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., 59 Temple Place, Suite 330, Boston, MA  02111-1307  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
  30 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
  31
  32 unsigned char MSVCRT_mbctype[257];
  33 int MSVCRT___mb_cur_max = 1;
  34
  35 static MSVCRT_wchar_t msvcrt_mbc_to_wc(unsigned int ch)
  36 {
  37   MSVCRT_wchar_t chW;
  38   char mbch[2];
  39   int n_chars;
  40
  41   if (ch <= 0xff) {
  42     mbch[0] = ch;
  43     n_chars = 1;
  44   } else {
  45     mbch[0] = (ch >> 8) & 0xff;
  46     mbch[1] = ch & 0xff;
  47     n_chars = 2;
  48   }
  49   if (!MultiByteToWideChar(msvcrt_current_lc_all_cp, 0, mbch, n_chars, &chW, 1))
  50   {
  51     WARN("MultiByteToWideChar failed on %x\n", ch);
  52     return 0;
  53   }
  54   return chW;
  55 }
  56
  57 /*********************************************************************
  58  *              __p__mbctype (MSVCRT.@)
  59  */
  60 unsigned char* __p__mbctype(void)
  61 {
  62   return MSVCRT_mbctype;
  63 }
  64
  65 /*********************************************************************
  66  *              __p___mb_cur_max(MSVCRT.@)
  67  */
  68 int* __p___mb_cur_max(void)
  69 {
  70   return &MSVCRT___mb_cur_max;
  71 }
  72
  73 /*********************************************************************
  74  *              _mbsnextc(MSVCRT.@)
  75  */
  76 unsigned int _mbsnextc(const unsigned char* str)
  77 {
  78   if(MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(*str))
  79     return *str << 8 | str[1];
  80   return *str; /* ASCII CP or SB char */
  81 }
  82
  83 /*********************************************************************
  84  *              _mbctolower(MSVCRT.@)
  85  */
  86 unsigned int _mbctolower(unsigned int c)
  87 {
  88     if (MSVCRT_isleadbyte(c))
  89     {
  90       FIXME("Handle MBC chars\n");
  91       return c;
  92     }
  93     return tolower(c); /* ASCII CP or SB char */
  94 }
  95
  96 /*********************************************************************
  97  *              _mbctoupper(MSVCRT.@)
  98  */
  99 unsigned int _mbctoupper(unsigned int c)
 100 {
 101     if (MSVCRT_isleadbyte(c))
 102     {
 103       FIXME("Handle MBC chars\n");
 104       return c;
 105     }
 106     return toupper(c); /* ASCII CP or SB char */
 107 }
 108
 109 /*********************************************************************
 110  *              _mbsdec(MSVCRT.@)
 111  */
 112 unsigned char* _mbsdec(const unsigned char* start, const unsigned char* cur)
 113 {
 114   if(MSVCRT___mb_cur_max > 1)
 115     return (unsigned char *)(_ismbstrail(start,cur-1) ? cur - 2 : cur -1);
 116
 117   return (unsigned char *)cur - 1; /* ASCII CP or SB char */
 118 }
 119
 120 /*********************************************************************
 121  *              _mbsinc(MSVCRT.@)
 122  */
 123 unsigned char* _mbsinc(const unsigned char* str)
 124 {
 125   if(MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(*str))
 126     return (unsigned char*)str + 2; /* MB char */
 127
 128   return (unsigned char*)str + 1; /* ASCII CP or SB char */
 129 }
 130
 131 /*********************************************************************
 132  *              _mbsninc(MSVCRT.@)
 133  */
 134 unsigned char* _mbsninc(const unsigned char* str, MSVCRT_size_t num)
 135 {
 136   if(!str || num < 1)
 137     return NULL;
 138   if(MSVCRT___mb_cur_max > 1)
 139   {
 140     while(num--)
 141       str = _mbsinc(str);
 142     return (unsigned char*)str;
 143   }
 144   return (unsigned char*)str + num; /* ASCII CP */
 145 }
 146
 147 /*********************************************************************
 148  *              _mbclen(MSVCRT.@)
 149  */
 150 unsigned int _mbclen(const unsigned char* str)
 151 {
 152   return MSVCRT_isleadbyte(*str) ? 2 : 1;
 153 }
 154
 155 /*********************************************************************
 156  *              mblen(MSVCRT.@)
 157  */
 158 int MSVCRT_mblen(const char* str, MSVCRT_size_t size)
 159 {
 160   if (str && *str && size)
 161   {
 162     if(MSVCRT___mb_cur_max == 1)
 163       return 1; /* ASCII CP */
