push 9758e6fe7ae8fbab538c98c718d6619029bb3457
[wine/hacks.git] / dlls / msvcrt / mbcs.c
blobc99d16727f5d36f81fdc22c8a60b0bf107ec1ff1
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 "msvcrt.h"
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] = { 0 };
34 static int g_mbcp_is_multibyte = 0;
36 int MSVCRT___mb_cur_max = 1;
37 extern int MSVCRT___lc_collate_cp;
39 /* It seems that the data about valid trail bytes is not available from kernel32
40 * so we have to store is here. The format is the same as for lead bytes in CPINFO */
41 struct cp_extra_info_t
43 int cp;
44 BYTE TrailBytes[MAX_LEADBYTES];
47 static struct cp_extra_info_t g_cpextrainfo[] =
49 {932, {0x40, 0x7e, 0x80, 0xfc, 0, 0}},
50 {936, {0x40, 0xfe, 0, 0}},
51 {949, {0x41, 0xfe, 0, 0}},
52 {950, {0x40, 0x7e, 0xa1, 0xfe, 0, 0}},
53 {20932, {1, 255, 0, 0}}, /* seems to give different results on different systems */
54 {0, {1, 255, 0, 0}} /* match all with FIXME */
57 static MSVCRT_wchar_t msvcrt_mbc_to_wc(unsigned int ch)
59 MSVCRT_wchar_t chW;
60 char mbch[2];
61 int n_chars;
63 if (ch <= 0xff) {
64 mbch[0] = ch;
65 n_chars = 1;
66 } else {
67 mbch[0] = (ch >> 8) & 0xff;
68 mbch[1] = ch & 0xff;
69 n_chars = 2;
71 if (!MultiByteToWideChar(MSVCRT___lc_codepage, 0, mbch, n_chars, &chW, 1))
73 WARN("MultiByteToWideChar failed on %x\n", ch);
74 return 0;
76 return chW;
79 static inline size_t u_strlen( const unsigned char *str )
81 return strlen( (const char*) str );
84 static inline unsigned char* u_strncat( unsigned char* dst, const unsigned char* src, size_t len )
86 return (unsigned char*)strncat( (char*)dst, (const char*)src, len);
89 static inline int u_strcmp( const unsigned char *s1, const unsigned char *s2 )
91 return strcmp( (const char*)s1, (const char*)s2 );
94 static inline int u_strcasecmp( const unsigned char *s1, const unsigned char *s2 )
96 return strcasecmp( (const char*)s1, (const char*)s2 );
99 static inline int u_strncmp( const unsigned char *s1, const unsigned char *s2, size_t len )
101 return strncmp( (const char*)s1, (const char*)s2, len );
104 static inline int u_strncasecmp( const unsigned char *s1, const unsigned char *s2, size_t len )
106 return strncasecmp( (const char*)s1, (const char*)s2, len );
109 static inline unsigned char *u_strchr( const unsigned char *s, unsigned char x )
111 return (unsigned char*) strchr( (const char*)s, x );
114 static inline unsigned char *u_strrchr( const unsigned char *s, unsigned char x )
116 return (unsigned char*) strrchr( (const char*)s, x );
119 static inline unsigned char *u_strtok( unsigned char *s, const unsigned char *delim )
121 return (unsigned char*) strtok( (char*)s, (const char*)delim );
124 static inline unsigned char *u__strset( unsigned char *s, unsigned char c )
126 return (unsigned char*) _strset( (char*)s, c);
129 static inline unsigned char *u__strnset( unsigned char *s, unsigned char c, size_t len )
131 return (unsigned char*) _strnset( (char*)s, c, len );
134 static inline size_t u_strcspn( const unsigned char *s, const unsigned char *rej )
136 return strcspn( (const char *)s, (const char*)rej );
139 /*********************************************************************
140 * __p__mbctype (MSVCRT.@)
142 unsigned char* CDECL __p__mbctype(void)
144 return MSVCRT_mbctype;
147 /*********************************************************************
148 * __p___mb_cur_max(MSVCRT.@)
150 int* CDECL __p___mb_cur_max(void)
152 return &MSVCRT___mb_cur_max;
155 /*********************************************************************
156 * _setmbcp (MSVCRT.@)
158 int CDECL _setmbcp(int cp)
160 int newcp;
161 CPINFO cpi;
162 BYTE *bytes;
163 WORD chartypes[256];
164 WORD *curr_type;
165 char bufA[256];
166 WCHAR bufW[256];
167 int charcount;
168 int ret;
169 int i;
171 switch (cp)
173 case _MB_CP_ANSI:
174 newcp = GetACP();
175 break;
176 case _MB_CP_OEM:
177 newcp = GetOEMCP();
178 break;
179 case _MB_CP_LOCALE:
180 newcp = MSVCRT___lc_codepage;
181 break;
182 case _MB_CP_SBCS:
183 newcp = 20127; /* ASCII */
184 break;
185 default:
186 newcp = cp;
187 break;
190 if (!GetCPInfo(newcp, &cpi))
192 WARN("Codepage %d not found\n", newcp);
193 *MSVCRT__errno() = MSVCRT_EINVAL;
194 return -1;
197 /* setup the _mbctype */
198 memset(MSVCRT_mbctype, 0, sizeof(MSVCRT_mbctype));
200 bytes = cpi.LeadByte;
