dlls/msvcrt/mbcs.c

   1 /*
   2  * msvcrt.dll mbcs functions
   3  *
   4  * Copyright 1999 Alexandre Julliard
   5  * Copyright 2000 Jon Griffths
   6  *
   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.
  11  *
  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.
  16  *
  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
  20  *
  21  * FIXME
  22  * Not currently binary compatible with win32. MSVCRT_mbctype must be
  23  * populated correctly and the ismb* functions should reference it.
  24  */
  25
  26 #include <stdio.h>
  27
  28 #include "msvcrt.h"
  29 #include "wine/unicode.h"
  30 #include "wine/debug.h"
  31
  32 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
  33
  34 unsigned char MSVCRT_mbctype[257] = { 0 };
  35
  36 /* It seems that the data about valid trail bytes is not available from kernel32
  37  * so we have to store is here. The format is the same as for lead bytes in CPINFO */
  38 struct cp_extra_info_t
  39 {
  40     int cp;
  41     BYTE TrailBytes[MAX_LEADBYTES];
  42 };
  43
  44 static struct cp_extra_info_t g_cpextrainfo[] =
  45 {
  46     {932, {0x40, 0x7e, 0x80, 0xfc, 0, 0}},
  47     {936, {0x40, 0xfe, 0, 0}},
  48     {949, {0x41, 0xfe, 0, 0}},
  49     {950, {0x40, 0x7e, 0xa1, 0xfe, 0, 0}},
  50     {1361, {0x31, 0x7e, 0x81, 0xfe, 0, 0}},
  51     {20932, {1, 255, 0, 0}},  /* seems to give different results on different systems */
  52     {0, {1, 255, 0, 0}}       /* match all with FIXME */
  53 };
  54
  55 /* Maps cp932 single byte character to multi byte character */
  56 static const unsigned char mbbtombc_932[] = {
  57   0x40,0x49,0x68,0x94,0x90,0x93,0x95,0x66,0x69,0x6a,0x96,0x7b,0x43,0x7c,0x44,0x5e,
  58   0x4f,0x50,0x51,0x52,0x53,0x54,0x55,0x56,0x57,0x58,0x46,0x47,0x83,0x81,0x84,0x48,
  59   0x97,0x60,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,
  60   0x6f,0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x6d,0x8f,0x6e,0x4f,0x76,
  61   0x77,0x78,0x79,0x6d,0x8f,0x6e,0x4f,0x51,0x65,0x81,0x82,0x83,0x84,0x85,0x86,0x87,
  62   0x88,0x89,0x8a,0x8b,0x8c,0x8d,0x8e,0x8f,0x90,0x91,0x92,0x93,0x94,0x95,0x50,
  63        0x42,0x75,0x76,0x41,0x45,0x92,0x40,0x42,0x44,0x46,0x48,0x83,0x85,0x87,0x62,
  64   0x5b,0x41,0x43,0x45,0x47,0x49,0x4a,0x4c,0x4e,0x50,0x52,0x54,0x56,0x58,0x5a,0x5c,
  65   0x5e,0x60,0x63,0x65,0x67,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,0x71,0x74,0x77,0x7a,0x7d,
  66   0x7e,0x80,0x81,0x82,0x84,0x86,0x88,0x89,0x8a,0x8b,0x8c,0x8d,0x8f,0x93,0x4a,0x4b };
  67
  68 /* Maps multibyte cp932 punctuation marks to single byte equivalents */
  69 static const unsigned char mbctombb_932_punct[] = {
  70   0x20,0xa4,0xa1,0x2c,0x2e,0xa5,0x3a,0x3b,0x3f,0x21,0xde,0xdf,0x00,0x00,0x00,0x5e,
  71   0x7e,0x5f,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0xb0,0x00,0x00,0x2f,0x00,
  72   0x00,0x00,0x7c,0x00,0x00,0x60,0x27,0x00,0x22,0x28,0x29,0x00,0x00,0x5b,0x5d,0x7b,
  73   0x7d,0x00,0x00,0x00,0x00,0xa2,0xa3,0x00,0x00,0x00,0x00,0x2b,0x2d,0x00,0x00,0x00,
  74   0x00,0x3d,0x00,0x3c,0x3e,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x5c,
  75   0x24,0x00,0x00,0x25,0x23,0x26,0x2a,0x40};
  76
  77 /* Maps multibyte cp932 hiragana/katakana to single-byte equivalents */
  78 static const unsigned char mbctombb_932_kana[] = {
  79   0xa7,0xb1,0xa8,0xb2,0xa9,0xb3,0xaa,0xb4,0xab,0xb5,0xb6,0xb6,0xb7,0xb7,0xb8,0xb8,
  80   0xb9,0xb9,0xba,0xba,0xbb,0xbb,0xbc,0xbc,0xbd,0xbd,0xbe,0xbe,0xbf,0xbf,0xc0,0xc0,
  81   0xc1,0xc1,0xaf,0xc2,0xc2,0xc3,0xc3,0xc4,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xca,
  82   0xca,0xcb,0xcb,0xcb,0xcc,0xcc,0xcc,0xcd,0xcd,0xcd,0xce,0xce,0xce,0xcf,0xd0,0xd1,
  83   0xd2,0xd3,0xac,0xd4,0xad,0xd5,0xae,0xd6,0xd7,0xd8,0xd9,0xda,0xdb,0xdc,0xdc,0xb2,
  84   0xb4,0xa6,0xdd,0xb3,0xb6,0xb9};
  85
  86 static MSVCRT_wchar_t msvcrt_mbc_to_wc(unsigned int ch)
  87 {
  88   MSVCRT_wchar_t chW;
  89   char mbch[2];
  90   int n_chars;
  91
  92   if (ch <= 0xff) {
  93     mbch[0] = ch;
  94     n_chars = 1;
  95   } else {
  96     mbch[0] = (ch >> 8) & 0xff;
  97     mbch[1] = ch & 0xff;
  98     n_chars = 2;
  99   }
 100   if (!MultiByteToWideChar(get_mbcinfo()->mbcodepage, 0, mbch, n_chars, &chW, 1))
 101   {
 102     WARN("MultiByteToWideChar failed on %x\n", ch);
 103     return 0;
 104   }
 105   return chW;
 106 }
 107
 108 static inline MSVCRT_size_t u_strlen( const unsigned char *str )
 109 {
 110   return strlen( (const char*) str );
 111 }
 112
 113 static inline unsigned char* u_strncat( unsigned char* dst, const unsigned char* src, MSVCRT_size_t len )
 114 {
 115   return (unsigned char*)strncat( (char*)dst, (const char*)src, len);
 116 }
 117
 118 static inline int u_strcmp( const unsigned char *s1, const unsigned char *s2 )
 119 {
 120   return strcmp( (const char*)s1, (const char*)s2 );
 121 }
 122
 123 static inline int u_strcasecmp( const unsigned char *s1, const unsigned char *s2 )
 124 {
 125   return strcasecmp( (const char*)s1, (const char*)s2 );
 126 }
 127
 128 static inline int u_strncmp( const unsigned char *s1, const unsigned char *s2, MSVCRT_size_t len )
 129 {
 130   return strncmp( (const char*)s1, (const char*)s2, len );
 131 }
 132
 133 static inline int u_strncasecmp( const unsigned char *s1, const unsigned char *s2, MSVCRT_size_t len )
 134 {
 135   return strncasecmp( (const char*)s1, (const char*)s2, len );
 136 }
 137
 138 static inline unsigned char *u_strchr( const unsigned char *s, unsigned char x )
 139 {
 140   return (unsigned char*) strchr( (const char*)s, x );
 141 }
 142
 143 static inline unsigned char *u_strrchr( const unsigned char *s, unsigned char x )
 144 {
 145   return (unsigned char*) strrchr( (const char*)s, x );
 146 }
 147
 148 static inline unsigned char *u_strtok( unsigned char *s, const unsigned char *delim )
 149 {
 150   return (unsigned char*) strtok( (char*)s, (const char*)delim );
 151 }
 152
 153 static inline unsigned char *u__strset( unsigned char *s, unsigned char c )
 154 {
 155   return (unsigned char*) _strset( (char*)s, c);
 156 }
 157
 158 static inline unsigned char *u__strnset( unsigned char *s, unsigned char c, MSVCRT_size_t len )
 159 {
 160   return (unsigned char*) MSVCRT__strnset( (char*)s, c, len );
 161 }
 162
 163 static inline MSVCRT_size_t u_strcspn( const unsigned char *s, const unsigned char *rej )
 164 {
 165   return strcspn( (const char *)s, (const char*)rej );
 166 }
 167
 168 /*********************************************************************
 169  *              __p__mbctype (MSVCRT.@)
 170  */
 171 unsigned char* CDECL __p__mbctype(void)
 172 {
 173   return get_mbcinfo()->mbctype;
 174 }
 175
 176 /*********************************************************************
 177  *              ___mb_cur_max_func(MSVCRT.@)
 178  */
 179 int* CDECL MSVCRT____mb_cur_max_func(void)
 180 {
 181   return &get_locinfo()->mb_cur_max;
 182 }
 183
 184 /* ___mb_cur_max_l_func - not exported in native msvcrt */
 185 int* CDECL ___mb_cur_max_l_func(MSVCRT__locale_t locale)
 186 {
 187   MSVCRT_pthreadlocinfo locinfo;
 188
 189   if(!locale)
 190     locinfo = get_locinfo();
 191   else
 192     locinfo = locale->locinfo;
 193
 194   return &locinfo->mb_cur_max;
 195 }
 196
 197 /*********************************************************************
 198  * INTERNAL: _setmbcp_l
 199  */
 200 int _setmbcp_l(int cp, LCID lcid, MSVCRT_pthreadmbcinfo mbcinfo)
 201 {
 202   const char format[] = ".%d";
 203
 204   int newcp;
 205   CPINFO cpi;
 206   BYTE *bytes;
 207   WORD chartypes[256];
