memory/codepage.c

   1 /*
   2  * Code page functions
   3  *
   4  * Copyright 2000 Alexandre Julliard
   5  */
   6
   7 #include <assert.h>
   8 #include <stdio.h>
   9 #include <stdlib.h>
  10 #include <string.h>
  11
  12 #include "winbase.h"
  13 #include "winerror.h"
  14 #include "winnls.h"
  15 #include "wine/unicode.h"
  16 #include "debugtools.h"
  17
  18 DEFAULT_DEBUG_CHANNEL(string);
  19
  20 /* current code pages */
  21 static const union cptable *ansi_cptable;
  22 static const union cptable *oem_cptable;
  23 static const union cptable *mac_cptable;
  24
  25 /* retrieve a code page table from the locale info */
  26 static const union cptable *get_locale_cp( LCID lcid, LCTYPE type )
  27 {
  28     const union cptable *table = NULL;
  29     char buf[32];
  30
  31     if (GetLocaleInfoA( lcid, type, buf, sizeof(buf) )) table = cp_get_table( atoi(buf) );
  32     return table;
  33 }
  34
  35 /* setup default codepage info before we can get at the locale stuff */
  36 static void init_codepages(void)
  37 {
  38     ansi_cptable = cp_get_table( 1252 );
  39     oem_cptable  = cp_get_table( 437 );
  40     mac_cptable  = cp_get_table( 10000 );
  41     assert( ansi_cptable );
  42     assert( oem_cptable );
  43     assert( mac_cptable );
  44 }
  45
  46 /* find the table for a given codepage, handling CP_ACP etc. pseudo-codepages */
  47 static const union cptable *get_codepage_table( unsigned int codepage )
  48 {
  49     const union cptable *ret = NULL;
  50
  51     if (!ansi_cptable) init_codepages();
  52
  53     switch(codepage)
  54     {
  55     case CP_ACP:        return ansi_cptable;
  56     case CP_OEMCP:      return oem_cptable;
  57     case CP_MACCP:      return mac_cptable;
  58     case CP_THREAD_ACP: return get_locale_cp( GetThreadLocale(), LOCALE_IDEFAULTANSICODEPAGE );
  59     case CP_UTF7:
  60     case CP_UTF8:
  61         break;
  62     default:
  63         if (codepage == ansi_cptable->info.codepage) return ansi_cptable;
  64         if (codepage == oem_cptable->info.codepage) return oem_cptable;
  65         if (codepage == mac_cptable->info.codepage) return mac_cptable;
  66         ret = cp_get_table( codepage );
  67         break;
  68     }
  69     return ret;
  70 }
  71
  72 /* initialize default code pages from locale info */
  73 /* FIXME: should be done in init_codepages, but it can't right now */
  74 /* since it needs KERNEL32 to be loaded for the locale info. */
  75 void CODEPAGE_Init(void)
  76 {
  77     extern void __wine_init_codepages( const union cptable *ansi, const union cptable *oem );
  78     const union cptable *table;
  79     LCID lcid = GetUserDefaultLCID();
  80
  81     if (!ansi_cptable) init_codepages();  /* just in case */
  82
  83     if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTANSICODEPAGE ))) ansi_cptable = table;
  84     if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTMACCODEPAGE ))) mac_cptable = table;
  85     if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTCODEPAGE ))) oem_cptable = table;
  86     __wine_init_codepages( ansi_cptable, oem_cptable );
  87
  88     TRACE( "ansi=%03d oem=%03d mac=%03d\n", ansi_cptable->info.codepage,
  89            oem_cptable->info.codepage, mac_cptable->info.codepage );
  90 }
  91
  92 /******************************************************************************
  93  *              GetACP   (KERNEL32.@)
  94  *
  95  * RETURNS
  96  *    Current ANSI code-page identifier, default if no current defined
  97  */
  98 UINT WINAPI GetACP(void)
  99 {
 100     if (!ansi_cptable) init_codepages();
 101     return ansi_cptable->info.codepage;
 102 }
 103
 104
 105 /***********************************************************************
 106  *              GetOEMCP   (KERNEL32.@)
 107  */
 108 UINT WINAPI GetOEMCP(void)
 109 {
 110     if (!oem_cptable) init_codepages();
 111     return oem_cptable->info.codepage;
