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 "msvcrt.h"
  27 #include "wine/unicode.h"
  28 #include "wine/debug.h"
  29 #include "msvcrt/mbctype.h"
  30
  31 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
  32
  33 unsigned char MSVCRT_mbctype[257];
  34 int MSVCRT___mb_cur_max = 1;
  35
  36 static MSVCRT_wchar_t msvcrt_mbc_to_wc(unsigned int ch)
  37 {
  38   MSVCRT_wchar_t chW;
  39   char mbch[2];
  40   int n_chars;
  41
  42   if (ch <= 0xff) {
  43     mbch[0] = ch;
  44     n_chars = 1;
  45   } else {
  46     mbch[0] = (ch >> 8) & 0xff;
  47     mbch[1] = ch & 0xff;
  48     n_chars = 2;
  49   }
  50   if (!MultiByteToWideChar(MSVCRT___lc_codepage, 0, mbch, n_chars, &chW, 1))
  51   {
  52     WARN("MultiByteToWideChar failed on %x\n", ch);
  53     return 0;
  54   }
  55   return chW;
  56 }
  57
  58 static inline size_t u_strlen( const unsigned char *str )
  59 {
  60   return strlen( (const char*) str );
  61 }
  62
  63 static inline unsigned char* u_strncat( unsigned char* dst, const unsigned char* src, size_t len )
  64 {
  65   return (unsigned char*)strncat( (char*)dst, (const char*)src, len);
  66 }
  67
  68 static inline int u_strcmp( const unsigned char *s1, const unsigned char *s2 )
  69 {
  70   return strcmp( (const char*)s1, (const char*)s2 );
  71 }
  72
  73 static inline int u_strcasecmp( const unsigned char *s1, const unsigned char *s2 )
  74 {
  75   return strcasecmp( (const char*)s1, (const char*)s2 );
  76 }
  77
  78 static inline int u_strncmp( const unsigned char *s1, const unsigned char *s2, size_t len )
  79 {
  80   return strncmp( (const char*)s1, (const char*)s2, len );
  81 }
  82
  83 static inline int u_strncasecmp( const unsigned char *s1, const unsigned char *s2, size_t len )
  84 {
  85   return strncasecmp( (const char*)s1, (const char*)s2, len );
  86 }
  87
  88 static inline unsigned char *u_strchr( const unsigned char *s, unsigned char x )
  89 {
  90   return (unsigned char*) strchr( (const char*)s, x );
  91 }
  92
  93 static inline unsigned char *u_strrchr( const unsigned char *s, unsigned char x )
  94 {
  95   return (unsigned char*) strrchr( (const char*)s, x );
  96 }
  97
  98 static inline unsigned char *u_strtok( unsigned char *s, const unsigned char *delim )
  99 {
 100   return (unsigned char*) strtok( (char*)s, (const char*)delim );
 101 }
 102
 103 static inline unsigned char *u__strset( unsigned char *s, unsigned char c )
 104 {
 105   return (unsigned char*) _strset( (char*)s, c);
 106 }
 107
 108 static inline unsigned char *u__strnset( unsigned char *s, unsigned char c, size_t len )
 109 {
 110   return (unsigned char*) _strnset( (char*)s, c, len );
 111 }
 112
 113 static inline size_t u_strcspn( const unsigned char *s, const unsigned char *rej )
 114 {
 115   return strcspn( (const char *)s, (const char*)rej );
 116 }
 117
 118 /*********************************************************************
 119  *              __p__mbctype (MSVCRT.@)
 120  */
 121 unsigned char* CDECL __p__mbctype(void)
 122 {
 123   return MSVCRT_mbctype;
 124 }
 125
 126 /*********************************************************************
 127  *              __p___mb_cur_max(MSVCRT.@)
 128  */
 129 int* CDECL __p___mb_cur_max(void)
 130 {
 131   return &MSVCRT___mb_cur_max;
 132 }
 133
 134 /*********************************************************************
 135  *              _mbsnextc(MSVCRT.@)
 136  */
 137 unsigned int CDECL _mbsnextc(const unsigned char* str)
 138 {
 139   if(MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(*str))
 140     return *str << 8 | str[1];
 141   return *str; /* ASCII CP or SB char */
 142 }
 143
 144 /*********************************************************************
 145  *              _mbctolower(MSVCRT.@)
 146  */
 147 unsigned int CDECL _mbctolower(unsigned int c)
 148 {
 149     if (MSVCRT_isleadbyte(c))
 150     {
 151       FIXME("Handle MBC chars\n");
 152       return c;
 153     }
 154     return tolower(c); /* ASCII CP or SB char */
 155 }
 156
 157 /*********************************************************************
 158  *              _mbctoupper(MSVCRT.@)
 159  */
 160 unsigned int CDECL _mbctoupper(unsigned int c)
 161 {
 162     if (MSVCRT_isleadbyte(c))
 163     {
 164       FIXME("Handle MBC chars\n");
 165       return c;
 166     }
 167     return toupper(c); /* ASCII CP or SB char */
 168 }
 169
 170 /*********************************************************************
 171  *              _mbsdec(MSVCRT.@)
 172  */
 173 unsigned char* CDECL _mbsdec(const unsigned char* start, const unsigned char* cur)
 174 {
 175   if(MSVCRT___mb_cur_max > 1)
 176     return (unsigned char *)(_ismbstrail(start,cur-1) ? cur - 2 : cur -1);
 177
 178   return (unsigned char *)cur - 1; /* ASCII CP or SB char */
 179 }
 180
 181 /*********************************************************************
 182  *              _mbsinc(MSVCRT.@)
 183  */
 184 unsigned char* CDECL _mbsinc(const unsigned char* str)
 185 {
 186   if(MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(*str))
 187     return (unsigned char*)str + 2; /* MB char */
 188
 189   return (unsigned char*)str + 1; /* ASCII CP or SB char */