 208   char bufA[256];
 209   WCHAR bufW[256];
 210   int charcount;
 211   int ret;
 212   int i;
 213
 214   if(!mbcinfo)
 215       mbcinfo = get_mbcinfo();
 216
 217   switch (cp)
 218   {
 219     case _MB_CP_ANSI:
 220       newcp = GetACP();
 221       break;
 222     case _MB_CP_OEM:
 223       newcp = GetOEMCP();
 224       break;
 225     case _MB_CP_LOCALE:
 226       newcp = get_locinfo()->lc_codepage;
 227       if(newcp)
 228           break;
 229       /* fall through (C locale) */
 230     case _MB_CP_SBCS:
 231       newcp = 20127;   /* ASCII */
 232       break;
 233     default:
 234       newcp = cp;
 235       break;
 236   }
 237
 238   if(lcid == -1) {
 239     sprintf(bufA, format, newcp);
 240     mbcinfo->mblcid = MSVCRT_locale_to_LCID(bufA);
 241   } else {
 242     mbcinfo->mblcid = lcid;
 243   }
 244
 245   if(mbcinfo->mblcid == -1)
 246   {
 247     WARN("Can't assign LCID to codepage (%d)\n", mbcinfo->mblcid);
 248     mbcinfo->mblcid = 0;
 249   }
 250
 251   if (!GetCPInfo(newcp, &cpi))
 252   {
 253     WARN("Codepage %d not found\n", newcp);
 254     *MSVCRT__errno() = MSVCRT_EINVAL;
 255     return -1;
 256   }
 257
 258   /* setup the _mbctype */
 259   memset(mbcinfo->mbctype, 0, sizeof(unsigned char[257]));
 260   memset(mbcinfo->mbcasemap, 0, sizeof(unsigned char[256]));
 261
 262   bytes = cpi.LeadByte;
 263   while (bytes[0] || bytes[1])
 264   {
 265     for (i = bytes[0]; i <= bytes[1]; i++)
 266       mbcinfo->mbctype[i + 1] |= _M1;
 267     bytes += 2;
 268   }
 269
 270   if (cpi.MaxCharSize > 1)
 271   {
 272     /* trail bytes not available through kernel32 but stored in a structure in msvcrt */
 273     struct cp_extra_info_t *cpextra = g_cpextrainfo;
 274
 275     mbcinfo->ismbcodepage = 1;
 276     while (TRUE)
 277     {
 278       if (cpextra->cp == 0 || cpextra->cp == newcp)
 279       {
 280         if (cpextra->cp == 0)
 281           FIXME("trail bytes data not available for DBCS codepage %d - assuming all bytes\n", newcp);
 282
 283         bytes = cpextra->TrailBytes;
 284         while (bytes[0] || bytes[1])
 285         {
 286           for (i = bytes[0]; i <= bytes[1]; i++)
 287             mbcinfo->mbctype[i + 1] |= _M2;
 288           bytes += 2;
 289         }
 290         break;
 291       }
 292       cpextra++;
 293     }
 294   }
 295   else
 296     mbcinfo->ismbcodepage = 0;
 297
 298   /* we can't use GetStringTypeA directly because we don't have a locale - only a code page
 299    */
 300   charcount = 0;
 301   for (i = 0; i < 256; i++)
 302     if (!(mbcinfo->mbctype[i + 1] & _M1))
 303       bufA[charcount++] = i;
 304
 305   ret = MultiByteToWideChar(newcp, 0, bufA, charcount, bufW, charcount);
 306   if (ret != charcount)
 307     ERR("MultiByteToWideChar of chars failed for cp %d, ret=%d (exp %d), error=%d\n", newcp, ret, charcount, GetLastError());
 308
 309   GetStringTypeW(CT_CTYPE1, bufW, charcount, chartypes);
 310
 311   charcount = 0;
 312   for (i = 0; i < 256; i++)
 313     if (!(mbcinfo->mbctype[i + 1] & _M1))
 314     {
 315       if (chartypes[charcount] & C1_UPPER)
 316       {
 317         mbcinfo->mbctype[i + 1] |= _SBUP;
 318         bufW[charcount] = tolowerW(bufW[charcount]);
 319       }
 320       else if (chartypes[charcount] & C1_LOWER)
 321       {
 322         mbcinfo->mbctype[i + 1] |= _SBLOW;
 323         bufW[charcount] = toupperW(bufW[charcount]);
 324       }
 325       charcount++;
 326     }
 327
 328   ret = WideCharToMultiByte(newcp, 0, bufW, charcount, bufA, charcount, NULL, NULL);
 329   if (ret != charcount)
 330     ERR("WideCharToMultiByte failed for cp %d, ret=%d (exp %d), error=%d\n", newcp, ret, charcount, GetLastError());
 331
 332   charcount = 0;
 333   for (i = 0; i < 256; i++)
 334   {
 335     if(!(mbcinfo->mbctype[i + 1] & _M1))
 336     {
 337       if(mbcinfo->mbctype[i] & (C1_UPPER|C1_LOWER))
 338         mbcinfo->mbcasemap[i] = bufA[charcount];
 339       charcount++;
 340     }
 341   }
 342
 343   if (newcp == 932)   /* CP932 only - set _MP and _MS */
 344   {
 345     /* On Windows it's possible to calculate the _MP and _MS from CT_CTYPE1
 346      * and CT_CTYPE3. But as of Wine 0.9.43 we return wrong values what makes
 347      * it hard. As this is set only for codepage 932 we hardcode it what gives
 348      * also faster execution.
 349      */
 350     for (i = 161; i <= 165; i++)
 351       mbcinfo->mbctype[i + 1] |= _MP;
 352     for (i = 166; i <= 223; i++)
 353       mbcinfo->mbctype[i + 1] |= _MS;
 354   }
 355
 356   mbcinfo->mbcodepage = newcp;
 357   if(MSVCRT_locale && mbcinfo == MSVCRT_locale->mbcinfo)
 358     memcpy(MSVCRT_mbctype, MSVCRT_locale->mbcinfo->mbctype, sizeof(MSVCRT_mbctype));
 359
 360   return 0;
 361 }
 362
 363 /*********************************************************************
 364  *              _setmbcp (MSVCRT.@)
 365  */
 366 int CDECL _setmbcp(int cp)
 367 {
 368     return _setmbcp_l(cp, -1, NULL);
 369 }
 370
 371 /*********************************************************************
 372  *              _getmbcp (MSVCRT.@)
 373  */
 374 int CDECL _getmbcp(void)
 375 {
 376   return get_mbcinfo()->mbcodepage;
 377 }
 378
 379 /*********************************************************************
 380  *              _mbsnextc(MSVCRT.@)
 381  */
 382 unsigned int CDECL _mbsnextc(const unsigned char* str)
 383 {
 384   if(_ismbblead(*str))
 385     return *str << 8 | str[1];
 386   return *str;
 387 }
 388
 389 /*********************************************************************
 390  *              _mbctolower(MSVCRT.@)
 391  */
 392 unsigned int CDECL _mbctolower(unsigned int c)
 393 {
 394     if (_ismbblead(c))
 395     {
 396       FIXME("Handle MBC chars\n");
 397       return c;
 398     }
 399     return tolower(c); /* ASCII CP or SB char */
 400 }
 401
 402 /*********************************************************************
 403  *              _mbctoupper(MSVCRT.@)
 404  */
 405 unsigned int CDECL _mbctoupper(unsigned int c)
 406 {
 407     if (_ismbblead(c))
 408     {
 409       FIXME("Handle MBC chars\n");
 410       return c;
 411     }
 412     return toupper(c); /* ASCII CP or SB char */
 413 }
 414
 415 /*********************************************************************
 416  *              _mbctombb (MSVCRT.@)
 417  */
 418 unsigned int CDECL _mbctombb(unsigned int c)
 419 {
 420     unsigned int value;
 421
 422     if(get_mbcinfo()->mbcodepage == 932)
 423     {
 424         if(c >= 0x829f && c <= 0x82f1)    /* Hiragana */
 425             return mbctombb_932_kana[c - 0x829f];
 426         if(c >= 0x8340 && c <= 0x8396 && c != 0x837f)    /* Katakana */
 427             return mbctombb_932_kana[c - 0x8340 - (c >= 0x837f ? 1 : 0)];
 428         if(c >= 0x8140 && c <= 0x8197)    /* Punctuation */
 429         {
 430             value = mbctombb_932_punct[c - 0x8140];
 431             return value ? value : c;
 432         }
 433         if((c >= 0x824f && c <= 0x8258) || /* Fullwidth digits */
 434            (c >= 0x8260 && c <= 0x8279))   /* Fullwidth capitals letters */
 435             return c - 0x821f;
 436         if(c >= 0x8281 && c <= 0x829a)     /* Fullwidth small letters */
 437             return c - 0x8220;
 438         /* all other cases return c */
 439     }
 440     return c;
 441 }
 442
 443 /*********************************************************************
 444  *              _mbcjistojms(MSVCRT.@)
 445  *
 446  *              Converts a jis character to sjis.