 164
 165     return !MSVCRT_isleadbyte(*str) ? 1 : (size>1 ? 2 : -1);
 166   }
 167   return 0;
 168 }
 169
 170 /*********************************************************************
 171  *              _mbslen(MSVCRT.@)
 172  */
 173 MSVCRT_size_t _mbslen(const unsigned char* str)
 174 {
 175   if(MSVCRT___mb_cur_max > 1)
 176   {
 177     MSVCRT_size_t len = 0;
 178     while(*str)
 179     {
 180       str += MSVCRT_isleadbyte(*str) ? 2 : 1;
 181       len++;
 182     }
 183     return len;
 184   }
 185   return strlen(str); /* ASCII CP */
 186 }
 187
 188 /*********************************************************************
 189  *              _mbstrlen(MSVCRT.@)
 190  */
 191 MSVCRT_size_t _mbstrlen(const char* str)
 192 {
 193   if(MSVCRT___mb_cur_max > 1)
 194   {
 195     MSVCRT_size_t len = 0;
 196     while(*str)
 197     {
 198       /* FIXME: According to the documentation we are supposed to test for
 199        * multi-byte character validity. Whatever that means
 200        */
 201       str += MSVCRT_isleadbyte(*str) ? 2 : 1;
 202       len++;
 203     }
 204     return len;
 205   }
 206   return strlen(str); /* ASCII CP */
 207 }
 208
 209 /*********************************************************************
 210  *              _mbccpy(MSVCRT.@)
 211  */
 212 void _mbccpy(unsigned char* dest, const unsigned char* src)
 213 {
 214   *dest++ = *src;
 215   if(MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(*src))
 216     *dest = *++src; /* MB char */
 217   else
 218     ERR("failure.. is this ok?\n");
 219 }
 220
 221 /*********************************************************************
 222  *              _mbsncpy(MSVCRT.@)
 223  */
 224 unsigned char* _mbsncpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 225 {
 226   unsigned char* ret = dst;
 227   if(!n)
 228     return dst;
 229   if(MSVCRT___mb_cur_max > 1)
 230   {
 231     while (*src && n)
 232     {
 233       n--;
 234       *dst++ = *src;
 235       if (MSVCRT_isleadbyte(*src++))
 236           *dst++ = *src++;
 237     }
 238   }
 239   else
 240   {
 241     while (n)
 242     {
 243         n--;
 244         if (!(*dst++ = *src++)) break;
 245     }
 246   }
 247   while (n--) *dst++ = 0;
 248   return ret;
 249 }
 250
 251 /*********************************************************************
 252  *              _mbsnbcpy(MSVCRT.@)
 253  */
 254 unsigned char* _mbsnbcpy(unsigned char* dst, const unsigned char* src, MSVCRT_size_t n)
 255 {
 256   unsigned char* ret = dst;
 257   if(!n)
 258     return dst;
 259   if(MSVCRT___mb_cur_max > 1)
 260   {
 261     while (*src && (n > 1))
 262     {
 263       n--;
 264       *dst++ = *src;
 265       if (MSVCRT_isleadbyte(*src++))
 266       {
 267         *dst++ = *src++;
 268         n--;
 269       }
 270     }
 271     if (*src && n && !MSVCRT_isleadbyte(*src))
 272     {
 273       /* If the last character is a multi-byte character then
 274        * we cannot copy it since we have only one byte left
 275        */
 276       *dst++ = *src;
 277       n--;
 278     }
 279   }
 280   else
 281   {
 282     while (n)
 283     {
 284         n--;
 285         if (!(*dst++ = *src++)) break;
 286     }
 287   }
 288   while (n--) *dst++ = 0;
 289   return ret;
 290 }
 291
 292 /*********************************************************************
 293  *              _mbscmp(MSVCRT.@)
 294  */
 295 int _mbscmp(const unsigned char* str, const unsigned char* cmp)
 296 {
 297   if(MSVCRT___mb_cur_max > 1)
 298   {
 299     unsigned int strc, cmpc;
 300     do {
 301       if(!*str)
 302         return *cmp ? -1 : 0;
 303       if(!*cmp)
 304         return 1;
 305       strc = _mbsnextc(str);
 306       cmpc = _mbsnextc(cmp);
 307       if(strc != cmpc)
 308         return strc < cmpc ? -1 : 1;
 309       str +=(strc > 255) ? 2 : 1;
 310       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 311     } while(1);
 312   }
 313   return strcmp(str, cmp); /* ASCII CP */
 314 }
 315
 316 /*********************************************************************
 317  *              _mbsicoll(MSVCRT.@)
 318  * FIXME: handle locales.