201 while (bytes[0] || bytes[1])
203 for (i = bytes[0]; i <= bytes[1]; i++)
204 MSVCRT_mbctype[i + 1] |= _M1;
205 bytes += 2;
208 if (cpi.MaxCharSize > 1)
210 /* trail bytes not available through kernel32 but stored in a structure in msvcrt */
211 struct cp_extra_info_t *cpextra = g_cpextrainfo;
213 g_mbcp_is_multibyte = 1;
214 while (TRUE)
216 if (cpextra->cp == 0 || cpextra->cp == newcp)
218 if (cpextra->cp == 0)
219 FIXME("trail bytes data not available for DBCS codepage %d - assuming all bytes\n", newcp);
221 bytes = cpextra->TrailBytes;
222 while (bytes[0] || bytes[1])
224 for (i = bytes[0]; i <= bytes[1]; i++)
225 MSVCRT_mbctype[i + 1] |= _M2;
226 bytes += 2;
228 break;
230 cpextra++;
233 else
234 g_mbcp_is_multibyte = 0;
236 /* we can't use GetStringTypeA directly because we don't have a locale - only a code page
238 charcount = 0;
239 for (i = 0; i < 256; i++)
240 if (!(MSVCRT_mbctype[i + 1] & _M1))
241 bufA[charcount++] = i;
243 ret = MultiByteToWideChar(newcp, 0, bufA, charcount, bufW, charcount);
244 if (ret != charcount)
245 ERR("MultiByteToWideChar of chars failed for cp %d, ret=%d (exp %d), error=%d\n", newcp, ret, charcount, GetLastError());
247 GetStringTypeW(CT_CTYPE1, bufW, charcount, chartypes);
249 curr_type = chartypes;
250 for (i = 0; i < 256; i++)
251 if (!(MSVCRT_mbctype[i + 1] & _M1))
253 if ((*curr_type) & C1_UPPER)
254 MSVCRT_mbctype[i + 1] |= _SBUP;
255 if ((*curr_type) & C1_LOWER)
256 MSVCRT_mbctype[i + 1] |= _SBLOW;
257 curr_type++;
260 if (newcp == 932) /* CP932 only - set _MP and _MS */
262 /* On Windows it's possible to calculate the _MP and _MS from CT_CTYPE1
263 * and CT_CTYPE3. But as of Wine 0.9.43 we return wrong values what makes
264 * it hard. As this is set only for codepage 932 we hardcode it what gives
265 * also faster execution.
267 for (i = 161; i <= 165; i++)
268 MSVCRT_mbctype[i + 1] |= _MP;
269 for (i = 166; i <= 223; i++)
270 MSVCRT_mbctype[i + 1] |= _MS;
273 MSVCRT___lc_collate_cp = MSVCRT___lc_codepage = newcp;
274 TRACE("(%d) -> %d\n", cp, MSVCRT___lc_codepage);
275 return 0;
278 /*********************************************************************
279 * _getmbcp (MSVCRT.@)
281 int CDECL _getmbcp(void)
283 return MSVCRT___lc_codepage;
286 /*********************************************************************
287 * _mbsnextc(MSVCRT.@)
289 unsigned int CDECL _mbsnextc(const unsigned char* str)
291 if(_ismbblead(*str))
292 return *str << 8 | str[1];
293 return *str;
296 /*********************************************************************
297 * _mbctolower(MSVCRT.@)
299 unsigned int CDECL _mbctolower(unsigned int c)
301 if (MSVCRT_isleadbyte(c))
303 FIXME("Handle MBC chars\n");
304 return c;
306 return tolower(c); /* ASCII CP or SB char */
309 /*********************************************************************
310 * _mbctoupper(MSVCRT.@)
312 unsigned int CDECL _mbctoupper(unsigned int c)
314 if (MSVCRT_isleadbyte(c))
316 FIXME("Handle MBC chars\n");
317 return c;
319 return toupper(c); /* ASCII CP or SB char */
322 /*********************************************************************
323 * _mbsdec(MSVCRT.@)
325 unsigned char* CDECL _mbsdec(const unsigned char* start, const unsigned char* cur)
327 if(MSVCRT___mb_cur_max > 1)
328 return (unsigned char *)(_ismbstrail(start,cur-1) ? cur - 2 : cur -1);
330 return (unsigned char *)cur - 1; /* ASCII CP or SB char */
333 /*********************************************************************
334 * _mbclen(MSVCRT.@)
336 unsigned int CDECL _mbclen(const unsigned char* str)
338 return _ismbblead(*str) ? 2 : 1;
341 /*********************************************************************
342 * _mbsinc(MSVCRT.@)
344 unsigned char* CDECL _mbsinc(const unsigned char* str)
346 return (unsigned char *)(str + _mbclen(str));
349 /*********************************************************************
350 * _mbsninc(MSVCRT.@)
352 unsigned char* CDECL _mbsninc(const unsigned char* str, MSVCRT_size_t num)
354 if(!str)
355 return NULL;
357 while (num > 0 && *str)
359 if (_ismbblead(*str))
361 if (!*(str+1))
362 break;
363 str++;
365 str++;
366 num--;
369 return (unsigned char*)str;
372 /*********************************************************************
373 * _mbslen(MSVCRT.@)
375 MSVCRT_size_t CDECL _mbslen(const unsigned char* str)