 447  *              Based on description from
 448  *              http://www.slayers.ne.jp/~oouchi/code/jistosjis.html
 449  */
 450 unsigned int CDECL _mbcjistojms(unsigned int c)
 451 {
 452   /* Conversion takes place only when codepage is 932.
 453      In all other cases, c is returned unchanged */
 454   if(get_mbcinfo()->mbcodepage == 932)
 455   {
 456     if(HIBYTE(c) >= 0x21 && HIBYTE(c) <= 0x7e &&
 457        LOBYTE(c) >= 0x21 && LOBYTE(c) <= 0x7e)
 458     {
 459       if(HIBYTE(c) % 2)
 460         c += 0x1f;
 461       else
 462         c += 0x7d;
 463
 464       if(LOBYTE(c) >= 0x7F)
 465         c += 0x1;
 466
 467       c = (((HIBYTE(c) - 0x21)/2 + 0x81) << 8) | LOBYTE(c);
 468
 469       if(HIBYTE(c) > 0x9f)
 470         c += 0x4000;
 471     }
 472     else
 473       return 0; /* Codepage is 932, but c can't be converted */
 474   }
 475
 476   return c;
 477 }
 478
 479 /*********************************************************************
 480  *              _mbcjmstojis(MSVCRT.@)
 481  *
 482  *              Converts a sjis character to jis.
 483  */
 484 unsigned int CDECL _mbcjmstojis(unsigned int c)
 485 {
 486   /* Conversion takes place only when codepage is 932.
 487      In all other cases, c is returned unchanged */
 488   if(get_mbcinfo()->mbcodepage == 932)
 489   {
 490     if(_ismbclegal(c) && HIBYTE(c) < 0xf0)
 491     {
 492       if(HIBYTE(c) >= 0xe0)
 493         c -= 0x4000;
 494
 495       c = (((HIBYTE(c) - 0x81)*2 + 0x21) << 8) | LOBYTE(c);
 496
 497       if(LOBYTE(c) > 0x7f)
 498         c -= 0x1;
 499
 500       if(LOBYTE(c) > 0x9d)
 501         c += 0x83;
 502       else
 503         c -= 0x1f;
 504     }
 505     else
 506       return 0; /* Codepage is 932, but c can't be converted */
 507   }
 508
 509   return c;
 510 }
 511
 512 /*********************************************************************
 513  *              _mbsdec(MSVCRT.@)
 514  */
 515 unsigned char* CDECL _mbsdec(const unsigned char* start, const unsigned char* cur)
 516 {
 517   if(get_mbcinfo()->ismbcodepage)
 518     return (unsigned char *)(_ismbstrail(start,cur-1) ? cur - 2 : cur -1);
 519
 520   return (unsigned char *)cur - 1; /* ASCII CP or SB char */
 521 }
 522
 523 /*********************************************************************
 524  *              _mbclen(MSVCRT.@)
 525  */
 526 unsigned int CDECL _mbclen(const unsigned char* str)
 527 {
 528   return _ismbblead(*str) ? 2 : 1;
 529 }
 530
 531 /*********************************************************************
 532  *              _mbsinc(MSVCRT.@)
 533  */
 534 unsigned char* CDECL _mbsinc(const unsigned char* str)
 535 {
 536   return (unsigned char *)(str + _mbclen(str));
 537 }
 538
 539 /*********************************************************************
 540  *              _mbsninc(MSVCRT.@)
 541  */
 542 unsigned char* CDECL _mbsninc(const unsigned char* str, MSVCRT_size_t num)
 543 {
 544   if(!str)
 545     return NULL;
 546
 547   while (num > 0 && *str)
 548   {
 549     if (_ismbblead(*str))
 550     {
 551       if (!*(str+1))
 552          break;
 553       str++;
 554     }
 555     str++;
 556     num--;
 557   }
 558
 559   return (unsigned char*)str;
 560 }
 561
 562 /*********************************************************************
 563  *              _mbslen(MSVCRT.@)
 564  */
 565 MSVCRT_size_t CDECL _mbslen(const unsigned char* str)
 566 {
 567   MSVCRT_size_t len = 0;
 568   while(*str)
 569   {
 570     if (_ismbblead(*str))
 571     {
 572       str++;
 573       if (!*str)  /* count only full chars */
 574         break;
 575     }
 576     str++;
 577     len++;
 578   }
 579   return len;
 580 }
 581
 582 /*********************************************************************
 583  *              _mbccpy(MSVCRT.@)
 584  */
 585 void CDECL _mbccpy(unsigned char* dest, const unsigned char* src)
 586 {
 587   *dest = *src;
 588   if(_ismbblead(*src))
 589     *++dest = *++src; /* MB char */
 590 }
 591
 592 /*********************************************************************
 593  *              _mbsncpy(MSVCRT.@)
 594  * REMARKS
 595  *  The parameter n is the number or characters to copy, not the size of
 596  *  the buffer. Use _mbsnbcpy for a function analogical to strncpy
 597  */
 598 unsigned char* CDECL _mbsncpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 599 {
 600   unsigned char* ret = dst;
 601   if(!n)
 602     return dst;
 603   if (get_mbcinfo()->ismbcodepage)
 604   {
 605     while (*src && n)
 606     {
 607       n--;
 608       if (_ismbblead(*src))
 609       {
 610         if (!*(src+1))
 611         {
 612             *dst++ = 0;
 613             *dst++ = 0;
 614             break;
 615         }
 616
 617         *dst++ = *src++;
 618       }
 619
 620       *dst++ = *src++;
 621     }
 622   }
 623   else
 624   {
 625     while (n)
 626     {
 627         n--;
 628         if (!(*dst++ = *src++)) break;
 629     }
 630   }
 631   while (n--) *dst++ = 0;
 632   return ret;
 633 }
 634
 635 /*********************************************************************
 636  *              _mbsnbcpy_s(MSVCRT.@)
 637  * REMARKS
 638  * Unlike _mbsnbcpy this function does not pad the rest of the dest
 639  * string with 0
 640  */
 641 int CDECL _mbsnbcpy_s(unsigned char* dst, MSVCRT_size_t size, const unsigned char* src, MSVCRT_size_t n)
 642 {
 643     MSVCRT_size_t pos = 0;
 644
 645     if(!dst || size == 0)
 646         return MSVCRT_EINVAL;
 647     if(!src)
 648     {
 649         dst[0] = '\0';
 650         return MSVCRT_EINVAL;
 651     }
 652     if(!n)
 653         return 0;
 654
 655     if(get_mbcinfo()->ismbcodepage)
 656     {
 657         int is_lead = 0;
 658         while (*src && n)
 659         {
 660             if(pos == size)
 661             {
 662                 dst[0] = '\0';
 663                 return MSVCRT_ERANGE;
 664             }
 665             is_lead = (!is_lead && _ismbblead(*src));
 666             n--;
 667             dst[pos++] = *src++;
 668         }
 669
 670         if (is_lead) /* if string ends with a lead, remove it */
 671             dst[pos - 1] = 0;
 672     }
 673     else
 674     {
 675         while (n)
 676         {
 677             n--;
 678             if(pos == size)
 679             {
 680                 dst[0] = '\0';
 681                 return MSVCRT_ERANGE;
 682             }
 683
 684             if(!(*src)) break;
 685             dst[pos++] = *src++;
 686         }
 687     }
 688
 689     if(pos < size)
 690         dst[pos] = '\0';
 691     else
 692     {
 693         dst[0] = '\0';
 694         return MSVCRT_ERANGE;
 695     }
 696
 697     return 0;
 698 }
 699
 700 /*********************************************************************
 701  *              _mbsnbcpy(MSVCRT.@)
 702  * REMARKS
 703  *  Like strncpy this function doesn't enforce the string to be
 704  *  NUL-terminated
 705  */
 706 unsigned char* CDECL _mbsnbcpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 707 {
 708   unsigned char* ret = dst;
 709   if(!n)
 710     return dst;
 711   if(get_mbcinfo()->ismbcodepage)
 712   {
 713     int is_lead = 0;
 714     while (*src && n)
 715     {
 716       is_lead = (!is_lead && _ismbblead(*src));
 717       n--;
 718       *dst++ = *src++;
 719     }
 720
 721     if (is_lead) /* if string ends with a lead, remove it */
 722         *(dst - 1) = 0;
 723   }
 724   else
 725   {
 726     while (n)
 727     {
 728         n--;
 729         if (!(*dst++ = *src++)) break;
 730     }
 731   }
 732   while (n--) *dst++ = 0;
 733   return ret;
 734 }
 735
 736 /*********************************************************************
 737  *              _mbscmp(MSVCRT.@)
 738  */
 739 int CDECL _mbscmp(const unsigned char* str, const unsigned char* cmp)
 740 {
 741   if(get_mbcinfo()->ismbcodepage)
 742   {
 743     unsigned int strc, cmpc;
 744     do {
 745       if(!*str)
 746         return *cmp ? -1 : 0;
 747       if(!*cmp)
 748         return 1;
 749       strc = _mbsnextc(str);
 750       cmpc = _mbsnextc(cmp);
 751       if(strc != cmpc)
 752         return strc < cmpc ? -1 : 1;
 753       str +=(strc > 255) ? 2 : 1;
 754       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 755     } while(1);
 756   }
 757   return u_strcmp(str, cmp); /* ASCII CP */
 758 }
 759
 760 /*********************************************************************
 761  *              _mbsicoll(MSVCRT.@)
 762  * FIXME: handle locales.