 112 }
 113
 114
 115 /***********************************************************************
 116  *           IsValidCodePage   (KERNEL32.@)
 117  */
 118 BOOL WINAPI IsValidCodePage( UINT codepage )
 119 {
 120     return cp_get_table( codepage ) != NULL;
 121 }
 122
 123
 124 /***********************************************************************
 125  *           IsDBCSLeadByteEx   (KERNEL32.@)
 126  */
 127 BOOL WINAPI IsDBCSLeadByteEx( UINT codepage, BYTE testchar )
 128 {
 129     const union cptable *table = get_codepage_table( codepage );
 130     return table && is_dbcs_leadbyte( table, testchar );
 131 }
 132
 133
 134 /***********************************************************************
 135  *           IsDBCSLeadByte   (KERNEL32.@)
 136  *           IsDBCSLeadByte   (KERNEL.207)
 137  */
 138 BOOL WINAPI IsDBCSLeadByte( BYTE testchar )
 139 {
 140     if (!ansi_cptable) init_codepages();
 141     return is_dbcs_leadbyte( ansi_cptable, testchar );
 142 }
 143
 144
 145 /***********************************************************************
 146  *           GetCPInfo   (KERNEL32.@)
 147  */
 148 BOOL WINAPI GetCPInfo( UINT codepage, LPCPINFO cpinfo )
 149 {
 150     const union cptable *table = get_codepage_table( codepage );
 151
 152     if (!table)
 153     {
 154         SetLastError( ERROR_INVALID_PARAMETER );
 155         return FALSE;
 156     }
 157     if (table->info.def_char & 0xff00)
 158     {
 159         cpinfo->DefaultChar[0] = table->info.def_char & 0xff00;
 160         cpinfo->DefaultChar[1] = table->info.def_char & 0x00ff;
 161     }
 162     else
 163     {
 164         cpinfo->DefaultChar[0] = table->info.def_char & 0xff;
 165         cpinfo->DefaultChar[1] = 0;
 166     }
 167     if ((cpinfo->MaxCharSize = table->info.char_size) == 2)
 168         memcpy( cpinfo->LeadByte, table->dbcs.lead_bytes, sizeof(cpinfo->LeadByte) );
 169     else
 170         cpinfo->LeadByte[0] = cpinfo->LeadByte[1] = 0;
 171
 172     return TRUE;
 173 }
 174
 175
 176 /***********************************************************************
 177  *              EnumSystemCodePagesA   (KERNEL32.@)
 178  */
 179 BOOL WINAPI EnumSystemCodePagesA( CODEPAGE_ENUMPROCA lpfnCodePageEnum, DWORD flags )
 180 {
 181     const union cptable *table;
 182     char buffer[10];
 183     int index = 0;
 184
 185     for (;;)
 186     {
 187         if (!(table = cp_enum_table( index++ ))) break;
 188         sprintf( buffer, "%d", table->info.codepage );
 189         if (!lpfnCodePageEnum( buffer )) break;
 190     }
 191     return TRUE;
 192 }
 193
 194
 195 /***********************************************************************
 196  *              EnumSystemCodePagesW   (KERNEL32.@)
 197  */
 198 BOOL WINAPI EnumSystemCodePagesW( CODEPAGE_ENUMPROCW lpfnCodePageEnum, DWORD flags )
 199 {
 200     const union cptable *table;
 201     WCHAR buffer[10], *p;
 202     int page, index = 0;
 203
 204     for (;;)
 205     {
 206         if (!(table = cp_enum_table( index++ ))) break;
 207         p = buffer + sizeof(buffer)/sizeof(WCHAR);
 208         *--p = 0;
 209         page = table->info.codepage;
 210         do
 211         {
 212             *--p = '0' + (page % 10);
 213             page /= 10;
 214         } while( page );
 215         if (!lpfnCodePageEnum( p )) break;
 216     }
 217     return TRUE;
 218 }
 219
 220
 221 /***********************************************************************
 222  *              MultiByteToWideChar   (KERNEL32.@)
 223  *
 224  * PARAMS
 225  *   page [in]    Codepage character set to convert from
 226  *   flags [in]   Character mapping flags
 227  *   src [in]     Source string buffer
 228  *   srclen [in]  Length of source string buffer
 229  *   dst [in]     Destination buffer
 230  *   dstlen [in]  Length of destination buffer
 231  *
 232  * NOTES
 233  *   The returned length includes the null terminator character.
