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