 763  */
 764 int CDECL _mbsicoll(const unsigned char* str, const unsigned char* cmp)
 765 {
 766   if(get_mbcinfo()->ismbcodepage)
 767   {
 768     unsigned int strc, cmpc;
 769     do {
 770       if(!*str)
 771         return *cmp ? -1 : 0;
 772       if(!*cmp)
 773         return 1;
 774       strc = _mbctolower(_mbsnextc(str));
 775       cmpc = _mbctolower(_mbsnextc(cmp));
 776       if(strc != cmpc)
 777         return strc < cmpc ? -1 : 1;
 778       str +=(strc > 255) ? 2 : 1;
 779       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 780     } while(1);
 781   }
 782   return u_strcasecmp(str, cmp); /* ASCII CP */
 783 }
 784
 785 /*********************************************************************
 786  *              _mbscoll(MSVCRT.@)
 787  * Performs a case-sensitive comparison according to the current code page
 788  * RETURN
 789  *   _NLSCMPERROR if error
 790  * FIXME: handle locales.
 791  */
 792 int CDECL _mbscoll(const unsigned char* str, const unsigned char* cmp)
 793 {
 794   if(get_mbcinfo()->ismbcodepage)
 795   {
 796     unsigned int strc, cmpc;
 797     do {
 798       if(!*str)
 799         return *cmp ? -1 : 0;
 800       if(!*cmp)
 801         return 1;
 802       strc = _mbsnextc(str);
 803       cmpc = _mbsnextc(cmp);
 804       if(strc != cmpc)
 805         return strc < cmpc ? -1 : 1;
 806       str +=(strc > 255) ? 2 : 1;
 807       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 808     } while(1);
 809   }
 810   return u_strcmp(str, cmp); /* ASCII CP */
 811 }
 812
 813
 814 /*********************************************************************
 815  *              _mbsicmp(MSVCRT.@)
 816  */
 817 int CDECL _mbsicmp(const unsigned char* str, const unsigned char* cmp)
 818 {
 819   if(get_mbcinfo()->ismbcodepage)
 820   {
 821     unsigned int strc, cmpc;
 822     do {
 823       if(!*str)
 824         return *cmp ? -1 : 0;
 825       if(!*cmp)
 826         return 1;
 827       strc = _mbctolower(_mbsnextc(str));
 828       cmpc = _mbctolower(_mbsnextc(cmp));
 829       if(strc != cmpc)
 830         return strc < cmpc ? -1 : 1;
 831       str +=(strc > 255) ? 2 : 1;
 832       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 833     } while(1);
 834   }
 835   return u_strcasecmp(str, cmp); /* ASCII CP */
 836 }
 837
 838 /*********************************************************************
 839  *              _mbsncmp(MSVCRT.@)
 840  */
 841 int CDECL _mbsncmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 842 {
 843   if(!len)
 844     return 0;
 845
 846   if(get_mbcinfo()->ismbcodepage)
 847   {
 848     unsigned int strc, cmpc;
 849     while(len--)
 850     {
 851       int inc;
 852       if(!*str)
 853         return *cmp ? -1 : 0;
 854       if(!*cmp)
 855         return 1;
 856       strc = _mbsnextc(str);
 857       cmpc = _mbsnextc(cmp);
 858       if(strc != cmpc)
 859         return strc < cmpc ? -1 : 1;
 860       inc=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 861       str += inc;
 862       cmp += inc;
 863     }
 864     return 0; /* Matched len chars */
 865   }
 866   return u_strncmp(str, cmp, len); /* ASCII CP */
 867 }
 868
 869 /*********************************************************************
 870  *              _mbsnbcmp(MSVCRT.@)
 871  */
 872 int CDECL _mbsnbcmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 873 {
 874   if (!len)
 875     return 0;
 876   if(get_mbcinfo()->ismbcodepage)
 877   {
 878     unsigned int strc, cmpc;
 879     while (len)
 880     {
 881       int clen;
 882       if(!*str)
 883         return *cmp ? -1 : 0;
 884       if(!*cmp)
 885         return 1;
 886       if (_ismbblead(*str))
 887       {
 888         strc=(len>=2)?_mbsnextc(str):0;
 889         clen=2;
 890       }
 891       else
 892       {
 893         strc=*str;
 894         clen=1;
 895       }
 896       if (_ismbblead(*cmp))
 897         cmpc=(len>=2)?_mbsnextc(cmp):0;
 898       else
 899         cmpc=*cmp;
 900       if(strc != cmpc)
 901         return strc < cmpc ? -1 : 1;
 902       len -= clen;
 903       str += clen;
 904       cmp += clen;
 905     }
 906     return 0; /* Matched len chars */
 907   }
 908   return u_strncmp(str,cmp,len);
 909 }
 910
 911 /*********************************************************************
 912  *              _mbsnicmp(MSVCRT.@)
 913  *
 914  * Compare two multibyte strings case insensitively to 'len' characters.
 915  */
 916 int CDECL _mbsnicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 917 {
 918   /* FIXME: No tolower() for mb strings yet */
 919   if(get_mbcinfo()->ismbcodepage)
 920   {
 921     unsigned int strc, cmpc;
 922     while(len--)
 923     {
 924       if(!*str)
 925         return *cmp ? -1 : 0;
 926       if(!*cmp)
 927         return 1;
 928       strc = _mbctolower(_mbsnextc(str));
 929       cmpc = _mbctolower(_mbsnextc(cmp));
 930       if(strc != cmpc)
 931         return strc < cmpc ? -1 : 1;
 932       str +=(strc > 255) ? 2 : 1;
 933       cmp +=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 934     }
 935     return 0; /* Matched len chars */
 936   }
 937   return u_strncasecmp(str, cmp, len); /* ASCII CP */
 938 }
 939
 940 /*********************************************************************
 941  *              _mbsnbicmp(MSVCRT.@)
 942  */
 943 int CDECL _mbsnbicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 944 {
 945   if (!len)
 946     return 0;
 947   if(get_mbcinfo()->ismbcodepage)
 948   {
 949     unsigned int strc, cmpc;
 950     while (len)
 951     {
 952       int clen;
 953       if(!*str)
 954         return *cmp ? -1 : 0;
 955       if(!*cmp)
 956         return 1;
 957       if (_ismbblead(*str))
 958       {
 959         strc=(len>=2)?_mbsnextc(str):0;
 960         clen=2;
 961       }
 962       else
 963       {
 964         strc=*str;
 965         clen=1;
 966       }
 967       if (_ismbblead(*cmp))
 968         cmpc=(len>=2)?_mbsnextc(cmp):0;
 969       else
 970         cmpc=*cmp;
 971       strc = _mbctolower(strc);
 972       cmpc = _mbctolower(cmpc);
 973       if(strc != cmpc)
 974         return strc < cmpc ? -1 : 1;
 975       len -= clen;
 976       str += clen;
 977       cmp += clen;
 978     }
 979     return 0; /* Matched len bytes */
 980   }
 981   return u_strncasecmp(str,cmp,len);
 982 }
 983
 984 /*********************************************************************
 985  *              _mbscat (MSVCRT.@)
 986  */
 987 unsigned char * CDECL _mbscat( unsigned char *dst, const unsigned char *src )
 988 {
 989     strcat( (char *)dst, (const char *)src );
 990     return dst;
 991 }
 992
 993 /*********************************************************************
 994  *              _mbscpy (MSVCRT.@)
 995  */
 996 unsigned char* CDECL _mbscpy( unsigned char *dst, const unsigned char *src )
 997 {
 998     strcpy( (char *)dst, (const char *)src );
 999     return dst;
1000 }
1001
1002 /*********************************************************************
1003  *              _mbsstr (MSVCRT.@)
1004  */
1005 unsigned char * CDECL _mbsstr(const unsigned char *haystack, const unsigned char *needle)
1006 {
1007     return (unsigned char *)strstr( (const char *)haystack, (const char *)needle );
1008 }
1009
1010 /*********************************************************************
1011  *              _mbschr(MSVCRT.@)
1012  *
1013  * Find a multibyte character in a multibyte string.