 190 }
 191
 192 /*********************************************************************
 193  *              _mbsninc(MSVCRT.@)
 194  */
 195 unsigned char* CDECL _mbsninc(const unsigned char* str, MSVCRT_size_t num)
 196 {
 197   if(!str || num < 1)
 198     return NULL;
 199   if(MSVCRT___mb_cur_max > 1)
 200   {
 201     while(num--)
 202       str = _mbsinc(str);
 203     return (unsigned char*)str;
 204   }
 205   return (unsigned char*)str + num; /* ASCII CP */
 206 }
 207
 208 /*********************************************************************
 209  *              _mbclen(MSVCRT.@)
 210  */
 211 unsigned int CDECL _mbclen(const unsigned char* str)
 212 {
 213   return MSVCRT_isleadbyte(*str) ? 2 : 1;
 214 }
 215
 216 /*********************************************************************
 217  *              mblen(MSVCRT.@)
 218  */
 219 int CDECL MSVCRT_mblen(const char* str, MSVCRT_size_t size)
 220 {
 221   if (str && *str && size)
 222   {
 223     if(MSVCRT___mb_cur_max == 1)
 224       return 1; /* ASCII CP */
 225
 226     return !MSVCRT_isleadbyte(*str) ? 1 : (size>1 ? 2 : -1);
 227   }
 228   return 0;
 229 }
 230
 231 /*********************************************************************
 232  *              _mbslen(MSVCRT.@)
 233  */
 234 MSVCRT_size_t CDECL _mbslen(const unsigned char* str)
 235 {
 236   if(MSVCRT___mb_cur_max > 1)
 237   {
 238     MSVCRT_size_t len = 0;
 239     while(*str)
 240     {
 241       str += MSVCRT_isleadbyte(*str) ? 2 : 1;
 242       len++;
 243     }
 244     return len;
 245   }
 246   return u_strlen(str); /* ASCII CP */
 247 }
 248
 249 /*********************************************************************
 250  *              _mbstrlen(MSVCRT.@)
 251  */
 252 MSVCRT_size_t CDECL _mbstrlen(const char* str)
 253 {
 254   if(MSVCRT___mb_cur_max > 1)
 255   {
 256     MSVCRT_size_t len = 0;
 257     while(*str)
 258     {
 259       /* FIXME: According to the documentation we are supposed to test for
 260        * multi-byte character validity. Whatever that means
 261        */
 262       str += MSVCRT_isleadbyte(*str) ? 2 : 1;
 263       len++;
 264     }
 265     return len;
 266   }
 267   return strlen(str); /* ASCII CP */
 268 }
 269
 270 /*********************************************************************
 271  *              _mbccpy(MSVCRT.@)
 272  */
 273 void CDECL _mbccpy(unsigned char* dest, const unsigned char* src)
 274 {
 275   *dest = *src;
 276   if(MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(*src))
 277     *++dest = *++src; /* MB char */
 278 }
 279
 280 /*********************************************************************
 281  *              _mbsncpy(MSVCRT.@)
 282  */
 283 unsigned char* CDECL _mbsncpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 284 {
 285   unsigned char* ret = dst;
 286   if(!n)
 287     return dst;
 288   if(MSVCRT___mb_cur_max > 1)
 289   {
 290     while (*src && n)
 291     {
 292       n--;
 293       *dst++ = *src;
 294       if (MSVCRT_isleadbyte(*src++))
 295           *dst++ = *src++;
 296     }
 297   }
 298   else
 299   {
 300     while (n)
 301     {
 302         n--;
 303         if (!(*dst++ = *src++)) break;
 304     }
 305   }
 306   while (n--) *dst++ = 0;
 307   return ret;
 308 }
 309
 310 /*********************************************************************
 311  *              _mbsnbcpy(MSVCRT.@)
 312  */
 313 unsigned char* CDECL _mbsnbcpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 314 {
 315   unsigned char* ret = dst;
 316   if(!n)
 317     return dst;
 318   if(MSVCRT___mb_cur_max > 1)
 319   {
 320     while (*src && (n > 1))
 321     {
 322       n--;
 323       *dst++ = *src;
 324       if (MSVCRT_isleadbyte(*src++))
 325       {
 326         *dst++ = *src++;
 327         n--;
 328       }
 329     }
 330     if (*src && n && !MSVCRT_isleadbyte(*src))
 331     {
 332       /* If the last character is a multi-byte character then
 333        * we cannot copy it since we have only one byte left
 334        */
 335       *dst++ = *src;
 336       n--;
 337     }
 338   }
 339   else
 340   {
 341     while (n)
 342     {
 343         n--;
 344         if (!(*dst++ = *src++)) break;
 345     }
 346   }
 347   while (n--) *dst++ = 0;
 348   return ret;
 349 }
 350
 351 /*********************************************************************
 352  *              _mbscmp(MSVCRT.@)
 353  */
 354 int CDECL _mbscmp(const unsigned char* str, const unsigned char* cmp)
 355 {
 356   if(MSVCRT___mb_cur_max > 1)
 357   {
 358     unsigned int strc, cmpc;
 359     do {
 360       if(!*str)
 361         return *cmp ? -1 : 0;
 362       if(!*cmp)
 363         return 1;
 364       strc = _mbsnextc(str);
 365       cmpc = _mbsnextc(cmp);
 366       if(strc != cmpc)
 367         return strc < cmpc ? -1 : 1;
 368       str +=(strc > 255) ? 2 : 1;
 369       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 370     } while(1);
 371   }
 372   return u_strcmp(str, cmp); /* ASCII CP */
 373 }
 374
 375 /*********************************************************************
 376  *              _mbsicoll(MSVCRT.@)
 377  * FIXME: handle locales.