377 MSVCRT_size_t len = 0;
378 while(*str)
380 if (_ismbblead(*str))
382 str++;
383 if (!*str) /* count only full chars */
384 break;
386 str++;
387 len++;
389 return len;
392 /*********************************************************************
393 * _mbccpy(MSVCRT.@)
395 void CDECL _mbccpy(unsigned char* dest, const unsigned char* src)
397 *dest = *src;
398 if(_ismbblead(*src))
399 *++dest = *++src; /* MB char */
402 /*********************************************************************
403 * _mbsncpy(MSVCRT.@)
404 * REMARKS
405 * The parameter n is the number or characters to copy, not the size of
406 * the buffer. Use _mbsnbcpy for a function analogical to strncpy
408 unsigned char* CDECL _mbsncpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
410 unsigned char* ret = dst;
411 if(!n)
412 return dst;
413 if (g_mbcp_is_multibyte)
415 while (*src && n)
417 n--;
418 if (_ismbblead(*src))
420 if (!*(src+1))
422 *dst++ = 0;
423 *dst++ = 0;
424 break;
427 *dst++ = *src++;
430 *dst++ = *src++;
433 else
435 while (n)
437 n--;
438 if (!(*dst++ = *src++)) break;
441 while (n--) *dst++ = 0;
442 return ret;
445 /*********************************************************************
446 * _mbsnbcpy_s(MSVCRT.@)
447 * REMARKS
448 * Unlike _mbsnbcpy this function does not pad the rest of the dest
449 * string with 0
451 int CDECL _mbsnbcpy_s(unsigned char* dst, MSVCRT_size_t size, const unsigned char* src, MSVCRT_size_t n)
453 MSVCRT_size_t pos = 0;
455 if(!dst || size == 0)
456 return MSVCRT_EINVAL;
457 if(!src)
459 dst[0] = '\0';
460 return MSVCRT_EINVAL;
462 if(!n)
463 return 0;
465 if(g_mbcp_is_multibyte)
467 int is_lead = 0;
468 while (*src && n)
470 if(pos == size)
472 dst[0] = '\0';
473 return MSVCRT_ERANGE;
475 is_lead = (!is_lead && _ismbblead(*src));
476 n--;
477 dst[pos++] = *src++;
480 if (is_lead) /* if string ends with a lead, remove it */
481 dst[pos - 1] = 0;
483 else
485 while (n)
487 n--;
488 if(pos == size)
490 dst[0] = '\0';
491 return MSVCRT_ERANGE;
494 if(!(*src)) break;
495 dst[pos++] = *src++;
499 if(pos < size)
500 dst[pos] = '\0';
501 else
503 dst[0] = '\0';
504 return MSVCRT_ERANGE;
507 return 0;
510 /*********************************************************************
511 * _mbsnbcpy(MSVCRT.@)
512 * REMARKS
513 * Like strncpy this function doesn't enforce the string to be
514 * NUL-terminated
516 unsigned char* CDECL _mbsnbcpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
518 unsigned char* ret = dst;
519 if(!n)
520 return dst;
521 if(g_mbcp_is_multibyte)
523 int is_lead = 0;
524 while (*src && n)
526 is_lead = (!is_lead && _ismbblead(*src));
527 n--;
528 *dst++ = *src++;
531 if (is_lead) /* if string ends with a lead, remove it */
532 *(dst - 1) = 0;
534 else
536 while (n)
538 n--;
539 if (!(*dst++ = *src++)) break;
542 while (n--) *dst++ = 0;
543 return ret;
546 /*********************************************************************
547 * _mbscmp(MSVCRT.@)
549 int CDECL _mbscmp(const unsigned char* str, const unsigned char* cmp)
551 if(MSVCRT___mb_cur_max > 1)
553 unsigned int strc, cmpc;
554 do {
555 if(!*str)
556 return *cmp ? -1 : 0;
557 if(!*cmp)
558 return 1;
559 strc = _mbsnextc(str);
560 cmpc = _mbsnextc(cmp);
561 if(strc != cmpc)
562 return strc < cmpc ? -1 : 1;
563 str +=(strc > 255) ? 2 : 1;
564 cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
565 } while(1);
567 return u_strcmp(str, cmp); /* ASCII CP */
570 /*********************************************************************
571 * _mbsicoll(MSVCRT.@)
572 * FIXME: handle locales.
574 int CDECL _mbsicoll(const unsigned char* str, const unsigned char* cmp)
576 if(MSVCRT___mb_cur_max > 1)
578 unsigned int strc, cmpc;
579 do {
580 if(!*str)
581 return *cmp ? -1 : 0;
582 if(!*cmp)
583 return 1;
584 strc = _mbctolower(_mbsnextc(str));
585 cmpc = _mbctolower(_mbsnextc(cmp));
586 if(strc != cmpc)
587 return strc < cmpc ? -1 : 1;
588 str +=(strc > 255) ? 2 : 1;
589 cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
590 } while(1);
592 return u_strcasecmp(str, cmp); /* ASCII CP */
595 /*********************************************************************
596 * _mbscoll(MSVCRT.@)
597 * Performs a case-sensitive comparison according to the current code page
598 * RETURN
599 * _NLSCMPERROR if error
600 * FIXME: handle locales.