1014  */
1015 unsigned char* CDECL _mbschr(const unsigned char* s, unsigned int x)
1016 {
1017   if(get_mbcinfo()->ismbcodepage)
1018   {
1019     unsigned int c;
1020     while (1)
1021     {
1022       c = _mbsnextc(s);
1023       if (c == x)
1024         return (unsigned char*)s;
1025       if (!c)
1026         return NULL;
1027       s += c > 255 ? 2 : 1;
1028     }
1029   }
1030   return u_strchr(s, x); /* ASCII CP */
1031 }
1032
1033 /*********************************************************************
1034  *              _mbsrchr(MSVCRT.@)
1035  */
1036 unsigned char* CDECL _mbsrchr(const unsigned char* s, unsigned int x)
1037 {
1038   if(get_mbcinfo()->ismbcodepage)
1039   {
1040     unsigned int c;
1041     unsigned char* match=NULL;
1042     if(!s)
1043       return NULL;
1044     while (1) {
1045       c = _mbsnextc(s);
1046       if (c == x)
1047         match=(unsigned char*)s;
1048       if (!c)
1049         return match;
1050       s +=(c > 255) ? 2 : 1;
1051     }
1052   }
1053   return u_strrchr(s, x);
1054 }
1055
1056 /*********************************************************************
1057  *              _mbstok(MSVCRT.@)
1058  *
1059  * Find and extract tokens from strings
1060  */
1061 unsigned char* CDECL _mbstok(unsigned char *str, const unsigned char *delim)
1062 {
1063     thread_data_t *data = msvcrt_get_thread_data();
1064     unsigned char *ret;
1065
1066     if(get_mbcinfo()->ismbcodepage)
1067     {
1068         unsigned int c;
1069
1070         if (!str)
1071             if (!(str = data->mbstok_next)) return NULL;
1072
1073         while ((c = _mbsnextc(str)) && _mbschr(delim, c)) {
1074             str += c > 255 ? 2 : 1;
1075         }
1076         if (!*str) return NULL;
1077         ret = str++;
1078         while ((c = _mbsnextc(str)) && !_mbschr(delim, c)) {
1079             str += c > 255 ? 2 : 1;
1080         }
1081         if (*str) {
1082             *str++ = 0;
1083             if (c > 255) *str++ = 0;
1084         }
1085         data->mbstok_next = str;
1086         return ret;
1087     }
1088     return u_strtok(str, delim); /* ASCII CP */
1089 }
1090
1091 /*********************************************************************
1092  *              _mbbtombc(MSVCRT.@)
1093  */
1094 unsigned int CDECL _mbbtombc(unsigned int c)
1095 {
1096   if(get_mbcinfo()->mbcodepage == 932)
1097   {
1098     if(c >= 0x20 && c <= 0x7e) {
1099       if((c >= 0x41 && c <= 0x5a) || (c >= 0x61 && c <= 0x7a) || (c >= 0x30 && c <= 0x39))
1100         return mbbtombc_932[c - 0x20] | 0x8200;
1101       else
1102         return mbbtombc_932[c - 0x20] | 0x8100;
1103     }
1104     else if(c >= 0xa1 && c <= 0xdf) {
1105       if(c >= 0xa6 && c <= 0xdd && c != 0xb0)
1106         return mbbtombc_932[c - 0xa1 + 0x5f] | 0x8300;
1107       else
1108         return mbbtombc_932[c - 0xa1 + 0x5f] | 0x8100;
1109     }
1110   }
1111   return c;  /* not Japanese or no MB char */
1112 }
1113
1114 /*********************************************************************
1115  *              _mbbtype(MSVCRT.@)
1116  */
1117 int CDECL _mbbtype(unsigned char c, int type)
1118 {
1119     if (type == 1)
1120     {
1121         if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
1122             return _MBC_SINGLE;
1123         else if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
1124             return _MBC_TRAIL;
1125         else
1126             return _MBC_ILLEGAL;
1127     }
1128     else
1129     {
1130         if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
1131             return _MBC_SINGLE;
1132         else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
1133             return _MBC_LEAD;
1134         else
1135             return _MBC_ILLEGAL;
1136     }
1137 }
1138
1139 /*********************************************************************
1140  *              _ismbbkana(MSVCRT.@)
1141  */
1142 int CDECL _ismbbkana(unsigned int c)
1143 {
1144   /* FIXME: use lc_ctype when supported, not lc_all */
1145   if(get_mbcinfo()->mbcodepage == 932)
1146   {
1147     /* Japanese/Katakana, CP 932 */
1148     return (c >= 0xa1 && c <= 0xdf);
1149   }
1150   return 0;
1151 }
1152
1153 /*********************************************************************
1154  *              _ismbcdigit(MSVCRT.@)
1155  */
1156 int CDECL _ismbcdigit(unsigned int ch)
1157 {
1158     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1159     return (get_char_typeW( wch ) & C1_DIGIT);
1160 }
1161
1162 /*********************************************************************
1163  *              _ismbcgraph(MSVCRT.@)
1164  */
1165 int CDECL _ismbcgraph(unsigned int ch)
1166 {
1167     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1168     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA));
1169 }
1170
1171 /*********************************************************************
1172  *              _ismbcalpha (MSVCRT.@)
1173  */
1174 int CDECL _ismbcalpha(unsigned int ch)
1175 {
1176     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1177     return (get_char_typeW( wch ) & C1_ALPHA);
1178 }
1179
1180 /*********************************************************************
1181  *              _ismbclower (MSVCRT.@)
1182  */
1183 int CDECL _ismbclower(unsigned int ch)
1184 {
1185     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1186     return (get_char_typeW( wch ) & C1_UPPER);
1187 }
1188
1189 /*********************************************************************
1190  *              _ismbcupper (MSVCRT.@)
1191  */
1192 int CDECL _ismbcupper(unsigned int ch)
1193 {
1194     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1195     return (get_char_typeW( wch ) & C1_LOWER);
1196 }
1197
1198 /*********************************************************************
1199  *              _ismbcsymbol(MSVCRT.@)
1200  */
1201 int CDECL _ismbcsymbol(unsigned int ch)
1202 {
1203     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1204     WORD ctype;
1205     if (!GetStringTypeW(CT_CTYPE3, &wch, 1, &ctype))
1206     {
1207         WARN("GetStringTypeW failed on %x\n", ch);
1208         return 0;
1209     }
1210     return ((ctype & C3_SYMBOL) != 0);
1211 }
1212
1213 /*********************************************************************
1214  *              _ismbcalnum (MSVCRT.@)
1215  */
1216 int CDECL _ismbcalnum(unsigned int ch)
1217 {
1218     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1219     return (get_char_typeW( wch ) & (C1_ALPHA | C1_DIGIT));
1220 }
1221
1222 /*********************************************************************
1223  *              _ismbcspace (MSVCRT.@)
1224  */
1225 int CDECL _ismbcspace(unsigned int ch)
1226 {
1227     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1228     return (get_char_typeW( wch ) & C1_SPACE);
1229 }
1230
1231 /*********************************************************************
1232  *              _ismbcprint (MSVCRT.@)
1233  */
1234 int CDECL _ismbcprint(unsigned int ch)
1235 {