 319  */
 320 int _mbsicoll(const unsigned char* str, const unsigned char* cmp)
 321 {
 322   if(MSVCRT___mb_cur_max > 1)
 323   {
 324     unsigned int strc, cmpc;
 325     do {
 326       if(!*str)
 327         return *cmp ? -1 : 0;
 328       if(!*cmp)
 329         return 1;
 330       strc = _mbctolower(_mbsnextc(str));
 331       cmpc = _mbctolower(_mbsnextc(cmp));
 332       if(strc != cmpc)
 333         return strc < cmpc ? -1 : 1;
 334       str +=(strc > 255) ? 2 : 1;
 335       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 336     } while(1);
 337   }
 338   return strcasecmp(str, cmp); /* ASCII CP */
 339 }
 340
 341
 342 /*********************************************************************
 343  *              _mbsicmp(MSVCRT.@)
 344  */
 345 int _mbsicmp(const unsigned char* str, const unsigned char* cmp)
 346 {
 347   if(MSVCRT___mb_cur_max > 1)
 348   {
 349     unsigned int strc, cmpc;
 350     do {
 351       if(!*str)
 352         return *cmp ? -1 : 0;
 353       if(!*cmp)
 354         return 1;
 355       strc = _mbctolower(_mbsnextc(str));
 356       cmpc = _mbctolower(_mbsnextc(cmp));
 357       if(strc != cmpc)
 358         return strc < cmpc ? -1 : 1;
 359       str +=(strc > 255) ? 2 : 1;
 360       cmp +=(strc > 255) ? 2 : 1; /* equal, use same increment */
 361     } while(1);
 362   }
 363   return strcasecmp(str, cmp); /* ASCII CP */
 364 }
 365
 366 /*********************************************************************
 367  *              _mbsncmp(MSVCRT.@)
 368  */
 369 int _mbsncmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 370 {
 371   if(!len)
 372     return 0;
 373
 374   if(MSVCRT___mb_cur_max > 1)
 375   {
 376     unsigned int strc, cmpc;
 377     while(len--)
 378     {
 379       int inc;
 380       if(!*str)
 381         return *cmp ? -1 : 0;
 382       if(!*cmp)
 383         return 1;
 384       strc = _mbsnextc(str);
 385       cmpc = _mbsnextc(cmp);
 386       if(strc != cmpc)
 387         return strc < cmpc ? -1 : 1;
 388       inc=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 389       str += inc;
 390       cmp += inc;
 391     }
 392     return 0; /* Matched len chars */
 393   }
 394   return strncmp(str, cmp, len); /* ASCII CP */
 395 }
 396
 397 /*********************************************************************
 398  *              _mbsnbcmp(MSVCRT.@)
 399  */
 400 int _mbsnbcmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 401 {
 402   if (!len)
 403     return 0;
 404   if(MSVCRT___mb_cur_max > 1)
 405   {
 406     unsigned int strc, cmpc;
 407     while (len)
 408     {
 409       int clen;
 410       if(!*str)
 411         return *cmp ? -1 : 0;
 412       if(!*cmp)
 413         return 1;
 414       if (MSVCRT_isleadbyte(*str))
 415       {
 416         strc=(len>=2)?_mbsnextc(str):0;
 417         clen=2;
 418       }
 419       else
 420       {
 421         strc=*str;
 422         clen=1;
 423       }
 424       if (MSVCRT_isleadbyte(*cmp))
 425         cmpc=(len>=2)?_mbsnextc(cmp):0;
 426       else
 427         cmpc=*str;
 428       if(strc != cmpc)
 429         return strc < cmpc ? -1 : 1;
 430       len -= clen;
 431       str += clen;
 432       cmp += clen;
 433     }
 434     return 0; /* Matched len chars */
 435   }
 436   return strncmp(str,cmp,len);
 437 }
 438
 439 /*********************************************************************
 440  *              _mbsnicmp(MSVCRT.@)
 441  *
 442  * Compare two multibyte strings case insensitively to 'len' characters.