602 int CDECL _mbscoll(const unsigned char* str, const unsigned char* cmp)
604 if(MSVCRT___mb_cur_max > 1)
606 unsigned int strc, cmpc;
607 do {
608 if(!*str)
609 return *cmp ? -1 : 0;
610 if(!*cmp)
611 return 1;
612 strc = _mbsnextc(str);
613 cmpc = _mbsnextc(cmp);
614 if(strc != cmpc)
615 return strc < cmpc ? -1 : 1;
616 str +=(strc > 255) ? 2 : 1;
617 cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
618 } while(1);
620 return u_strcmp(str, cmp); /* ASCII CP */
624 /*********************************************************************
625 * _mbsicmp(MSVCRT.@)
627 int CDECL _mbsicmp(const unsigned char* str, const unsigned char* cmp)
629 if(MSVCRT___mb_cur_max > 1)
631 unsigned int strc, cmpc;
632 do {
633 if(!*str)
634 return *cmp ? -1 : 0;
635 if(!*cmp)
636 return 1;
637 strc = _mbctolower(_mbsnextc(str));
638 cmpc = _mbctolower(_mbsnextc(cmp));
639 if(strc != cmpc)
640 return strc < cmpc ? -1 : 1;
641 str +=(strc > 255) ? 2 : 1;
642 cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
643 } while(1);
645 return u_strcasecmp(str, cmp); /* ASCII CP */
648 /*********************************************************************
649 * _mbsncmp(MSVCRT.@)
651 int CDECL _mbsncmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
653 if(!len)
654 return 0;
656 if(MSVCRT___mb_cur_max > 1)
658 unsigned int strc, cmpc;
659 while(len--)
661 int inc;
662 if(!*str)
663 return *cmp ? -1 : 0;
664 if(!*cmp)
665 return 1;
666 strc = _mbsnextc(str);
667 cmpc = _mbsnextc(cmp);
668 if(strc != cmpc)
669 return strc < cmpc ? -1 : 1;
670 inc=(strc > 255) ? 2 : 1; /* Equal, use same increment */
671 str += inc;
672 cmp += inc;
674 return 0; /* Matched len chars */
676 return u_strncmp(str, cmp, len); /* ASCII CP */
679 /*********************************************************************
680 * _mbsnbcmp(MSVCRT.@)
682 int CDECL _mbsnbcmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
684 if (!len)
685 return 0;
686 if(MSVCRT___mb_cur_max > 1)
688 unsigned int strc, cmpc;
689 while (len)
691 int clen;
692 if(!*str)
693 return *cmp ? -1 : 0;
694 if(!*cmp)
695 return 1;
696 if (MSVCRT_isleadbyte(*str))
698 strc=(len>=2)?_mbsnextc(str):0;
699 clen=2;
701 else
703 strc=*str;
704 clen=1;
706 if (MSVCRT_isleadbyte(*cmp))
707 cmpc=(len>=2)?_mbsnextc(cmp):0;
708 else
709 cmpc=*str;
710 if(strc != cmpc)
711 return strc < cmpc ? -1 : 1;
712 len -= clen;
713 str += clen;
714 cmp += clen;
716 return 0; /* Matched len chars */
718 return u_strncmp(str,cmp,len);
721 /*********************************************************************
722 * _mbsnicmp(MSVCRT.@)
724 * Compare two multibyte strings case insensitively to 'len' characters.
726 int CDECL _mbsnicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
728 /* FIXME: No tolower() for mb strings yet */
729 if(MSVCRT___mb_cur_max > 1)
731 unsigned int strc, cmpc;
732 while(len--)
734 if(!*str)
735 return *cmp ? -1 : 0;
736 if(!*cmp)
737 return 1;
738 strc = _mbctolower(_mbsnextc(str));
739 cmpc = _mbctolower(_mbsnextc(cmp));
740 if(strc != cmpc)
741 return strc < cmpc ? -1 : 1;
742 str +=(strc > 255) ? 2 : 1;
743 cmp +=(strc > 255) ? 2 : 1; /* Equal, use same increment */
745 return 0; /* Matched len chars */
747 return u_strncasecmp(str, cmp, len); /* ASCII CP */
750 /*********************************************************************
751 * _mbsnbicmp(MSVCRT.@)
753 int CDECL _mbsnbicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
755 if (!len)
756 return 0;
757 if(MSVCRT___mb_cur_max > 1)
759 unsigned int strc, cmpc;
760 while (len)
762 int clen;
763 if(!*str)
764 return *cmp ? -1 : 0;
765 if(!*cmp)
766 return 1;
767 if (MSVCRT_isleadbyte(*str))
769 strc=(len>=2)?_mbsnextc(str):0;
770 clen=2;
772 else
774 strc=*str;
775 clen=1;
777 if (MSVCRT_isleadbyte(*cmp))
778 cmpc=(len>=2)?_mbsnextc(cmp):0;
779 else
780 cmpc=*str;
781 strc = _mbctolower(strc);
782 cmpc = _mbctolower(cmpc);
783 if(strc != cmpc)
784 return strc < cmpc ? -1 : 1;
785 len -= clen;
786 str += clen;
787 cmp += clen;
789 return 0; /* Matched len bytes */
791 return u_strncasecmp(str,cmp,len);
794 /*********************************************************************
795 * _mbscat (MSVCRT.@)
797 unsigned char * CDECL _mbscat( unsigned char *dst, const unsigned char *src )
799 strcat( (char *)dst, (const char *)src );
800 return dst;
803 /*********************************************************************
804 * _mbscpy (MSVCRT.@)
806 unsigned char* CDECL _mbscpy( unsigned char *dst, const unsigned char *src )
808 strcpy( (char *)dst, (const char *)src );
809 return dst;
812 /*********************************************************************
813 * _mbsstr (MSVCRT.@)
815 unsigned char * CDECL _mbsstr(const unsigned char *haystack, const unsigned char *needle)
817 return (unsigned char *)strstr( (const char *)haystack, (const char *)needle );
820 /*********************************************************************
821 * _mbschr(MSVCRT.@)
823 * Find a multibyte character in a multibyte string.