1236     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1237     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA | C1_SPACE));
1238 }
1239
1240 /*********************************************************************
1241  *              _ismbcpunct(MSVCRT.@)
1242  */
1243 int CDECL _ismbcpunct(unsigned int ch)
1244 {
1245     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
1246     return (get_char_typeW( wch ) & C1_PUNCT);
1247 }
1248
1249 /*********************************************************************
1250  *              _ismbchira(MSVCRT.@)
1251  */
1252 int CDECL _ismbchira(unsigned int c)
1253 {
1254   /* FIXME: use lc_ctype when supported, not lc_all */
1255   if(get_mbcinfo()->mbcodepage == 932)
1256   {
1257     /* Japanese/Hiragana, CP 932 */
1258     return (c >= 0x829f && c <= 0x82f1);
1259   }
1260   return 0;
1261 }
1262
1263 /*********************************************************************
1264  *              _ismbckata(MSVCRT.@)
1265  */
1266 int CDECL _ismbckata(unsigned int c)
1267 {
1268   /* FIXME: use lc_ctype when supported, not lc_all */
1269   if(get_mbcinfo()->mbcodepage == 932)
1270   {
1271     if(c < 256)
1272       return _ismbbkana(c);
1273     /* Japanese/Katakana, CP 932 */
1274     return (c >= 0x8340 && c <= 0x8396 && c != 0x837f);
1275   }
1276   return 0;
1277 }
1278
1279 /*********************************************************************
1280  *              _ismbblead(MSVCRT.@)
1281  */
1282 int CDECL _ismbblead(unsigned int c)
1283 {
1284   return (get_mbcinfo()->mbctype[(c&0xff) + 1] & _M1) != 0;
1285 }
1286
1287
1288 /*********************************************************************
1289  *              _ismbbtrail(MSVCRT.@)
1290  */
1291 int CDECL _ismbbtrail(unsigned int c)
1292 {
1293   return (get_mbcinfo()->mbctype[(c&0xff) + 1] & _M2) != 0;
1294 }
1295
1296 /*********************************************************************
1297  *              _ismbclegal(MSVCRT.@)
1298  */
1299 int CDECL _ismbclegal(unsigned int c)
1300 {
1301     return _ismbblead(HIBYTE(c)) && _ismbbtrail(LOBYTE(c));
1302 }
1303
1304 /*********************************************************************
1305  *              _ismbslead(MSVCRT.@)
1306  */
1307 int CDECL _ismbslead(const unsigned char* start, const unsigned char* str)
1308 {
1309   int lead = 0;
1310
1311   if(!get_mbcinfo()->ismbcodepage)
1312     return 0;
1313
1314   /* Lead bytes can also be trail bytes so we need to analyse the string
1315    */
1316   while (start <= str)
1317   {
1318     if (!*start)
1319       return 0;
1320     lead = !lead && _ismbblead(*start);
1321     start++;
1322   }
1323
1324   return lead ? -1 : 0;
1325 }
1326
1327 /*********************************************************************
1328  *              _ismbstrail(MSVCRT.@)
1329  */
1330 int CDECL _ismbstrail(const unsigned char* start, const unsigned char* str)
1331 {
1332   /* Note: this function doesn't check _ismbbtrail */
1333   if ((str > start) && _ismbslead(start, str-1))
1334     return -1;
1335   else
1336     return 0;
1337 }
1338
1339 /*********************************************************************
1340  *              _mbsbtype (MSVCRT.@)
1341  */
1342 int CDECL _mbsbtype(const unsigned char *str, MSVCRT_size_t count)
1343 {
1344   int lead = 0;
1345   const unsigned char *end = str + count;
1346
1347   /* Lead bytes can also be trail bytes so we need to analyse the string.
1348    * Also we must return _MBC_ILLEGAL for chars past the end of the string
1349    */
1350   while (str < end) /* Note: we skip the last byte - will check after the loop */
1351   {
1352     if (!*str)
1353       return _MBC_ILLEGAL;
1354     lead = get_mbcinfo()->ismbcodepage && !lead && _ismbblead(*str);
1355     str++;
1356   }
1357
1358   if (lead)
1359     if (_ismbbtrail(*str))
1360       return _MBC_TRAIL;
1361     else
1362       return _MBC_ILLEGAL;
1363   else
1364     if (_ismbblead(*str))
1365       return _MBC_LEAD;
1366     else
1367       return _MBC_SINGLE;
1368 }
1369
1370 /*********************************************************************
1371  *              _mbsset(MSVCRT.@)
1372  */
1373 unsigned char* CDECL _mbsset(unsigned char* str, unsigned int c)
1374 {
1375   unsigned char* ret = str;
1376
1377   if(!get_mbcinfo()->ismbcodepage || c < 256)
1378     return u__strset(str, c); /* ASCII CP or SB char */
1379
1380   c &= 0xffff; /* Strip high bits */
1381
1382   while(str[0] && str[1])
1383   {
1384     *str++ = c >> 8;
1385     *str++ = c & 0xff;
1386   }
1387   if(str[0])
1388     str[0] = '\0'; /* FIXME: OK to shorten? */
1389
1390   return ret;
1391 }
1392
1393 /*********************************************************************
1394  *              _mbsnbset(MSVCRT.@)
1395  */
1396 unsigned char* CDECL _mbsnbset(unsigned char *str, unsigned int c, MSVCRT_size_t len)
1397 {
1398     unsigned char *ret = str;
1399
1400     if(!len)
1401         return ret;
1402
1403     if(!get_mbcinfo()->ismbcodepage || c < 256)
1404         return u__strnset(str, c, len); /* ASCII CP or SB char */
1405
1406     c &= 0xffff; /* Strip high bits */
1407
1408     while(str[0] && str[1] && (len > 1))
1409     {
1410         *str++ = c >> 8;
1411         len--;
1412         *str++ = c & 0xff;
1413         len--;
1414     }
1415     if(len && str[0]) {
1416         /* as per msdn pad with a blank character */
1417         str[0] = ' ';
1418     }
1419
1420     return ret;
1421 }
1422
1423 /*********************************************************************
1424  *              _mbsnset(MSVCRT.@)
1425  */
1426 unsigned char* CDECL _mbsnset(unsigned char* str, unsigned int c, MSVCRT_size_t len)
1427 {
1428   unsigned char *ret = str;
1429
1430   if(!len)
1431     return ret;
1432
1433   if(!get_mbcinfo()->ismbcodepage || c < 256)
1434     return u__strnset(str, c, len); /* ASCII CP or SB char */
1435
1436   c &= 0xffff; /* Strip high bits */
1437
1438   while(str[0] && str[1] && len--)
1439   {
1440     *str++ = c >> 8;
1441     *str++ = c & 0xff;
1442   }
1443   if(len && str[0])
1444     str[0] = '\0'; /* FIXME: OK to shorten? */
1445
1446   return ret;
1447 }
1448
1449 /*********************************************************************
1450  *              _mbsnccnt(MSVCRT.@)
1451  * 'c' is for 'character'.
1452  */
1453 MSVCRT_size_t CDECL _mbsnccnt(const unsigned char* str, MSVCRT_size_t len)
1454 {
1455   MSVCRT_size_t ret;
1456   if(get_mbcinfo()->ismbcodepage)
1457   {
1458     ret=0;
1459     while(*str && len-- > 0)
1460     {
1461       if(_ismbblead(*str))
1462       {
1463         if (!len)
1464           break;
1465         len--;
1466         str++;
1467       }
1468       str++;
1469       ret++;
1470     }
1471     return ret;
1472   }
1473   ret=u_strlen(str);
1474   return min(ret, len); /* ASCII CP */
1475 }
1476
1477 /*********************************************************************
1478  *              _mbsnbcnt(MSVCRT.@)
1479  * 'b' is for byte count.