 378  */
 379 int CDECL _mbsicoll(const unsigned char* str, const unsigned char* cmp)
 380 {
 381   if(MSVCRT___mb_cur_max > 1)
 382   {
 383     unsigned int strc, cmpc;
 384     do {
 385       if(!*str)
 386         return *cmp ? -1 : 0;
 387       if(!*cmp)
 388         return 1;
 389       strc = _mbctolower(_mbsnextc(str));
 390       cmpc = _mbctolower(_mbsnextc(cmp));
 391       if(strc != cmpc)
 392         return strc < cmpc ? -1 : 1;
 393       str +=(strc > 255) ? 2 : 1;
 394       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 395     } while(1);
 396   }
 397   return u_strcasecmp(str, cmp); /* ASCII CP */
 398 }
 399
 400 /*********************************************************************
 401  *              _mbscoll(MSVCRT.@)
 402  * Performs a case-sensitive comparison according to the current code page
 403  * RETURN
 404  *   _NLSCMPERROR if error
 405  * FIXME: handle locales.
 406  */
 407 int CDECL _mbscoll(const unsigned char* str, const unsigned char* cmp)
 408 {
 409   if(MSVCRT___mb_cur_max > 1)
 410   {
 411     unsigned int strc, cmpc;
 412     do {
 413       if(!*str)
 414         return *cmp ? -1 : 0;
 415       if(!*cmp)
 416         return 1;
 417       strc = _mbsnextc(str);
 418       cmpc = _mbsnextc(cmp);
 419       if(strc != cmpc)
 420         return strc < cmpc ? -1 : 1;
 421       str +=(strc > 255) ? 2 : 1;
 422       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 423     } while(1);
 424   }
 425   return u_strcmp(str, cmp); /* ASCII CP */
 426 }
 427
 428
 429 /*********************************************************************
 430  *              _mbsicmp(MSVCRT.@)
 431  */
 432 int CDECL _mbsicmp(const unsigned char* str, const unsigned char* cmp)
 433 {
 434   if(MSVCRT___mb_cur_max > 1)
 435   {
 436     unsigned int strc, cmpc;
 437     do {
 438       if(!*str)
 439         return *cmp ? -1 : 0;
 440       if(!*cmp)
 441         return 1;
 442       strc = _mbctolower(_mbsnextc(str));
 443       cmpc = _mbctolower(_mbsnextc(cmp));
 444       if(strc != cmpc)
 445         return strc < cmpc ? -1 : 1;
 446       str +=(strc > 255) ? 2 : 1;
 447       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 448     } while(1);
 449   }
 450   return u_strcasecmp(str, cmp); /* ASCII CP */
 451 }
 452
 453 /*********************************************************************
 454  *              _mbsncmp(MSVCRT.@)
 455  */
 456 int CDECL _mbsncmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 457 {
 458   if(!len)
 459     return 0;
 460
 461   if(MSVCRT___mb_cur_max > 1)
 462   {
 463     unsigned int strc, cmpc;
 464     while(len--)
 465     {
 466       int inc;
 467       if(!*str)
 468         return *cmp ? -1 : 0;
 469       if(!*cmp)
 470         return 1;
 471       strc = _mbsnextc(str);
 472       cmpc = _mbsnextc(cmp);
 473       if(strc != cmpc)
 474         return strc < cmpc ? -1 : 1;
 475       inc=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 476       str += inc;
 477       cmp += inc;
 478     }
 479     return 0; /* Matched len chars */
 480   }
 481   return u_strncmp(str, cmp, len); /* ASCII CP */
 482 }
 483
 484 /*********************************************************************
 485  *              _mbsnbcmp(MSVCRT.@)
 486  */
 487 int CDECL _mbsnbcmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 488 {
 489   if (!len)
 490     return 0;
 491   if(MSVCRT___mb_cur_max > 1)
 492   {
 493     unsigned int strc, cmpc;
 494     while (len)
 495     {
 496       int clen;
 497       if(!*str)
 498         return *cmp ? -1 : 0;
 499       if(!*cmp)
 500         return 1;
 501       if (MSVCRT_isleadbyte(*str))
 502       {
 503         strc=(len>=2)?_mbsnextc(str):0;
 504         clen=2;
 505       }
 506       else
 507       {
 508         strc=*str;
 509         clen=1;
 510       }
 511       if (MSVCRT_isleadbyte(*cmp))
 512         cmpc=(len>=2)?_mbsnextc(cmp):0;
 513       else
 514         cmpc=*str;
 515       if(strc != cmpc)
 516         return strc < cmpc ? -1 : 1;
 517       len -= clen;
 518       str += clen;
 519       cmp += clen;
 520     }
 521     return 0; /* Matched len chars */
 522   }
 523   return u_strncmp(str,cmp,len);
 524 }
 525
 526 /*********************************************************************
 527  *              _mbsnicmp(MSVCRT.@)
 528  *
 529  * Compare two multibyte strings case insensitively to 'len' characters.