825 unsigned char* CDECL _mbschr(const unsigned char* s, unsigned int x)
827 if(MSVCRT___mb_cur_max > 1)
829 unsigned int c;
830 while (1)
832 c = _mbsnextc(s);
833 if (c == x)
834 return (unsigned char*)s;
835 if (!c)
836 return NULL;
837 s += c > 255 ? 2 : 1;
840 return u_strchr(s, x); /* ASCII CP */
843 /*********************************************************************
844 * _mbsrchr(MSVCRT.@)
846 unsigned char* CDECL _mbsrchr(const unsigned char* s, unsigned int x)
848 if(MSVCRT___mb_cur_max > 1)
850 unsigned int c;
851 unsigned char* match=NULL;
852 if(!s)
853 return NULL;
854 while (1) {
855 c = _mbsnextc(s);
856 if (c == x)
857 match=(unsigned char*)s;
858 if (!c)
859 return match;
860 s +=(c > 255) ? 2 : 1;
863 return u_strrchr(s, x);
866 /*********************************************************************
867 * _mbstok(MSVCRT.@)
869 * Find and extract tokens from strings
871 unsigned char* CDECL _mbstok(unsigned char *str, const unsigned char *delim)
873 thread_data_t *data = msvcrt_get_thread_data();
874 unsigned char *ret;
876 if(MSVCRT___mb_cur_max > 1)
878 unsigned int c;
880 if (!str)
881 if (!(str = data->mbstok_next)) return NULL;
883 while ((c = _mbsnextc(str)) && _mbschr(delim, c)) {
884 str += c > 255 ? 2 : 1;
886 if (!*str) return NULL;
887 ret = str++;
888 while ((c = _mbsnextc(str)) && !_mbschr(delim, c)) {
889 str += c > 255 ? 2 : 1;
891 if (*str) {
892 *str++ = 0;
893 if (c > 255) *str++ = 0;
895 data->mbstok_next = str;
896 return ret;
898 return u_strtok(str, delim); /* ASCII CP */
901 /*********************************************************************
902 * mbtowc(MSVCRT.@)
904 int CDECL MSVCRT_mbtowc(MSVCRT_wchar_t *dst, const char* str, MSVCRT_size_t n)
906 /* temp var needed because MultiByteToWideChar wants non NULL destination */
907 MSVCRT_wchar_t tmpdst = '\0';
909 if(n <= 0 || !str)
910 return 0;
911 if(!MultiByteToWideChar(CP_ACP, 0, str, n, &tmpdst, 1))
912 return -1;
913 if(dst)
914 *dst = tmpdst;
915 /* return the number of bytes from src that have been used */
916 if(!*str)
917 return 0;
918 if(n >= 2 && MSVCRT_isleadbyte(*str) && str[1])
919 return 2;
920 return 1;
923 /*********************************************************************
924 * _mbbtombc(MSVCRT.@)
926 unsigned int CDECL _mbbtombc(unsigned int c)
928 if(MSVCRT___mb_cur_max > 1 &&
929 ((c >= 0x20 && c <=0x7e) ||(c >= 0xa1 && c <= 0xdf)))
931 /* FIXME: I can't get this function to return anything
932 * different from what I pass it...