 234  *
 235  * RETURNS
 236  *   Success: If dstlen > 0, number of characters written to destination
 237  *            buffer.  If dstlen == 0, number of characters needed to do
 238  *            conversion.
 239  *   Failure: 0. Occurs if not enough space is available.
 240  *
 241  * ERRORS
 242  *   ERROR_INSUFFICIENT_BUFFER
 243  *   ERROR_INVALID_PARAMETER
 244  *   ERROR_NO_UNICODE_TRANSLATION
 245  *
 246  */
 247 INT WINAPI MultiByteToWideChar( UINT page, DWORD flags, LPCSTR src, INT srclen,
 248                                 LPWSTR dst, INT dstlen )
 249 {
 250     const union cptable *table;
 251     int ret;
 252
 253     if (!src || (!dst && dstlen))
 254     {
 255         SetLastError( ERROR_INVALID_PARAMETER );
 256         return 0;
 257     }
 258
 259     if (srclen == -1) srclen = strlen(src) + 1;
 260
 261     if (flags & MB_USEGLYPHCHARS) FIXME("MB_USEGLYPHCHARS not supported\n");
 262
 263     switch(page)
 264     {
 265     case CP_UTF7:
 266         FIXME("UTF not supported\n");
 267         SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
 268         return 0;
 269     case CP_UTF8:
 270         ret = utf8_mbstowcs( flags, src, srclen, dst, dstlen );
 271         break;
 272     default:
 273         if (!(table = get_codepage_table( page )))
 274         {
 275             SetLastError( ERROR_INVALID_PARAMETER );
 276             return 0;
 277         }
 278         ret = cp_mbstowcs( table, flags, src, srclen, dst, dstlen );
 279         break;
 280     }
 281
 282     if (ret < 0)
 283     {
 284         switch(ret)
 285         {
 286         case -1: SetLastError( ERROR_INSUFFICIENT_BUFFER ); break;
 287         case -2: SetLastError( ERROR_NO_UNICODE_TRANSLATION ); break;
 288         }
 289         ret = 0;
 290     }
 291     return ret;
 292 }
 293
 294
 295 /***********************************************************************
 296  *              WideCharToMultiByte   (KERNEL32.@)
 297  *
 298  * PARAMS
 299  *   page [in]    Codepage character set to convert to
 300  *   flags [in]   Character mapping flags
 301  *   src [in]     Source string buffer
 302  *   srclen [in]  Length of source string buffer
 303  *   dst [in]     Destination buffer
 304  *   dstlen [in]  Length of destination buffer
 305  *   defchar [in] Default character to use for conversion if no exact
 306  *                  conversion can be made
 307  *   used [out]   Set if default character was used in the conversion
 308  *
 309  * NOTES
 310  *   The returned length includes the null terminator character.
 311  *
 312  * RETURNS
 313  *   Success: If dstlen > 0, number of characters written to destination
 314  *            buffer.  If dstlen == 0, number of characters needed to do
 315  *            conversion.
 316  *   Failure: 0. Occurs if not enough space is available.
