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