935 return c; /* ASCII CP or no MB char */
938 /*********************************************************************
939 * _mbbtype(MSVCRT.@)
941 int CDECL _mbbtype(unsigned char c, int type)
943 if (type == 1)
945 if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
946 return _MBC_SINGLE;
947 else if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
948 return _MBC_TRAIL;
949 else
950 return _MBC_ILLEGAL;
952 else
954 if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
955 return _MBC_SINGLE;
956 else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
957 return _MBC_LEAD;
958 else
959 return _MBC_ILLEGAL;
963 /*********************************************************************
964 * _ismbbkana(MSVCRT.@)
966 int CDECL _ismbbkana(unsigned int c)
968 /* FIXME: use lc_ctype when supported, not lc_all */
969 if(MSVCRT___lc_codepage == 932)
971 /* Japanese/Katakana, CP 932 */
972 return (c >= 0xa1 && c <= 0xdf);
974 return 0;
977 /*********************************************************************
978 * _ismbcdigit(MSVCRT.@)
980 int CDECL _ismbcdigit(unsigned int ch)
982 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
983 return (get_char_typeW( wch ) & C1_DIGIT);
986 /*********************************************************************
987 * _ismbcgraph(MSVCRT.@)
989 int CDECL _ismbcgraph(unsigned int ch)
991 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
992 return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA));
995 /*********************************************************************
996 * _ismbcalpha (MSVCRT.@)
998 int CDECL _ismbcalpha(unsigned int ch)
1000 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1001 return (get_char_typeW( wch ) & C1_ALPHA);
1004 /*********************************************************************
1005 * _ismbclower (MSVCRT.@)
1007 int CDECL _ismbclower(unsigned int ch)
1009 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1010 return (get_char_typeW( wch ) & C1_UPPER);
1013 /*********************************************************************
1014 * _ismbcupper (MSVCRT.@)
1016 int CDECL _ismbcupper(unsigned int ch)
1018 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1019 return (get_char_typeW( wch ) & C1_LOWER);
1022 /*********************************************************************
1023 * _ismbcsymbol(MSVCRT.@)
1025 int CDECL _ismbcsymbol(unsigned int ch)
1027 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1028 WORD ctype;
1029 if (!GetStringTypeW(CT_CTYPE3, &wch, 1, &ctype))
1031 WARN("GetStringTypeW failed on %x\n", ch);
1032 return 0;
1034 return ((ctype & C3_SYMBOL) != 0);
1037 /*********************************************************************
1038 * _ismbcalnum (MSVCRT.@)
1040 int CDECL _ismbcalnum(unsigned int ch)
1042 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1043 return (get_char_typeW( wch ) & (C1_ALPHA | C1_DIGIT));
1046 /*********************************************************************
1047 * _ismbcspace (MSVCRT.@)
1049 int CDECL _ismbcspace(unsigned int ch)
1051 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1052 return (get_char_typeW( wch ) & C1_SPACE);
1055 /*********************************************************************
1056 * _ismbcprint (MSVCRT.@)
1058 int CDECL _ismbcprint(unsigned int ch)
1060 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1061 return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA | C1_SPACE));
1064 /*********************************************************************
1065 * _ismbcpunct(MSVCRT.@)
1067 int CDECL _ismbcpunct(unsigned int ch)
1069 MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1070 return (get_char_typeW( wch ) & C1_PUNCT);
1073 /*********************************************************************
1074 * _ismbchira(MSVCRT.@)
1076 int CDECL _ismbchira(unsigned int c)
1078 /* FIXME: use lc_ctype when supported, not lc_all */
1079 if(MSVCRT___lc_codepage == 932)
1081 /* Japanese/Hiragana, CP 932 */
1082 return (c >= 0x829f && c <= 0x82f1);
1084 return 0;
1087 /*********************************************************************
1088 * _ismbckata(MSVCRT.@)
1090 int CDECL _ismbckata(unsigned int c)
1092 /* FIXME: use lc_ctype when supported, not lc_all */
1093 if(MSVCRT___lc_codepage == 932)
1095 if(c < 256)
1096 return _ismbbkana(c);
1097 /* Japanese/Katakana, CP 932 */
1098 return (c >= 0x8340 && c <= 0x8396 && c != 0x837f);
1100 return 0;
1103 /*********************************************************************
1104 * _ismbblead(MSVCRT.@)
1106 int CDECL _ismbblead(unsigned int c)
1108 return (MSVCRT_mbctype[(c&0xff) + 1] & _M1) != 0;
1112 /*********************************************************************
1113 * _ismbbtrail(MSVCRT.@)
1115 int CDECL _ismbbtrail(unsigned int c)
1117 return (MSVCRT_mbctype[(c&0xff) + 1] & _M2) != 0;
1120 /*********************************************************************
1121 * _ismbslead(MSVCRT.@)
1123 int CDECL _ismbslead(const unsigned char* start, const unsigned char* str)
1125 int lead = 0;
1127 if(!g_mbcp_is_multibyte)
1128 return 0;
1130 /* Lead bytes can also be trail bytes so we need to analyse the string
1132 while (start <= str)
1134 if (!*start)
1135 return 0;
1136 lead = !lead && _ismbblead(*start);
1137 start++;
1140 return lead ? -1 : 0;
1143 /*********************************************************************
1144 * _ismbstrail(MSVCRT.@)
1146 int CDECL _ismbstrail(const unsigned char* start, const unsigned char* str)
1148 /* Note: this function doesn't check _ismbbtrail */
1149 if ((str > start) && _ismbslead(start, str-1))
1150 return -1;
1151 else
1152 return 0;
1155 /*********************************************************************
1156 * _mbsbtype (MSVCRT.@)
1158 int CDECL _mbsbtype(const unsigned char *str, MSVCRT_size_t count)
1160 int lead = 0;
1161 const unsigned char *end = str + count;
1162 int mbcp = g_mbcp_is_multibyte;
1164 /* Lead bytes can also be trail bytes so we need to analyse the string.