 530  */
 531 int CDECL _mbsnicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 532 {
 533   /* FIXME: No tolower() for mb strings yet */
 534   if(MSVCRT___mb_cur_max > 1)
 535   {
 536     unsigned int strc, cmpc;
 537     while(len--)
 538     {
 539       if(!*str)
 540         return *cmp ? -1 : 0;
 541       if(!*cmp)
 542         return 1;
 543       strc = _mbctolower(_mbsnextc(str));
 544       cmpc = _mbctolower(_mbsnextc(cmp));
 545       if(strc != cmpc)
 546         return strc < cmpc ? -1 : 1;
 547       str +=(strc > 255) ? 2 : 1;
 548       cmp +=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 549     }
 550     return 0; /* Matched len chars */
 551   }
 552   return u_strncasecmp(str, cmp, len); /* ASCII CP */
 553 }
 554
 555 /*********************************************************************
 556  *              _mbsnbicmp(MSVCRT.@)
 557  */
 558 int CDECL _mbsnbicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 559 {
 560   if (!len)
 561     return 0;
 562   if(MSVCRT___mb_cur_max > 1)
 563   {
 564     unsigned int strc, cmpc;
 565     while (len)
 566     {
 567       int clen;
 568       if(!*str)
 569         return *cmp ? -1 : 0;
 570       if(!*cmp)
 571         return 1;
 572       if (MSVCRT_isleadbyte(*str))
 573       {
 574         strc=(len>=2)?_mbsnextc(str):0;
 575         clen=2;
 576       }
 577       else
 578       {
 579         strc=*str;
 580         clen=1;
 581       }
 582       if (MSVCRT_isleadbyte(*cmp))
 583         cmpc=(len>=2)?_mbsnextc(cmp):0;
 584       else
 585         cmpc=*str;
 586       strc = _mbctolower(strc);
 587       cmpc = _mbctolower(cmpc);
 588       if(strc != cmpc)
 589         return strc < cmpc ? -1 : 1;
 590       len -= clen;
 591       str += clen;
 592       cmp += clen;
 593     }
 594     return 0; /* Matched len bytes */
 595   }
 596   return u_strncmp(str,cmp,len);
 597 }
 598
 599 /*********************************************************************
 600  *              _mbscat (MSVCRT.@)
 601  */
 602 unsigned char * CDECL _mbscat( unsigned char *dst, const unsigned char *src )
 603 {
 604     strcat( (char *)dst, (const char *)src );
 605     return dst;
 606 }
 607
 608 /*********************************************************************
 609  *              _mbscpy (MSVCRT.@)
 610  */
 611 unsigned char* CDECL _mbscpy( unsigned char *dst, const unsigned char *src )
 612 {
 613     strcpy( (char *)dst, (const char *)src );
 614     return dst;
 615 }
 616
 617 /*********************************************************************
 618  *              _mbsstr (MSVCRT.@)
 619  */
 620 unsigned char * CDECL _mbsstr(const unsigned char *haystack, const unsigned char *needle)
 621 {
 622     return (unsigned char *)strstr( (const char *)haystack, (const char *)needle );
 623 }
 624
 625 /*********************************************************************
 626  *              _mbschr(MSVCRT.@)
 627  *
 628  * Find a multibyte character in a multibyte string.