 317  *
 318  * ERRORS
 319  *   ERROR_INSUFFICIENT_BUFFER
 320  *   ERROR_INVALID_PARAMETER
 321  */
 322 INT WINAPI WideCharToMultiByte( UINT page, DWORD flags, LPCWSTR src, INT srclen,
 323                                 LPSTR dst, INT dstlen, LPCSTR defchar, BOOL *used )
 324 {
 325     const union cptable *table;
 326     int ret, used_tmp;
 327
 328     if (!src || (!dst && dstlen))
 329     {
 330         SetLastError( ERROR_INVALID_PARAMETER );
 331         return 0;
 332     }
 333
 334     if (srclen == -1) srclen = strlenW(src) + 1;
 335
 336     switch(page)
 337     {
 338     case CP_UTF7:
 339         FIXME("UTF-7 not supported\n");
 340         SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
 341         return 0;
 342     case CP_UTF8:
 343         ret = utf8_wcstombs( src, srclen, dst, dstlen );
 344         break;
 345     default:
 346         if (!(table = get_codepage_table( page )))
 347         {
 348             SetLastError( ERROR_INVALID_PARAMETER );
 349             return 0;
 350         }
 351         ret = cp_wcstombs( table, flags, src, srclen, dst, dstlen,
 352                            defchar, used ? &used_tmp : NULL );
 353         if (used) *used = used_tmp;
 354         break;
 355     }
 356
 357     if (ret == -1)
 358     {
 359         SetLastError( ERROR_INSUFFICIENT_BUFFER );
 360         ret = 0;
 361     }
 362     return ret;
 363 }
 364
 365
 366 /******************************************************************************
 367  *              GetStringTypeW   (KERNEL32.@)
 368  *
 369  */
 370 BOOL WINAPI GetStringTypeW( DWORD type, LPCWSTR src, INT count, LPWORD chartype )
 371 {
 372     if (count == -1) count = strlenW(src) + 1;
 373     switch(type)
 374     {
 375     case CT_CTYPE1:
 376         while (count--) *chartype++ = get_char_typeW( *src++ ) & 0xfff;
 377         break;
 378     case CT_CTYPE2:
 379         while (count--) *chartype++ = get_char_typeW( *src++ ) >> 12;
 380         break;
 381     case CT_CTYPE3:
 382         FIXME("CT_CTYPE3 not supported.\n");
 383     default:
 384         SetLastError( ERROR_INVALID_PARAMETER );
 385         return FALSE;
 386     }
 387     return TRUE;
 388 }
 389
 390
 391 /******************************************************************************
 392  *              GetStringTypeExW   (KERNEL32.@)
 393  */
 394 BOOL WINAPI GetStringTypeExW( LCID locale, DWORD type, LPCWSTR src, INT count, LPWORD chartype )
 395 {
 396     /* locale is ignored for Unicode */
 397     return GetStringTypeW( type, src, count, chartype );
 398 }
 399
 400 /******************************************************************************
 401  *              GetStringTypeA  [KERNEL32.@]
 402  */
 403 BOOL WINAPI GetStringTypeA(LCID locale, DWORD type, LPCSTR src, INT count, LPWORD chartype)
 404 {
 405     char buf[20];
 406     UINT cp;
 407     INT countW;
 408     LPWSTR srcW;
 409     BOOL ret = FALSE;
 410
 411     if(count == -1) count = strlen(src) + 1;
 412
 413     if(!GetLocaleInfoA(locale, LOCALE_IDEFAULTANSICODEPAGE | LOCALE_NOUSEROVERRIDE,
 414                    buf, sizeof(buf)))
 415     {
 416         FIXME("For locale %04lx using current ANSI code page\n", locale);
 417         cp = GetACP();
 418     }
 419     else
 420         cp = atoi(buf);
 421
 422     countW = MultiByteToWideChar(cp, 0, src, count, NULL, 0);
 423     if((srcW = HeapAlloc(GetProcessHeap(), 0, countW * sizeof(WCHAR))))
 424     {
 425         MultiByteToWideChar(cp, 0, src, count, srcW, countW);
 426         ret = GetStringTypeW(type, srcW, count, chartype);
 427         HeapFree(GetProcessHeap(), 0, srcW);
 428     }
 429     return ret;
 430 }
 431
 432 /******************************************************************************
 433  *              GetStringTypeExA        [KERNEL32.@]
 434  */
 435 BOOL WINAPI GetStringTypeExA(LCID locale, DWORD type, LPCSTR src, INT count, LPWORD chartype)
 436 {
 437     return GetStringTypeA(locale, type, src, count, chartype);
 438 }