 443  */
 444 int _mbsnicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 445 {
 446   /* FIXME: No tolower() for mb strings yet */
 447   if(MSVCRT___mb_cur_max > 1)
 448   {
 449     unsigned int strc, cmpc;
 450     while(len--)
 451     {
 452       if(!*str)
 453         return *cmp ? -1 : 0;
 454       if(!*cmp)
 455         return 1;
 456       strc = _mbctolower(_mbsnextc(str));
 457       cmpc = _mbctolower(_mbsnextc(cmp));
 458       if(strc != cmpc)
 459         return strc < cmpc ? -1 : 1;
 460       str +=(strc > 255) ? 2 : 1;
 461       cmp +=(strc > 255) ? 2 : 1; /* Equal, use same increment */
 462     }
 463     return 0; /* Matched len chars */
 464   }
 465   return strncasecmp(str, cmp, len); /* ASCII CP */
 466 }
 467
 468 /*********************************************************************
 469  *              _mbsnbicmp(MSVCRT.@)
 470  */
 471 int _mbsnbicmp(const unsigned char* str, const unsigned char* cmp, MSVCRT_size_t len)
 472 {
 473   if (!len)
 474     return 0;
 475   if(MSVCRT___mb_cur_max > 1)
 476   {
 477     unsigned int strc, cmpc;
 478     while (len)
 479     {
 480       int clen;
 481       if(!*str)
 482         return *cmp ? -1 : 0;
 483       if(!*cmp)
 484         return 1;
 485       if (MSVCRT_isleadbyte(*str))
 486       {
 487         strc=(len>=2)?_mbsnextc(str):0;
 488         clen=2;
 489       }
 490       else
 491       {
 492         strc=*str;
 493         clen=1;
 494       }
 495       if (MSVCRT_isleadbyte(*cmp))
 496         cmpc=(len>=2)?_mbsnextc(cmp):0;
 497       else
 498         cmpc=*str;
 499       strc = _mbctolower(strc);
 500       cmpc = _mbctolower(cmpc);
 501       if(strc != cmpc)
 502         return strc < cmpc ? -1 : 1;
 503       len -= clen;
 504       str += clen;
 505       cmp += clen;
 506     }
 507     return 0; /* Matched len bytes */
 508   }
 509   return strncmp(str,cmp,len);
 510 }
 511
 512 /*********************************************************************
 513  *              _mbschr(MSVCRT.@)
 514  *
 515  * Find a multibyte character in a multibyte string.
 516  */
 517 unsigned char* _mbschr(const unsigned char* s, unsigned int x)
 518 {
 519   if(MSVCRT___mb_cur_max > 1)
 520   {
 521     unsigned int c;
 522     while (1)
 523     {
 524       c = _mbsnextc(s);
 525       if (c == x)
 526         return (unsigned char*)s;
 527       if (!c)
 528         return NULL;
 529       s += c > 255 ? 2 : 1;
 530     }
 531   }
 532   return strchr(s, x); /* ASCII CP */
 533 }
 534
 535 /*********************************************************************
 536  *              _mbsrchr(MSVCRT.@)
 537  */
 538 unsigned char* _mbsrchr(const unsigned char* s, unsigned int x)
 539 {
 540   if(MSVCRT___mb_cur_max > 1)
 541   {
 542     unsigned int c;
 543     unsigned char* match=NULL;
 544     if(!s)
 545       return NULL;
 546     while (1) {
 547       c = _mbsnextc(s);
 548       if (c == x)
 549         match=(unsigned char*)s;
 550       if (!c)
 551         return match;
 552       s +=(c > 255) ? 2 : 1;
 553     }
 554   }
 555   return strrchr(s,x);
 556 }
 557
 558 /*********************************************************************
 559  *              _mbstok(MSVCRT.@)
 560  *
 561  * Find and extract tokens from strings
 562  */
 563 unsigned char* _mbstok(unsigned char *str, const unsigned char *delim)
 564 {
 565     thread_data_t *data = msvcrt_get_thread_data();
 566     char *ret;
 567
 568     if(MSVCRT___mb_cur_max > 1)
 569     {
 570         unsigned int c;
 571
 572         if (!str)
 573             if (!(str = data->mbstok_next)) return NULL;
 574
 575         while ((c = _mbsnextc(str)) && _mbschr(delim, c)) {
 576             str += c > 255 ? 2 : 1;
 577         }
 578         if (!*str) return NULL;
 579         ret = str++;
 580         while ((c = _mbsnextc(str)) && !_mbschr(delim, c)) {
 581             str += c > 255 ? 2 : 1;
 582         }
 583         if (*str) {
 584             *str++ = 0;
 585             if (c > 255) *str++ = 0;
 586         }
 587         data->mbstok_next = str;
 588         return ret;
 589     }
 590     return strtok(str, delim);  /* ASCII CP */
 591 }
 592
 593 /*********************************************************************
 594  *              mbtowc(MSVCRT.@)
 595  */
 596 int MSVCRT_mbtowc(MSVCRT_wchar_t *dst, const char* str, MSVCRT_size_t n)
 597 {
 598     /* temp var needed because MultiByteToWideChar wants non NULL destination */
 599     MSVCRT_wchar_t tmpdst = '\0';
 600
 601     if(n <= 0 || !str)
 602         return 0;