 629  */
 630 unsigned char* CDECL _mbschr(const unsigned char* s, unsigned int x)
 631 {
 632   if(MSVCRT___mb_cur_max > 1)
 633   {
 634     unsigned int c;
 635     while (1)
 636     {
 637       c = _mbsnextc(s);
 638       if (c == x)
 639         return (unsigned char*)s;
 640       if (!c)
 641         return NULL;
 642       s += c > 255 ? 2 : 1;
 643     }
 644   }
 645   return u_strchr(s, x); /* ASCII CP */
 646 }
 647
 648 /*********************************************************************
 649  *              _mbsrchr(MSVCRT.@)
 650  */
 651 unsigned char* CDECL _mbsrchr(const unsigned char* s, unsigned int x)
 652 {
 653   if(MSVCRT___mb_cur_max > 1)
 654   {
 655     unsigned int c;
 656     unsigned char* match=NULL;
 657     if(!s)
 658       return NULL;
 659     while (1) {
 660       c = _mbsnextc(s);
 661       if (c == x)
 662         match=(unsigned char*)s;
 663       if (!c)
 664         return match;
 665       s +=(c > 255) ? 2 : 1;
 666     }
 667   }
 668   return u_strrchr(s, x);
 669 }
 670
 671 /*********************************************************************
 672  *              _mbstok(MSVCRT.@)
 673  *
 674  * Find and extract tokens from strings
 675  */
 676 unsigned char* CDECL _mbstok(unsigned char *str, const unsigned char *delim)
 677 {
 678     thread_data_t *data = msvcrt_get_thread_data();
 679     unsigned char *ret;
 680
 681     if(MSVCRT___mb_cur_max > 1)
 682     {
 683         unsigned int c;
 684
 685         if (!str)
 686             if (!(str = data->mbstok_next)) return NULL;
 687
 688         while ((c = _mbsnextc(str)) && _mbschr(delim, c)) {
 689             str += c > 255 ? 2 : 1;
 690         }
 691         if (!*str) return NULL;
 692         ret = str++;
 693         while ((c = _mbsnextc(str)) && !_mbschr(delim, c)) {
 694             str += c > 255 ? 2 : 1;
 695         }
 696         if (*str) {
 697             *str++ = 0;
 698             if (c > 255) *str++ = 0;
 699         }
 700         data->mbstok_next = str;
 701         return ret;
 702     }
 703     return u_strtok(str, delim); /* ASCII CP */
 704 }
 705
 706 /*********************************************************************
 707  *              mbtowc(MSVCRT.@)
 708  */
 709 int CDECL MSVCRT_mbtowc(MSVCRT_wchar_t *dst, const char* str, MSVCRT_size_t n)
 710 {
 711     /* temp var needed because MultiByteToWideChar wants non NULL destination */
 712     MSVCRT_wchar_t tmpdst = '\0';
 713
 714     if(n <= 0 || !str)
 715         return 0;
 716     if(!MultiByteToWideChar(CP_ACP, 0, str, n, &tmpdst, 1))
 717         return -1;
 718     if(dst)
 719         *dst = tmpdst;
 720     /* return the number of bytes from src that have been used */
 721     if(!*str)
 722         return 0;
 723     if(n >= 2 && MSVCRT_isleadbyte(*str) && str[1])
 724         return 2;
 725     return 1;
 726 }
 727
 728 /*********************************************************************
 729  *              _mbbtombc(MSVCRT.@)
 730  */
 731 unsigned int CDECL _mbbtombc(unsigned int c)
 732 {
 733   if(MSVCRT___mb_cur_max > 1 &&
 734      ((c >= 0x20 && c <=0x7e) ||(c >= 0xa1 && c <= 0xdf)))
 735   {
 736     /* FIXME: I can't get this function to return anything
 737      * different from what I pass it...
 738      */
 739   }
 740   return c;  /* ASCII CP or no MB char */
 741 }
 742
 743 /*********************************************************************
 744  *              _mbbtype(MSVCRT.@)
 745  */
 746 int CDECL _mbbtype(unsigned char c, int type)
 747 {
 748     if (type == 1)
 749     {
 750         if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
 751             return _MBC_SINGLE;
 752         else if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
 753             return _MBC_TRAIL;
 754         else
 755             return _MBC_ILLEGAL;
 756     }
 757     else
 758     {
 759         if ((c >= 0x20 && c <= 0x7e) || (c >= 0xa1 && c <= 0xdf))
 760             return _MBC_SINGLE;
 761         else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
 762             return _MBC_LEAD;
 763         else
 764             return _MBC_ILLEGAL;
 765     }
 766 }
 767
 768 /*********************************************************************
 769  *              _ismbbkana(MSVCRT.@)
 770  */
 771 int CDECL _ismbbkana(unsigned int c)
 772 {
 773   /* FIXME: use lc_ctype when supported, not lc_all */
 774   if(MSVCRT___lc_codepage == 932)
 775   {
 776     /* Japanese/Katakana, CP 932 */
 777     return (c >= 0xa1 && c <= 0xdf);
 778   }
 779   return 0;
 780 }
 781
 782 /*********************************************************************
 783  *              _ismbcdigit(MSVCRT.@)
 784  */
 785 int CDECL _ismbcdigit(unsigned int ch)
 786 {
 787     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 788     return (get_char_typeW( wch ) & C1_DIGIT);
 789 }
 790
 791 /*********************************************************************
 792  *              _ismbcgraph(MSVCRT.@)
 793  */
 794 int CDECL _ismbcgraph(unsigned int ch)
 795 {
 796     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 797     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA));
