mbcinfo = get_mbcinfo();
2933 else
2934 mbcinfo = locale->mbcinfo;
2936 if(mbcinfo->mbcodepage == 932)
2938 /* JIS level-2 */
2939 return _ismbclegal_l(c, locale) && c >= 0x989f && c <= 0xeaa4;
2942 return 0;
2945 /*********************************************************************
2946 * _ismbcl2 (MSVCRT.@)
2948 int CDECL _ismbcl2(unsigned int c)
2950 return _ismbcl2_l(c, NULL);