 603     if(!MultiByteToWideChar(CP_ACP, 0, str, n, &tmpdst, 1))
 604         return -1;
 605     if(dst)
 606         *dst = tmpdst;
 607     /* return the number of bytes from src that have been used */
 608     if(!*str)
 609         return 0;
 610     if(n >= 2 && MSVCRT_isleadbyte(*str) && str[1])
 611         return 2;
 612     return 1;
 613 }
 614
 615 /*********************************************************************
 616  *              _mbbtombc(MSVCRT.@)
 617  */
 618 unsigned int _mbbtombc(unsigned int c)
 619 {
 620   if(MSVCRT___mb_cur_max > 1 &&
 621      ((c >= 0x20 && c <=0x7e) ||(c >= 0xa1 && c <= 0xdf)))
 622   {
 623     /* FIXME: I can't get this function to return anything
 624      * different from what I pass it...
 625      */
 626   }
 627   return c;  /* ASCII CP or no MB char */
 628 }
 629
 630 /*********************************************************************
 631  *              _ismbbkana(MSVCRT.@)
 632  */
 633 int _ismbbkana(unsigned int c)
 634 {
 635   /* FIXME: use lc_ctype when supported, not lc_all */
 636   if(msvcrt_current_lc_all_cp == 932)
 637   {
 638     /* Japanese/Katakana, CP 932 */
 639     return (c >= 0xa1 && c <= 0xdf);
 640   }
 641   return 0;
 642 }
 643
 644 /*********************************************************************
 645  *              _ismbcdigit(MSVCRT.@)
 646  */
 647 int _ismbcdigit(unsigned int ch)
 648 {
 649     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 650     return (get_char_typeW( wch ) & C1_DIGIT);
 651 }
 652
 653 /*********************************************************************
 654  *              _ismbcgraph(MSVCRT.@)
 655  */
 656 int _ismbcgraph(unsigned int ch)
 657 {
 658     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 659     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA));
 660 }
 661
 662 /*********************************************************************
 663  *              _ismbcalpha (MSVCRT.@)
 664  */
 665 int _ismbcalpha(unsigned int ch)
 666 {
 667     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 668     return (get_char_typeW( wch ) & C1_ALPHA);
 669 }
 670
 671 /*********************************************************************
 672  *              _ismbclower (MSVCRT.@)
 673  */
 674 int _ismbclower(unsigned int ch)
 675 {
 676     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 677     return (get_char_typeW( wch ) & C1_UPPER);
 678 }
 679
 680 /*********************************************************************
 681  *              _ismbcupper (MSVCRT.@)
 682  */
 683 int _ismbcupper(unsigned int ch)
 684 {
 685     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 686     return (get_char_typeW( wch ) & C1_LOWER);
 687 }
 688
 689 /*********************************************************************
 690  *              _ismbcsymbol(MSVCRT.@)
 691  */
 692 int _ismbcsymbol(unsigned int ch)
 693 {
 694     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 695     WORD ctype;
 696     if (!GetStringTypeW(CT_CTYPE3, &wch, 1, &ctype))
 697     {
 698         WARN("GetStringTypeW failed on %x\n", ch);
 699         return 0;
 700     }
 701     return ((ctype & C3_SYMBOL) != 0);
 702 }
 703
 704 /*********************************************************************
 705  *              _ismbcalnum (MSVCRT.@)
 706  */
 707 int _ismbcalnum(unsigned int ch)
 708 {
 709     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 710     return (get_char_typeW( wch ) & (C1_ALPHA | C1_DIGIT));
 711 }
 712
 713 /*********************************************************************
 714  *              _ismbcspace (MSVCRT.@)
 715  */
 716 int _ismbcspace(unsigned int ch)
 717 {
 718     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 719     return (get_char_typeW( wch ) & C1_SPACE);
 720 }
 721
 722 /*********************************************************************
 723  *              _ismbcprint (MSVCRT.@)
 724  */
 725 int _ismbcprint(unsigned int ch)
 726 {
 727     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 728     return (get_char_typeW( wch ) & (C1_UPPER | C1_LOWER | C1_DIGIT | C1_PUNCT | C1_ALPHA | C1_SPACE));
 729 }
 730
 731 /*********************************************************************
 732  *              _ismbcpunct(MSVCRT.@)
 733  */
 734 int _ismbcpunct(unsigned int ch)
 735 {
 736     MSVCRT_wchar_t wch = msvcrt_mbc_to_wc( ch );
 737     return (get_char_typeW( wch ) & C1_PUNCT);
 738 }
 739
 740 /*********************************************************************