 798 }
 799
 800 /*********************************************************************
 801  *              _ismbcalpha (MSVCRT.@)
 802  */
 803 int CDECL _ismbcalpha(unsigned int ch)
 804 {
 805     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 806     return (get_char_typeW( wch ) & C1_ALPHA);
 807 }
 808
 809 /*********************************************************************
 810  *              _ismbclower (MSVCRT.@)
 811  */
 812 int CDECL _ismbclower(unsigned int ch)
 813 {
 814     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 815     return (get_char_typeW( wch ) & C1_UPPER);
 816 }
 817
 818 /*********************************************************************
 819  *              _ismbcupper (MSVCRT.@)
 820  */
 821 int CDECL _ismbcupper(unsigned int ch)
 822 {
 823     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 824     return (get_char_typeW( wch ) & C1_LOWER);
 825 }
 826
 827 /*********************************************************************
 828  *              _ismbcsymbol(MSVCRT.@)
 829  */
 830 int CDECL _ismbcsymbol(unsigned int ch)
 831 {
 832     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 833     WORD ctype;
 834     if (!GetStringTypeW(CT_CTYPE3, &wch, 1, &ctype))
 835     {
 836         WARN("GetStringTypeW failed on %x\n", ch);
 837         return 0;
 838     }
 839     return ((ctype & C3_SYMBOL) != 0);
 840 }
 841
 842 /*********************************************************************
 843  *              _ismbcalnum (MSVCRT.@)
 844  */
 845 int CDECL _ismbcalnum(unsigned int ch)
 846 {
 847     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 848     return (get_char_typeW( wch ) & (C1_ALPHA | C1_DIGIT));
 849 }
 850
 851 /*********************************************************************
 852  *              _ismbcspace (MSVCRT.@)
 853  */
 854 int CDECL _ismbcspace(unsigned int ch)
 855 {
 856     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 857     return (get_char_typeW( wch ) & C1_SPACE);
 858 }
 859
 860 /*********************************************************************
 861  *              _ismbcprint (MSVCRT.@)
 862  */
 863 int CDECL _ismbcprint(unsigned int ch)
 864 {
 865     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 866     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA | C1_SPACE));
 867 }
 868
 869 /*********************************************************************
 870  *              _ismbcpunct(MSVCRT.@)
 871  */
 872 int CDECL _ismbcpunct(unsigned int ch)
 873 {
 874     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 875     return (get_char_typeW( wch ) & C1_PUNCT);
 876 }
 877
 878 /*********************************************************************
 879  *              _ismbchira(MSVCRT.@)
 880  */
 881 int CDECL _ismbchira(unsigned int c)
 882 {
 883   /* FIXME: use lc_ctype when supported, not lc_all */
 884   if(MSVCRT___lc_codepage == 932)
 885   {
 886     /* Japanese/Hiragana, CP 932 */
 887     return (c >= 0x829f && c <= 0x82f1);
 888   }
 889   return 0;
 890 }
 891
 892 /*********************************************************************
 893  *              _ismbckata(MSVCRT.@)
 894  */
 895 int CDECL _ismbckata(unsigned int c)
 896 {
 897   /* FIXME: use lc_ctype when supported, not lc_all */
 898   if(MSVCRT___lc_codepage == 932)
 899   {
 900     if(c < 256)
 901       return _ismbbkana(c);
 902     /* Japanese/Katakana, CP 932 */
 903     return (c >= 0x8340 && c <= 0x8396 && c != 0x837f);
 904   }
 905   return 0;
 906 }
 907
 908 /*********************************************************************
 909  *              _ismbblead(MSVCRT.@)
 910  */
 911 int CDECL _ismbblead(unsigned int c)
 912 {
 913   /* FIXME: should reference MSVCRT_mbctype */
 914   return MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(c);
 915 }
 916
 917
 918 /*********************************************************************
 919  *              _ismbbtrail(MSVCRT.@)
 920  */
 921 int CDECL _ismbbtrail(unsigned int c)
 922 {
 923   /* FIXME: should reference MSVCRT_mbctype */
 924   return !_ismbblead(c);
 925 }
 926
 927 /*********************************************************************
 928  *              _ismbslead(MSVCRT.@)
 929  */
 930 int CDECL _ismbslead(const unsigned char* start, const unsigned char* str)
 931 {
 932   /* Lead bytes can also be trail bytes if caller messed up
 933    * iterating through the string...
 934    */
 935   if(MSVCRT___mb_cur_max > 1)
 936   {
 937     while(start < str)
 938       start += MSVCRT_isleadbyte(*str) ? 2 : 1;
 939
 940     if(start == str)
 941       return MSVCRT_isleadbyte(*str);
 942   }
 943   return 0; /* Must have been a trail, we skipped it */
 944 }
 945
 946 /*********************************************************************
 947  *              _ismbstrail(MSVCRT.@)
 948  */
 949 int CDECL _ismbstrail(const unsigned char* start, const unsigned char* str)
 950 {
 951   /* Must not be a lead, and must be preceded by one */
 952   return !_ismbslead(start, str) && MSVCRT_isleadbyte(str[-1]);
 953 }
 954