1480  */
1481 MSVCRT_size_t CDECL _mbsnbcnt(const unsigned char* str, MSVCRT_size_t len)
1482 {
1483   MSVCRT_size_t ret;
1484   if(get_mbcinfo()->ismbcodepage)
1485   {
1486     const unsigned char* xstr = str;
1487     while(*xstr && len-- > 0)
1488     {
1489       if (_ismbblead(*xstr++))
1490         xstr++;
1491     }
1492     return xstr-str;
1493   }
1494   ret=u_strlen(str);
1495   return min(ret, len); /* ASCII CP */
1496 }
1497
1498 /*********************************************************************
1499  *              _mbsnbcat(MSVCRT.@)
1500  */
1501 unsigned char* CDECL _mbsnbcat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1502 {
1503     if(get_mbcinfo()->ismbcodepage)
1504     {
1505         unsigned char *res = dst;
1506         while (*dst) {
1507             if (_ismbblead(*dst++)) {
1508                 if (*dst) {
1509                     dst++;
1510                 } else {
1511                     /* as per msdn overwrite the lead byte in front of '\0' */
1512                     dst--;
1513                     break;
1514                 }
1515             }
1516         }
1517         while (*src && len--) *dst++ = *src++;
1518         *dst = '\0';
1519         return res;
1520     }
1521     return u_strncat(dst, src, len); /* ASCII CP */
1522 }
1523
1524 int CDECL _mbsnbcat_s(unsigned char *dst, MSVCRT_size_t size, const unsigned char *src, MSVCRT_size_t len)
1525 {
1526     unsigned char *ptr = dst;
1527     MSVCRT_size_t i;
1528
1529     if (!dst && !size && !src && !len)
1530         return 0;
1531
1532     if (!dst || !size || !src)
1533     {
1534         if (dst && size)
1535             *dst = '\0';
1536
1537         *MSVCRT__errno() = MSVCRT_EINVAL;
1538         return MSVCRT_EINVAL;
1539     }
1540
1541     /* Find the null terminator of the destination buffer. */
1542     while (size && *ptr)
1543         size--, ptr++;
1544
1545     if (!size)
1546     {
1547         *dst = '\0';
1548         *MSVCRT__errno() = MSVCRT_EINVAL;
1549         return MSVCRT_EINVAL;
1550     }
1551
1552     /* If necessary, check that the character preceding the null terminator is
1553      * a lead byte and move the pointer back by one for later overwrite. */
1554     if (ptr != dst && get_mbcinfo()->ismbcodepage && _ismbblead(*(ptr - 1)))
1555         size++, ptr--;
1556
1557     for (i = 0; *src && i < len; i++)
1558     {
1559         *ptr++ = *src++;
1560         size--;
1561
1562         if (!size)
1563         {
1564             *dst = '\0';
1565             *MSVCRT__errno() = MSVCRT_ERANGE;
1566             return MSVCRT_ERANGE;
1567         }
1568     }
1569
1570     *ptr = '\0';
1571     return 0;
1572 }
1573
1574 /*********************************************************************
1575  *              _mbsncat(MSVCRT.@)
1576  */
1577 unsigned char* CDECL _mbsncat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1578 {
1579   if(get_mbcinfo()->ismbcodepage)
1580   {
1581     unsigned char *res = dst;
1582     while (*dst)
1583     {
1584       if (_ismbblead(*dst++))
1585         dst++;
1586     }
1587     while (*src && len--)
1588     {
1589       *dst++ = *src;
1590       if(_ismbblead(*src++))
1591         *dst++ = *src++;
1592     }
1593     *dst = '\0';
1594     return res;
1595   }
1596   return u_strncat(dst, src, len); /* ASCII CP */
1597 }
1598
1599
1600 /*********************************************************************
1601  *              _mbslwr(MSVCRT.@)
1602  */
1603 unsigned char* CDECL _mbslwr(unsigned char* s)
1604 {
1605   unsigned char *ret = s;
1606   if (!s)
1607     return NULL;
1608   if (get_mbcinfo()->ismbcodepage)
1609   {
1610     unsigned int c;
1611     while (*s)
1612     {
1613       c = _mbctolower(_mbsnextc(s));
1614       /* Note that I assume that the size of the character is unchanged */
1615       if (c > 255)
1616       {
1617           *s++=(c>>8);
1618           c=c & 0xff;
1619       }
1620       *s++=c;
1621     }
1622   }
1623   else for ( ; *s; s++) *s = tolower(*s);
1624   return ret;
1625 }
1626
1627
1628 /*********************************************************************
1629  *              _mbslwr_s(MSVCRT.@)
1630  */
1631 int CDECL _mbslwr_s(unsigned char* s, MSVCRT_size_t len)
1632 {
1633   if (!s && !len)
1634   {
1635     return 0;
1636   }
1637   else if (!s || !len)
1638   {
1639     *MSVCRT__errno() = MSVCRT_EINVAL;
1640     return MSVCRT_EINVAL;
1641   }
1642   if (get_mbcinfo()->ismbcodepage)
1643   {
1644     unsigned int c;
1645     for ( ; *s && len > 0; len--)
1646     {
1647       c = _mbctolower(_mbsnextc(s));
1648       /* Note that I assume that the size of the character is unchanged */
1649       if (c > 255)
1650       {
1651           *s++=(c>>8);
1652           c=c & 0xff;
1653       }
1654       *s++=c;
1655     }
1656   }
1657   else for ( ; *s && len > 0; s++, len--) *s = tolower(*s);
1658   if (*s)
1659   {
1660     *s = '\0';
1661     *MSVCRT__errno() = MSVCRT_EINVAL;
1662     return MSVCRT_EINVAL;
1663   }
1664   return 0;
1665 }
1666
1667
1668 /*********************************************************************
1669  *              _mbsupr(MSVCRT.@)
1670  */
1671 unsigned char* CDECL _mbsupr(unsigned char* s)
1672 {
1673   unsigned char *ret = s;
1674   if (!s)
1675     return NULL;
1676   if (get_mbcinfo()->ismbcodepage)
1677   {
1678     unsigned int c;
1679     while (*s)
1680     {
1681       c = _mbctoupper(_mbsnextc(s));
1682       /* Note that I assume that the size of the character is unchanged */
1683       if (c > 255)
1684       {
1685           *s++=(c>>8);
1686           c=c & 0xff;
1687       }
1688       *s++=c;
1689     }
1690   }
1691   else for ( ; *s; s++) *s = toupper(*s);
1692   return ret;
1693 }
1694
1695
1696 /*********************************************************************
1697  *              _mbsupr_s(MSVCRT.@)
1698  */
1699 int CDECL _mbsupr_s(unsigned char* s, MSVCRT_size_t len)
1700 {
1701   if (!s && !len)
1702   {
1703     return 0;
1704   }
1705   else if (!s || !len)
1706   {
1707     *MSVCRT__errno() = MSVCRT_EINVAL;
1708     return MSVCRT_EINVAL;
1709   }
1710   if (get_mbcinfo()->ismbcodepage)
1711   {
1712     unsigned int c;
1713     for ( ; *s && len > 0; len--)
1714     {
1715       c = _mbctoupper(_mbsnextc(s));
1716       /* Note that I assume that the size of the character is unchanged */
1717       if (c > 255)
1718       {
1719           *s++=(c>>8);
1720           c=c & 0xff;
1721       }
1722       *s++=c;
1723     }
1724   }
1725   else for ( ; *s && len > 0; s++, len--) *s = toupper(*s);
1726   if (*s)
1727   {
1728     *s = '\0';
1729     *MSVCRT__errno() = MSVCRT_EINVAL;
1730     return MSVCRT_EINVAL;
1731   }
1732   return 0;
1733 }
1734
1735
1736 /*********************************************************************
1737  *              _mbsspn (MSVCRT.@)
1738  */
1739 MSVCRT_size_t CDECL _mbsspn(const unsigned char* string, const unsigned char* set)
1740 {
1741     const unsigned char *p, *q;
1742
1743     for (p = string; *p; p++)
1744     {
1745         if (_ismbblead(*p))
1746         {
1747             for (q = set; *q; q++)
1748             {
1749                 if (!q[1])
1750                     break;
1751                 if ((*p == *q) &&  (p[1] == q[1]))
1752                     break;
1753                 q++;
1754             }
1755             if (!q[0] || !q[1]) break;
1756         }
1757         else
1758         {
1759             for (q = set; *q; q++)
1760                 if (*p == *q)
1761                     break;
1762             if (!*q) break;
1763         }
1764     }
1765     return p - string;
1766 }
1767
1768 /*********************************************************************
1769  *              _mbsspnp (MSVCRT.@)
1770  */
1771 unsigned char* CDECL _mbsspnp(const unsigned char* string, const unsigned char* set)
1772 {
1773     const unsigned char *p, *q;
1774
1775     for (p = string; *p; p++)
1776     {
1777         if (_ismbblead(*p))
1778         {
1779             for (q = set; *q; q++)
1780             {
1781                 if (!q[1])
1782                     break;
1783                 if ((*p == *q) &&  (p[1] == q[1]))
1784                     break;
1785                 q++;
1786             }
1787             if (!q[0] || !q[1]) break;
1788         }