 741  *              _ismbchira(MSVCRT.@)
 742  */
 743 int _ismbchira(unsigned int c)
 744 {
 745   /* FIXME: use lc_ctype when supported, not lc_all */
 746   if(msvcrt_current_lc_all_cp == 932)
 747   {
 748     /* Japanese/Hiragana, CP 932 */
 749     return (c >= 0x829f && c <= 0x82f1);
 750   }
 751   return 0;
 752 }
 753
 754 /*********************************************************************
 755  *              _ismbckata(MSVCRT.@)
 756  */
 757 int _ismbckata(unsigned int c)
 758 {
 759   /* FIXME: use lc_ctype when supported, not lc_all */
 760   if(msvcrt_current_lc_all_cp == 932)
 761   {
 762     if(c < 256)
 763       return _ismbbkana(c);
 764     /* Japanese/Katakana, CP 932 */
 765     return (c >= 0x8340 && c <= 0x8396 && c != 0x837f);
 766   }
 767   return 0;
 768 }
 769
 770 /*********************************************************************
 771  *              _ismbblead(MSVCRT.@)
 772  */
 773 int _ismbblead(unsigned int c)
 774 {
 775   /* FIXME: should reference MSVCRT_mbctype */
 776   return MSVCRT___mb_cur_max > 1 && MSVCRT_isleadbyte(c);
 777 }
 778
 779
 780 /*********************************************************************
 781  *              _ismbbtrail(MSVCRT.@)
 782  */
 783 int _ismbbtrail(unsigned int c)
 784 {
 785   /* FIXME: should reference MSVCRT_mbctype */
 786   return !_ismbblead(c);
 787 }
 788
 789 /*********************************************************************
 790  *              _ismbslead(MSVCRT.@)
 791  */
 792 int _ismbslead(const unsigned char* start, const unsigned char* str)
 793 {
 794   /* Lead bytes can also be trail bytes if caller messed up
 795    * iterating through the string...
 796    */
 797   if(MSVCRT___mb_cur_max > 1)
 798   {
 799     while(start < str)
 800       start += MSVCRT_isleadbyte(*str) ? 2 : 1;
 801
 802     if(start == str)
 803       return MSVCRT_isleadbyte(*str);
 804   }
 805   return 0; /* Must have been a trail, we skipped it */
 806 }
 807
 808 /*********************************************************************
 809  *              _ismbstrail(MSVCRT.@)
 810  */
 811 int _ismbstrail(const unsigned char* start, const unsigned char* str)
 812 {
 813   /* Must not be a lead, and must be preceded by one */
 814   return !_ismbslead(start, str) && MSVCRT_isleadbyte(str[-1]);
 815 }
 816
 817 /*********************************************************************
 818  *              _mbsset(MSVCRT.@)
 819  */
 820 unsigned char* _mbsset(unsigned char* str, unsigned int c)
 821 {
 822   unsigned char* ret = str;
 823
 824   if(MSVCRT___mb_cur_max == 1 || c < 256)
 825     return _strset(str, c); /* ASCII CP or SB char */
 826
 827   c &= 0xffff; /* Strip high bits */
 828
 829   while(str[0] && str[1])
 830   {
 831     *str++ = c >> 8;
 832     *str++ = c & 0xff;
 833   }
 834   if(str[0])
 835     str[0] = '\0'; /* FIXME: OK to shorten? */
 836
 837   return ret;
 838 }
 839
 840 /*********************************************************************
 841  *              _mbsnbset(MSVCRT.@)
 842  */
 843 unsigned char* _mbsnbset(unsigned char *str, unsigned int c, MSVCRT_size_t len)
 844 {
 845     unsigned char *ret = str;
 846
 847     if(!len)
 848         return ret;
 849
 850     if(MSVCRT___mb_cur_max == 1 || c < 256)
 851         return _strnset(str, c, len); /* ASCII CP or SB char */
 852
 853     c &= 0xffff; /* Strip high bits */
 854
 855     while(str[0] && str[1] && (len > 1))
 856     {
 857         *str++ = c >> 8;
 858         len--;
 859         *str++ = c & 0xff;
 860         len--;
 861     }
 862     if(len && str[0]) {
 863         /* as per msdn pad with a blank character */
 864         str[0] = ' ';
 865     }
 866
 867     return ret;
 868 }
 869
 870 /*********************************************************************
 871  *              _mbsnset(MSVCRT.@)
 872  */
 873 unsigned char* _mbsnset(unsigned char* str, unsigned int c, MSVCRT_size_t len)
 874 {
 875   unsigned char *ret = str;
 876
 877   if(!len)
 878     return ret;
 879
 880   if(MSVCRT___mb_cur_max == 1 || c < 256)
 881     return _strnset(str, c, len); /* ASCII CP or SB char */
 882
 883   c &= 0xffff; /* Strip high bits */
 884
 885   while(str[0] && str[1] && len--)
 886   {
 887     *str++ = c >> 8;
 888     *str++ = c & 0xff;
 889   }
 890   if(len && str[0])
 891     str[0] = '\0'; /* FIXME: OK to shorten? */
 892
 893   return ret;
 894 }
 895
 896 /*********************************************************************
 897  *              _mbsnccnt(MSVCRT.@)
 898  * 'c' is for 'character'.