 955 /*********************************************************************
 956  *              _mbsset(MSVCRT.@)
 957  */
 958 unsigned char* CDECL _mbsset(unsigned char* str, unsigned int c)
 959 {
 960   unsigned char* ret = str;
 961
 962   if(MSVCRT___mb_cur_max == 1 || c < 256)
 963     return u__strset(str, c); /* ASCII CP or SB char */
 964
 965   c &= 0xffff; /* Strip high bits */
 966
 967   while(str[0] && str[1])
 968   {
 969     *str++ = c >> 8;
 970     *str++ = c & 0xff;
 971   }
 972   if(str[0])
 973     str[0] = '\0'; /* FIXME: OK to shorten? */
 974
 975   return ret;
 976 }
 977
 978 /*********************************************************************
 979  *              _mbsnbset(MSVCRT.@)
 980  */
 981 unsigned char* CDECL _mbsnbset(unsigned char *str, unsigned int c, MSVCRT_size_t len)
 982 {
 983     unsigned char *ret = str;
 984
 985     if(!len)
 986         return ret;
 987
 988     if(MSVCRT___mb_cur_max == 1 || c < 256)
 989         return u__strnset(str, c, len); /* ASCII CP or SB char */
 990
 991     c &= 0xffff; /* Strip high bits */
 992
 993     while(str[0] && str[1] && (len > 1))
 994     {
 995         *str++ = c >> 8;
 996         len--;
 997         *str++ = c & 0xff;
 998         len--;
 999     }
1000     if(len && str[0]) {
1001         /* as per msdn pad with a blank character */
1002         str[0] = ' ';
1003     }
1004
1005     return ret;
1006 }
1007
1008 /*********************************************************************
1009  *              _mbsnset(MSVCRT.@)
1010  */
1011 unsigned char* CDECL _mbsnset(unsigned char* str, unsigned int c, MSVCRT_size_t len)
1012 {
1013   unsigned char *ret = str;
1014
1015   if(!len)
1016     return ret;
1017
1018   if(MSVCRT___mb_cur_max == 1 || c < 256)
1019     return u__strnset(str, c, len); /* ASCII CP or SB char */
1020
1021   c &= 0xffff; /* Strip high bits */
1022
1023   while(str[0] && str[1] && len--)
1024   {
1025     *str++ = c >> 8;
1026     *str++ = c & 0xff;
1027   }
1028   if(len && str[0])
1029     str[0] = '\0'; /* FIXME: OK to shorten? */
1030
1031   return ret;
1032 }
1033
1034 /*********************************************************************
1035  *              _mbsnccnt(MSVCRT.@)
1036  * 'c' is for 'character'.
1037  */
1038 MSVCRT_size_t CDECL _mbsnccnt(const unsigned char* str, MSVCRT_size_t len)
1039 {
1040   MSVCRT_size_t ret;
1041   if(MSVCRT___mb_cur_max > 1)
1042   {
1043     ret=0;
1044     while(*str && len-- > 0)
1045     {
1046       if(MSVCRT_isleadbyte(*str))
1047       {
1048         if (!len)
1049           break;
1050         len--;
1051         str++;
1052       }
1053       str++;
1054       ret++;
1055     }
1056     return ret;
1057   }
1058   ret=u_strlen(str);
1059   return min(ret, len); /* ASCII CP */
1060 }
1061
1062 /*********************************************************************
1063  *              _mbsnbcnt(MSVCRT.@)
1064  * 'b' is for byte count.
1065  */
1066 MSVCRT_size_t CDECL _mbsnbcnt(const unsigned char* str, MSVCRT_size_t len)
1067 {
1068   MSVCRT_size_t ret;
1069   if(MSVCRT___mb_cur_max > 1)
1070   {
1071     const unsigned char* xstr = str;
1072     while(*xstr && len-- > 0)
1073     {
1074       if (MSVCRT_isleadbyte(*xstr++))
1075         xstr++;
1076     }
1077     return xstr-str;
1078   }
1079   ret=u_strlen(str);
1080   return min(ret, len); /* ASCII CP */
1081 }
1082
1083 /*********************************************************************
1084  *              _mbsnbcat(MSVCRT.@)
1085  */
1086 unsigned char* CDECL _mbsnbcat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1087 {
1088     if(MSVCRT___mb_cur_max > 1)
1089     {
1090         unsigned char *res = dst;
1091         while (*dst) {
1092             if (MSVCRT_isleadbyte(*dst++)) {
1093                 if (*dst) {
1094                     dst++;
1095                 } else {
1096                     /* as per msdn overwrite the lead byte in front of '\0' */
1097                     dst--;
1098                     break;
1099                 }
1100             }
1101         }
1102         while (*src && len--) *dst++ = *src++;
1103         *dst = '\0';
1104         return res;
1105     }
1106     return u_strncat(dst, src, len); /* ASCII CP */
1107 }
1108
1109
1110 /*********************************************************************
1111  *              _mbsncat(MSVCRT.@)
1112  */
1113 unsigned char* CDECL _mbsncat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
1114 {
1115   if(MSVCRT___mb_cur_max > 1)
1116   {
1117     unsigned char *res = dst;
1118     while (*dst)
1119     {
1120       if (MSVCRT_isleadbyte(*dst++))
1121         dst++;
1122     }
1123     while (*src && len--)
1124     {
1125       *dst++ = *src;
1126       if(MSVCRT_isleadbyte(*src++))
1127         *dst++ = *src++;
1128     }
1129     *dst = '\0';
1130     return res;
1131   }
1132   return u_strncat(dst, src, len); /* ASCII CP */
1133 }
1134
1135
1136 /*********************************************************************
1137  *              _mbslwr(MSVCRT.@)
1138  */
1139 unsigned char* CDECL _mbslwr(unsigned char* s)
1140 {
1141   unsigned char *ret = s;
1142   if (!s)