1165 * Also we must return _MBC_ILLEGAL for chars past the end of the string
1167 while (str < end) /* Note: we skip the last byte - will check after the loop */
1169 if (!*str)
1170 return _MBC_ILLEGAL;
1171 lead = mbcp && !lead && _ismbblead(*str);
1172 str++;
1175 if (lead)
1176 if (_ismbbtrail(*str))
1177 return _MBC_TRAIL;
1178 else
1179 return _MBC_ILLEGAL;
1180 else
1181 if (_ismbblead(*str))
1182 return _MBC_LEAD;
1183 else
1184 return _MBC_SINGLE;
1187 /*********************************************************************
1188 * _mbsset(MSVCRT.@)
1190 unsigned char* CDECL _mbsset(unsigned char* str, unsigned int c)
1192 unsigned char* ret = str;
1194 if(MSVCRT___mb_cur_max == 1 || c < 256)
1195 return u__strset(str, c); /* ASCII CP or SB char */
1197 c &= 0xffff; /* Strip high bits */
1199 while(str[0] && str[1])
1201 *str++ = c >> 8;
1202 *str++ = c & 0xff;
1204 if(str[0])
1205 str[0] = '\0'; /* FIXME: OK to shorten? */
1207 return ret;
1210 /*********************************************************************
1211 * _mbsnbset(MSVCRT.@)
1213 unsigned char* CDECL _mbsnbset(unsigned char *str, unsigned int c, MSVCRT_size_t len)
1215 unsigned char *ret = str;
1217 if(!len)
1218 return ret;
1220 if(MSVCRT___mb_cur_max == 1 || c < 256)
1221 return u__strnset(str, c, len); /* ASCII CP or SB char */
1223 c &= 0xffff; /* Strip high bits */
1225 while(str[0] && str[1] && (len > 1))
1227 *str++ = c >> 8;
1228 len--;
1229 *str++ = c & 0xff;
1230 len--;
1232 if(len && str[0]) {
1233 /* as per msdn pad with a blank character */
1234 str[0] = ' ';
1237 return ret;
1240 /*********************************************************************
1241 * _mbsnset(MSVCRT.@)
1243 unsigned char* CDECL _mbsnset(unsigned char* str, unsigned int c, MSVCRT_size_t len)
1245 unsigned char *ret = str;
1247 if(!len)
1248 return ret;
1250 if(MSVCRT___mb_cur_max == 1 || c < 256)
1251 return u__strnset(str, c, len); /* ASCII CP or SB char */
1253 c &= 0xffff; /* Strip high bits */
1255 while(str[0] && str[1] && len--)
1257 *str++ = c >> 8;
1258 *str++ = c & 0xff;
1260 if(len && str[0])
1261 str[0] = '\0'; /* FIXME: OK to shorten? */
1263 return ret;
1266 /*********************************************************************
1267 * _mbsnccnt(MSVCRT.@)
1268 * 'c' is for 'character'.
1270 MSVCRT_size_t CDECL _mbsnccnt(const unsigned char* str, MSVCRT_size_t len)
1272 MSVCRT_size_t ret;
1273 if(MSVCRT___mb_cur_max > 1)
1275 ret=0;
1276 while(*str && len-- > 0)
1278 if(MSVCRT_isleadbyte(*str))
1280 if (!len)
1281 break;
1282 len--;
1283 str++;
1285 str++;
1286 ret++;
1288 return ret;
1290 ret=u_strlen(str);
1291 return min(ret, len); /* ASCII CP */
1294 /*********************************************************************
1295 * _mbsnbcnt(MSVCRT.@)
1296 * 'b' is for byte count.
1298 MSVCRT_size_t CDECL _mbsnbcnt(const unsigned char* str, MSVCRT_size_t len)
1300 MSVCRT_size_t ret;
1301 if(MSVCRT___mb_cur_max > 1)
1303 const unsigned char* xstr = str;
1304 while(*xstr && len-- > 0)
1306 if (MSVCRT_isleadbyte(*xstr++))
1307 xstr++;
1309 return xstr-str;
1311 ret=u_strlen(str);
1312 return min(ret, len); /* ASCII CP */
1315 /*********************************************************************
1316 * _mbsnbcat(MSVCRT.@)
1318 unsigned char* CDECL _mbsnbcat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1320 if(MSVCRT___mb_cur_max > 1)
1322 unsigned char *res = dst;
1323 while (*dst) {
1324 if (MSVCRT_isleadbyte(*dst++)) {
1325 if (*dst) {
1326 dst++;
1327 } else {
1328 /* as per msdn overwrite the lead byte in front of '\0' */
1329 dst--;
1330 break;
1334 while (*src && len--) *dst++ = *src++;
1335 *dst = '\0';
1336 return res;
1338 return u_strncat(dst, src, len); /* ASCII CP */
1342 /*********************************************************************
1343 * _mbsncat(MSVCRT.@)
1345 unsigned char* CDECL _mbsncat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1347 if(MSVCRT___mb_cur_max > 1)
1349 unsigned char *res = dst;
1350 while (*dst)
1352 if (MSVCRT_isleadbyte(*dst++))
1353 dst++;
1355 while (*src && len--)
1357 *dst++ = *src;
1358 if(MSVCRT_isleadbyte(*src++))
1359 *dst++ = *src++;
1361 *dst = '\0';
1362 return res;
1364 return u_strncat(dst, src, len); /* ASCII CP */
1368 /*********************************************************************
1369 * _mbslwr(MSVCRT.@)
1371 unsigned char* CDECL _mbslwr(unsigned char* s)
1373 unsigned char *ret = s;
1374 if (!s)
1375 return NULL;
1376 if (MSVCRT___mb_cur_max > 1)
1378 unsigned int c;
1379 while (*s)
1381 c = _mbctolower(_mbsnextc(s));