 899  */
 900 MSVCRT_size_t _mbsnccnt(const unsigned char* str, MSVCRT_size_t len)
 901 {
 902   MSVCRT_size_t ret;
 903   if(MSVCRT___mb_cur_max > 1)
 904   {
 905     ret=0;
 906     while(*str && len-- > 0)
 907     {
 908       if(MSVCRT_isleadbyte(*str))
 909       {
 910         if (!len)
 911           break;
 912         len--;
 913         str++;
 914       }
 915       str++;
 916       ret++;
 917     }
 918     return ret;
 919   }
 920   ret=strlen(str);
 921   return min(ret, len); /* ASCII CP */
 922 }
 923
 924 /*********************************************************************
 925  *              _mbsnbcnt(MSVCRT.@)
 926  * 'b' is for byte count.
 927  */
 928 MSVCRT_size_t _mbsnbcnt(const unsigned char* str, MSVCRT_size_t len)
 929 {
 930   MSVCRT_size_t ret;
 931   if(MSVCRT___mb_cur_max > 1)
 932   {
 933     const unsigned char* xstr = str;
 934     while(*xstr && len-- > 0)
 935     {
 936       if (MSVCRT_isleadbyte(*xstr++))
 937         xstr++;
 938     }
 939     return xstr-str;
 940   }
 941   ret=strlen(str);
 942   return min(ret, len); /* ASCII CP */
 943 }
 944
 945
 946 /*********************************************************************
 947  *              _mbsnbcat(MSVCRT.@)
 948  */
 949 unsigned char* _mbsnbcat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
 950 {
 951     if(MSVCRT___mb_cur_max > 1)
 952     {
 953         char *res = dst;
 954         while (*dst) {
 955             if (MSVCRT_isleadbyte(*dst++)) {
 956                 if (*dst) {
 957                     dst++;
 958                 } else {
 959                     /* as per msdn overwrite the lead byte in front of '\0' */
 960                     dst--;
 961                     break;
 962                 }
 963             }
 964         }
 965         while (*src && len--) *dst++ = *src++;
 966         *dst = '\0';
 967         return res;
 968     }
 969     return strncat(dst, src, len); /* ASCII CP */
 970 }
 971
 972
 973 /*********************************************************************
 974  *              _mbsncat(MSVCRT.@)
 975  */
 976 unsigned char* _mbsncat(unsigned char* dst, const unsigned char* src, MSVCRT_size_t len)
 977 {
 978   if(MSVCRT___mb_cur_max > 1)
 979   {
 980     char *res = dst;
 981     while (*dst)
 982     {
 983       if (MSVCRT_isleadbyte(*dst++))
 984         dst++;
 985     }
 986     while (*src && len--)
 987     {
 988       *dst++ = *src;
 989       if(MSVCRT_isleadbyte(*src++))
 990         *dst++ = *src++;
 991     }
 992     *dst = '\0';
 993     return res;
 994   }
 995   return strncat(dst, src, len); /* ASCII CP */
 996 }
 997
 998
 999 /*********************************************************************
1000  *              _mbslwr(MSVCRT.@)
1001  */
1002 unsigned char* _mbslwr(unsigned char* s)
1003 {
1004   if (!s)
1005     return NULL;
1006   if (MSVCRT___mb_cur_max > 1)
1007   {
1008     unsigned int c;
1009     unsigned char* p=s;
1010     while (*s)
1011     {
1012       c = _mbctolower(_mbsnextc(s));
1013       /* Note that I assume that the size of the character is unchanged */
1014       if (c > 255)
1015       {
1016           *s++=(c>>8);
1017           c=c & 0xff;
1018       }
1019       *s++=c;
1020     }
1021     return p;
1022   }
1023   return _strlwr(s);
1024 }
1025
1026
1027 /*********************************************************************
1028  *              _mbsupr(MSVCRT.@)
1029  */