1789         else
1790         {
1791             for (q = set; *q; q++)
1792                 if (*p == *q)
1793                     break;
1794             if (!*q) break;
1795         }
1796     }
1797     if (*p == '\0')
1798         return NULL;
1799     return (unsigned char *)p;
1800 }
1801
1802 /*********************************************************************
1803  *              _mbscspn(MSVCRT.@)
1804  */
1805 MSVCRT_size_t CDECL _mbscspn(const unsigned char* str, const unsigned char* cmp)
1806 {
1807   if (get_mbcinfo()->ismbcodepage)
1808     FIXME("don't handle double character case\n");
1809   return u_strcspn(str, cmp);
1810 }
1811
1812 /*********************************************************************
1813  *              _mbsrev (MSVCRT.@)
1814  */
1815 unsigned char* CDECL _mbsrev(unsigned char* str)
1816 {
1817     int i, len = _mbslen(str);
1818     unsigned char *p, *temp=MSVCRT_malloc(len*2);
1819
1820     if(!temp)
1821         return str;
1822
1823     /* unpack multibyte string to temp buffer */
1824     p=str;
1825     for(i=0; i<len; i++)
1826     {
1827         if (_ismbblead(*p))
1828         {
1829             temp[i*2]=*p++;
1830             temp[i*2+1]=*p++;
1831         }
1832         else
1833         {
1834             temp[i*2]=*p++;
1835             temp[i*2+1]=0;
1836         }
1837     }
1838
1839     /* repack it in the reverse order */
1840     p=str;
1841     for(i=len-1; i>=0; i--)
1842     {
1843         if(_ismbblead(temp[i*2]))
1844         {
1845             *p++=temp[i*2];
1846             *p++=temp[i*2+1];
1847         }
1848         else
1849         {
1850             *p++=temp[i*2];
1851         }
1852     }
1853
1854     MSVCRT_free(temp);
1855
1856     return str;
1857 }
1858
1859 /*********************************************************************
1860  *              _mbspbrk (MSVCRT.@)
1861  */
1862 unsigned char* CDECL _mbspbrk(const unsigned char* str, const unsigned char* accept)
1863 {
1864     const unsigned char* p;
1865
1866     while(*str)
1867     {
1868         for(p = accept; *p; p += (_ismbblead(*p)?2:1) )
1869         {
1870             if (*p == *str)
1871                 if( !_ismbblead(*p) || ( *(p+1) == *(str+1) ) )
1872                      return (unsigned char*)str;
1873         }
1874         str += (_ismbblead(*str)?2:1);
1875     }
1876     return NULL;
1877 }
1878
1879
1880 /*
1881  * Functions depending on locale codepage
1882  */
1883
1884 /*********************************************************************
1885  *              mblen(MSVCRT.@)
1886  * REMARKS
1887  *  Unlike most of the multibyte string functions this function uses
1888  *  the locale codepage, not the codepage set by _setmbcp
1889  */
1890 int CDECL MSVCRT_mblen(const char* str, MSVCRT_size_t size)
1891 {
1892   if (str && *str && size)
1893   {
1894     if(get_locinfo()->mb_cur_max == 1)
1895       return 1; /* ASCII CP */
1896
1897     return !MSVCRT_isleadbyte(*str) ? 1 : (size>1 ? 2 : -1);
1898   }
1899   return 0;
1900 }
1901
1902 /*********************************************************************
1903  *              _mbstrlen_l(MSVCRT.@)
1904  */
1905 MSVCRT_size_t CDECL _mbstrlen_l(const char* str, MSVCRT__locale_t locale)
1906 {
1907     MSVCRT_pthreadlocinfo locinfo;
1908
1909     if(!locale)
1910         locinfo = get_locinfo();
1911     else
1912         locinfo = locale->locinfo;
1913
1914     if(locinfo->mb_cur_max > 1) {
1915         MSVCRT_size_t len;
1916         len = MultiByteToWideChar(locinfo->lc_codepage, MB_ERR_INVALID_CHARS,
1917                                   str, -1, NULL, 0);
1918         if (!len) {
1919             *MSVCRT__errno() = MSVCRT_EILSEQ;
1920             return -1;
1921         }
1922         return len - 1;
1923     }
1924
1925     return strlen(str);
1926 }
1927
1928 /*********************************************************************
1929  *              _mbstrlen(MSVCRT.@)
1930  */
1931 MSVCRT_size_t CDECL _mbstrlen(const char* str)
1932 {
1933     return _mbstrlen_l(str, NULL);
1934 }
1935
1936 /*********************************************************************
1937  *              _mbtowc_l(MSVCRT.@)
1938  */
1939 int CDECL MSVCRT_mbtowc_l(MSVCRT_wchar_t *dst, const char* str, MSVCRT_size_t n, MSVCRT__locale_t locale)
1940 {
1941     MSVCRT_pthreadlocinfo locinfo;
1942     MSVCRT_wchar_t tmpdst = '\0';
1943
1944     if(!locale)
1945         locinfo = get_locinfo();
1946     else
1947         locinfo = locale->locinfo;
1948
1949     if(n <= 0 || !str)
1950         return 0;
1951     if(!MultiByteToWideChar(locinfo->lc_codepage, 0, str, n, &tmpdst, 1))
1952         return -1;
1953     if(dst)
1954         *dst = tmpdst;
1955     /* return the number of bytes from src that have been used */
1956     if(!*str)
1957         return 0;
1958     if(n >= 2 && MSVCRT__isleadbyte_l(*str, locale) && str[1])
1959         return 2;
1960     return 1;
1961 }
1962
1963 /*********************************************************************
1964  *              mbtowc(MSVCRT.@)
1965  */
1966 int CDECL MSVCRT_mbtowc(MSVCRT_wchar_t *dst, const char* str, MSVCRT_size_t n)
1967 {
1968     return MSVCRT_mbtowc_l(dst, str, n, NULL);
1969 }
1970
1971 /*********************************************************************
1972  *              _mbstowcs_l(MSVCRT.@)
1973  */
1974 MSVCRT_size_t CDECL MSVCRT__mbstowcs_l(MSVCRT_wchar_t *wcstr, const char *mbstr,
1975         MSVCRT_size_t count, MSVCRT__locale_t locale)
1976 {
1977     MSVCRT_pthreadlocinfo locinfo;
1978     MSVCRT_size_t i, size;
1979
1980     if(!locale)
1981         locinfo = get_locinfo();
1982     else
1983         locinfo = locale->locinfo;
1984
1985     /* Ignore count parameter */
1986     if(!wcstr)
1987         return MultiByteToWideChar(locinfo->lc_codepage, 0, mbstr, -1, NULL, 0)-1;
1988
1989     for(i=0, size=0; i<count; i++) {
1990         if(mbstr[size] == '\0')
1991             break;
1992
1993         size += (MSVCRT__isleadbyte_l(mbstr[size], locale) ? 2 : 1);
1994     }
1995
1996     size = MultiByteToWideChar(locinfo->lc_codepage, 0,
1997             mbstr, size, wcstr, count);
1998
1999     if(size<count && wcstr)
2000         wcstr[size] = '\0';
2001
2002     return size;
2003 }
2004
2005 /*********************************************************************
2006  *              mbstowcs(MSVCRT.@)
2007  */
2008 MSVCRT_size_t CDECL MSVCRT_mbstowcs(MSVCRT_wchar_t *wcstr,
2009         const char *mbstr, MSVCRT_size_t count)
2010 {
2011     return MSVCRT__mbstowcs_l(wcstr, mbstr, count, NULL);
2012 }
2013
2014 /*********************************************************************
2015  *              _mbstowcs_s_l(MSVCRT.@)
2016  */
2017 int CDECL MSVCRT__mbstowcs_s_l(MSVCRT_size_t *ret, MSVCRT_wchar_t *wcstr,
2018         MSVCRT_size_t size, const char *mbstr,
2019         MSVCRT_size_t count, MSVCRT__locale_t locale)
2020 {
2021     MSVCRT_size_t conv;
2022
2023     if(!wcstr && !size) {
2024         conv = MSVCRT__mbstowcs_l(NULL, mbstr, 0, locale);
2025         if(ret)
2026             *ret = conv+1;
2027         return 0;
2028     }
2029
2030     if (!MSVCRT_CHECK_PMT(wcstr != NULL)) return MSVCRT_EINVAL;
2031     if (!MSVCRT_CHECK_PMT(mbstr != NULL)) {
2032         if(size) wcstr[0] = '\0';
2033         return MSVCRT_EINVAL;
2034     }
2035
2036     if(count==MSVCRT__TRUNCATE || size<count)
2037         conv = size;
2038     else
2039         conv = count;
2040
2041     conv = MSVCRT__mbstowcs_l(wcstr, mbstr, conv, locale);
2042     if(conv<size)
2043         wcstr[conv++] = '\0';
2044     else if(conv==size && (count==MSVCRT__TRUNCATE || wcstr[conv-1]=='\0'))
2045         wcstr[conv-1] = '\0';
2046     else {
2047         MSVCRT_INVALID_PMT("wcstr[size] is too small", MSVCRT_ERANGE);
2048         if(size)
2049             wcstr[0] = '\0';
2050         return MSVCRT_ERANGE;
2051     }
2052
2053     if(ret)
2054         *ret = conv;
2055     return 0;
2056 }
2057
2058 /*********************************************************************
2059  *              mbstowcs_s(MSVCRT.@)
2060  */
2061 int CDECL MSVCRT__mbstowcs_s(MSVCRT_size_t *ret, MSVCRT_wchar_t *wcstr,
2062         MSVCRT_size_t size, const char *mbstr, MSVCRT_size_t count)
2063 {
2064     return MSVCRT__mbstowcs_s_l(ret, wcstr, size, mbstr, count, NULL);
2065 }