1382 /* Note that I assume that the size of the character is unchanged */
1383 if (c > 255)
1385 *s++=(c>>8);
1386 c=c & 0xff;
1388 *s++=c;
1391 else for ( ; *s; s++) *s = tolower(*s);
1392 return ret;
1396 /*********************************************************************
1397 * _mbsupr(MSVCRT.@)
1399 unsigned char* CDECL _mbsupr(unsigned char* s)
1401 unsigned char *ret = s;
1402 if (!s)
1403 return NULL;
1404 if (MSVCRT___mb_cur_max > 1)
1406 unsigned int c;
1407 while (*s)
1409 c = _mbctoupper(_mbsnextc(s));
1410 /* Note that I assume that the size of the character is unchanged */
1411 if (c > 255)
1413 *s++=(c>>8);
1414 c=c & 0xff;
1416 *s++=c;
1419 else for ( ; *s; s++) *s = toupper(*s);
1420 return ret;
1424 /*********************************************************************
1425 * _mbsspn (MSVCRT.@)
1427 MSVCRT_size_t CDECL _mbsspn(const unsigned char* string, const unsigned char* set)
1429 const unsigned char *p, *q;
1431 for (p = string; *p; p++)
1433 if (MSVCRT_isleadbyte(*p))
1435 for (q = set; *q; q++)
1437 if (!q[1])
1438 break;
1439 if ((*p == *q) && (p[1] == q[1]))
1440 break;
1441 q++;
1443 if (!q[0] || !q[1]) break;
1445 else
1447 for (q = set; *q; q++)
1448 if (*p == *q)
1449 break;
1450 if (!*q) break;
1453 return p - string;
1456 /*********************************************************************
1457 * _mbsspnp (MSVCRT.@)
1459 unsigned char* CDECL _mbsspnp(const unsigned char* string, const unsigned char* set)
1461 const unsigned char *p, *q;
1463 for (p = string; *p; p++)
1465 if (MSVCRT_isleadbyte(*p))
1467 for (q = set; *q; q++)
1469 if (!q[1])
1470 break;
1471 if ((*p == *q) && (p[1] == q[1]))
1472 break;
1473 q++;
1475 if (!q[0] || !q[1]) break;
1477 else
1479 for (q = set; *q; q++)
1480 if (*p == *q)
1481 break;
1482 if (!*q) break;
1485 if (*p == '\0')
1486 return NULL;
1487 return (unsigned char *)p;
1490 /*********************************************************************
1491 * _mbscspn(MSVCRT.@)
1493 MSVCRT_size_t CDECL _mbscspn(const unsigned char* str, const unsigned char* cmp)
1495 if (MSVCRT___mb_cur_max > 1)
1496 FIXME("don't handle double character case\n");
1497 return u_strcspn(str, cmp);
1500 /*********************************************************************
1501 * _mbsrev (MSVCRT.@)
1503 unsigned char* CDECL _mbsrev(unsigned char* str)
1505 int i, len = _mbslen(str);
1506 unsigned char *p, *temp=MSVCRT_malloc(len*2);
1508 if(!temp)
1509 return str;
1511 /* unpack multibyte string to temp buffer */
1512 p=str;
1513 for(i=0; i<len; i++)
1515 if (MSVCRT_isleadbyte(*p))
1517 temp[i*2]=*p++;
1518 temp[i*2+1]=*p++;
1520 else
1522 temp[i*2]=*p++;
1523 temp[i*2+1]=0;
1527 /* repack it in the reverse order */
1528 p=str;
1529 for(i=len-1; i>=0; i--)
1531 if(MSVCRT_isleadbyte(temp[i*2]))
1533 *p++=temp[i*2];
1534 *p++=temp[i*2+1];
1536 else
1538 *p++=temp[i*2];
1542 MSVCRT_free(temp);
1544 return str;
1547 /*********************************************************************
1548 * _mbspbrk (MSVCRT.@)
1550 unsigned char* CDECL _mbspbrk(const unsigned char* str, const unsigned char* accept)
1552 const unsigned char* p;
1554 while(*str)
1556 for(p = accept; *p; p += (MSVCRT_isleadbyte(*p)?2:1) )
1558 if (*p == *str)
1559 if( !MSVCRT_isleadbyte(*p) || ( *(p+1) == *(str+1) ) )
1560 return (unsigned char*)str;
1562 str += (MSVCRT_isleadbyte(*str)?2:1);
1564 return NULL;
1569 * Functions depending on locale codepage
1572 /*********************************************************************
1573 * mblen(MSVCRT.@)
1574 * REMARKS
1575 * Unlike most of the multibyte string functions this function uses
1576 * the locale codepage, not the codepage set by _setmbcp
1578 int CDECL MSVCRT_mblen(const char* str, MSVCRT_size_t size)
1580 if (str && *str && size)
1582 if(MSVCRT___mb_cur_max == 1)
1583 return 1; /* ASCII CP */
1585 return !MSVCRT_isleadbyte(*str) ? 1 : (size>1 ? 2 : -1);
1587 return 0;
1590 /*********************************************************************
1591 * _mbstrlen(MSVCRT.@)
1592 * REMARKS
1593 * Unlike most of the multibyte string functions this function uses
1594 * the locale codepage, not the codepage set by _setmbcp
1596 MSVCRT_size_t CDECL _mbstrlen(const char* str)
1598 if(MSVCRT___mb_cur_max > 1)
1600 MSVCRT_size_t len = 0;
1601 while(*str)
1603 /* FIXME: According to the documentation we are supposed to test for
1604 * multi-byte character validity. Whatever that means
1606 str += MSVCRT_isleadbyte(*str) ? 2 : 1;
1607 len++;
1609 return len;
1611 return strlen(str); /* ASCII CP */