1143     return NULL;
1144   if (MSVCRT___mb_cur_max > 1)
1145   {
1146     unsigned int c;
1147     while (*s)
1148     {
1149       c = _mbctolower(_mbsnextc(s));
1150       /* Note that I assume that the size of the character is unchanged */
1151       if (c > 255)
1152       {
1153           *s++=(c>>8);
1154           c=c & 0xff;
1155       }
1156       *s++=c;
1157     }
1158   }
1159   else for ( ; *s; s++) *s = tolower(*s);
1160   return ret;
1161 }
1162
1163
1164 /*********************************************************************
1165  *              _mbsupr(MSVCRT.@)
1166  */
1167 unsigned char* CDECL _mbsupr(unsigned char* s)
1168 {
1169   unsigned char *ret = s;
1170   if (!s)
1171     return NULL;
1172   if (MSVCRT___mb_cur_max > 1)
1173   {
1174     unsigned int c;
1175     while (*s)
1176     {
1177       c = _mbctoupper(_mbsnextc(s));
1178       /* Note that I assume that the size of the character is unchanged */
1179       if (c > 255)
1180       {
1181           *s++=(c>>8);
1182           c=c & 0xff;
1183       }
1184       *s++=c;
1185     }
1186   }
1187   else for ( ; *s; s++) *s = toupper(*s);
1188   return ret;
1189 }
1190
1191
1192 /*********************************************************************
1193  *              _mbsspn (MSVCRT.@)
1194  */
1195 MSVCRT_size_t CDECL _mbsspn(const unsigned char* string, const unsigned char* set)
1196 {
1197     const unsigned char *p, *q;
1198
1199     for (p = string; *p; p++)
1200     {
1201         if (MSVCRT_isleadbyte(*p))
1202         {
1203             for (q = set; *q; q++)
1204             {
1205                 if (!q[1])
1206                     break;
1207                 if ((*p == *q) &&  (p[1] == q[1]))
1208                     break;
1209                 q++;
1210             }
1211             if (!q[0] || !q[1]) break;
1212         }
1213         else
1214         {
1215             for (q = set; *q; q++)
1216                 if (*p == *q)
1217                     break;
1218             if (!*q) break;
1219         }
1220     }
1221     return p - string;
1222 }
1223
1224 /*********************************************************************
1225  *              _mbsspnp (MSVCRT.@)
1226  */
1227 const unsigned char* CDECL _mbsspnp(const unsigned char* string, const unsigned char* set)
1228 {
1229     const unsigned char *p, *q;
1230
1231     for (p = string; *p; p++)
1232     {
1233         if (MSVCRT_isleadbyte(*p))
1234         {
1235             for (q = set; *q; q++)
1236             {
1237                 if (!q[1])
1238                     break;
1239                 if ((*p == *q) &&  (p[1] == q[1]))
1240                     break;
1241                 q++;
1242             }
1243             if (!q[0] || !q[1]) break;
1244         }
1245         else
1246         {
1247             for (q = set; *q; q++)
1248                 if (*p == *q)
1249                     break;
1250             if (!*q) break;
1251         }
1252     }
1253     if (*p == '\0')
1254         return NULL;
1255     return p;
1256 }
1257
1258 /*********************************************************************
1259  *              _mbscspn(MSVCRT.@)
1260  */
1261 MSVCRT_size_t CDECL _mbscspn(const unsigned char* str, const unsigned char* cmp)
1262 {
1263   if (MSVCRT___mb_cur_max > 1)
1264     FIXME("don't handle double character case\n");
1265   return u_strcspn(str, cmp);
1266 }
1267
1268 /*********************************************************************
1269  *              _mbsrev (MSVCRT.@)
1270  */
1271 unsigned char* CDECL _mbsrev(unsigned char* str)
1272 {
1273     int i, len = _mbslen(str);
1274     unsigned char *p, *temp=MSVCRT_malloc(len*2);
1275
1276     if(!temp)
1277         return str;
1278
1279     /* unpack multibyte string to temp buffer */
1280     p=str;
1281     for(i=0; i<len; i++)
1282     {
1283         if (MSVCRT_isleadbyte(*p))
1284         {
1285             temp[i*2]=*p++;
1286             temp[i*2+1]=*p++;
1287         }
1288         else
1289         {
1290             temp[i*2]=*p++;
1291             temp[i*2+1]=0;
1292         }
1293     }
1294
1295     /* repack it in the reverse order */
1296     p=str;
1297     for(i=len-1; i>=0; i--)
1298     {
1299         if(MSVCRT_isleadbyte(temp[i*2]))
1300         {
1301             *p++=temp[i*2];
1302             *p++=temp[i*2+1];
1303         }
1304         else
1305         {
1306             *p++=temp[i*2];
1307         }
1308     }
1309
1310     MSVCRT_free(temp);
1311
1312     return str;
1313 }
1314
1315 /*********************************************************************
1316  *              _mbspbrk (MSVCRT.@)
1317  */
1318 unsigned char* CDECL _mbspbrk(const unsigned char* str, const unsigned char* accept)
1319 {
1320     const unsigned char* p;
1321
1322     while(*str)
1323     {
1324         for(p = accept; *p; p += (MSVCRT_isleadbyte(*p)?2:1) )
1325         {
1326             if (*p == *str)
1327                 if( !MSVCRT_isleadbyte(*p) || ( *(p+1) == *(str+1) ) )
1328                      return (unsigned char*)str;
1329         }
1330         str += (MSVCRT_isleadbyte(*str)?2:1);
1331     }
1332     return NULL;
1333 }