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