1030 unsigned char* _mbsupr(unsigned char* s)
1031 {
1032   if (!s)
1033     return NULL;
1034   if (MSVCRT___mb_cur_max > 1)
1035   {
1036     unsigned int c;
1037     unsigned char* p=s;
1038     while (*s)
1039     {
1040       c = _mbctoupper(_mbsnextc(s));
1041       /* Note that I assume that the size of the character is unchanged */
1042       if (c > 255)
1043       {
1044           *s++=(c>>8);
1045           c=c & 0xff;
1046       }
1047       *s++=c;
1048     }
1049     return p;
1050   }
1051   return _strupr(s);
1052 }
1053
1054
1055 /*********************************************************************
1056  *              _mbsspn (MSVCRT.@)
1057  */
1058 MSVCRT_size_t _mbsspn(const unsigned char* string, const unsigned char* set)
1059 {
1060   const unsigned char *p, *q;
1061
1062   for (p = string; *p; p++)
1063     {
1064       if (MSVCRT_isleadbyte(*p))
1065         {
1066           for (q = set; *q; q++)
1067             {
1068               if (!q[1])
1069                 break;
1070               if ((*p == *q) &&  (p[1] == q[1]))
1071                 break;
1072               q++;
1073             }
1074           if (*++p == '\0')
1075             break;
1076         }
1077       else
1078         for (q = set; *q; q++)
1079           if (*p == *q)
1080             break;
1081     }
1082   return p - string;
1083 }
1084
1085 /*********************************************************************
1086  *              _mbscspn(MSVCRT.@)
1087  */
1088 MSVCRT_size_t _mbscspn(const unsigned char* str, const unsigned char* cmp)
1089 {
1090   if (MSVCRT___mb_cur_max > 1)
1091     FIXME("don't handle double character case\n");
1092   return strcspn(str, cmp);
1093 }
1094
1095 /*********************************************************************
1096  *              _mbsrev (MSVCRT.@)
1097  */
1098 unsigned char* _mbsrev(unsigned char* str)
1099 {
1100     int i, len = _mbslen(str);
1101     unsigned char *p, *temp=MSVCRT_malloc(len*2);
1102
1103     if(!temp)
1104         return str;
1105
1106     /* unpack multibyte string to temp buffer */
1107     p=str;
1108     for(i=0; i<len; i++)
1109     {
1110         if (MSVCRT_isleadbyte(*p))
1111         {
1112             temp[i*2]=*p++;
1113             temp[i*2+1]=*p++;
1114         }
1115         else
1116         {
1117             temp[i*2]=*p++;
1118             temp[i*2+1]=0;
1119         }
1120     }
1121
1122     /* repack it in the reverse order */
1123     p=str;
1124     for(i=len-1; i>=0; i--)
1125     {
1126         if(MSVCRT_isleadbyte(temp[i*2]))
1127         {
1128             *p++=temp[i*2];
1129             *p++=temp[i*2+1];
1130         }
1131         else
1132         {
1133             *p++=temp[i*2];
1134         }
1135     }
1136
1137     MSVCRT_free(temp);
1138
1139     return str;
1140 }
1141
1142 /*********************************************************************
1143  *              _mbspbrk (MSVCRT.@)
1144  */
1145 unsigned char* _mbspbrk(const unsigned char* str, const unsigned char* accept)
1146 {
1147     const unsigned char* p;
1148
1149     while(*str)
1150     {
1151         for(p = accept; *p; p += (MSVCRT_isleadbyte(*p)?2:1) )
1152         {
1153             if (*p == *str)
1154                 if( !MSVCRT_isleadbyte(*p) || ( *(p+1) == *(str+1) ) )
1155                      return (unsigned char*)str;
1156         }
1157         str += (MSVCRT_isleadbyte(*str)?2:1);
1158     }
1159     return NULL;
1160 }