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     const union cptable *table;
  78     LCID lcid = GetUserDefaultLCID();
  79
  80     if (!ansi_cptable) init_codepages();  /* just in case */
  81
  82     if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTANSICODEPAGE ))) ansi_cptable = table;
  83     if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTMACCODEPAGE ))) mac_cptable = table;
  84     if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTCODEPAGE ))) oem_cptable = table;
  85
  86     TRACE( "ansi=%03d oem=%03d mac=%03d\n", ansi_cptable->info.codepage,
  87            oem_cptable->info.codepage, mac_cptable->info.codepage );
  88 }
  89
  90 /******************************************************************************
  91  *              GetACP   (KERNEL32)
  92  *
  93  * RETURNS
  94  *    Current ANSI code-page identifier, default if no current defined
  95  */
  96 UINT WINAPI GetACP(void)
  97 {
  98     if (!ansi_cptable) init_codepages();
  99     return ansi_cptable->info.codepage;
 100 }
 101
 102
 103 /***********************************************************************
 104  *              GetOEMCP   (KERNEL32)
 105  */
 106 UINT WINAPI GetOEMCP(void)
 107 {
 108     if (!oem_cptable) init_codepages();
 109     return oem_cptable->info.codepage;
 110 }
 111
 112
 113 /***********************************************************************
 114  *           IsValidCodePage   (KERNEL32)
 115  */
 116 BOOL WINAPI IsValidCodePage( UINT codepage )
 117 {
 118     return cp_get_table( codepage ) != NULL;
 119 }
 120
 121
 122 /***********************************************************************
 123  *           IsDBCSLeadByteEx   (KERNEL32)
 124  */
 125 BOOL WINAPI IsDBCSLeadByteEx( UINT codepage, BYTE testchar )
 126 {
 127     const union cptable *table = get_codepage_table( codepage );
 128     return table && is_dbcs_leadbyte( table, testchar );
 129 }
 130
 131
 132 /***********************************************************************
 133  *           IsDBCSLeadByte   (KERNEL32)
 134  */
 135 BOOL WINAPI IsDBCSLeadByte( BYTE testchar )
 136 {
 137     if (!ansi_cptable) init_codepages();
 138     return is_dbcs_leadbyte( ansi_cptable, testchar );
 139 }
 140
 141
 142 /***********************************************************************
 143  *           GetCPInfo   (KERNEL32)
 144  */
 145 BOOL WINAPI GetCPInfo( UINT codepage, LPCPINFO cpinfo )
 146 {
 147     const union cptable *table = get_codepage_table( codepage );
 148
 149     if (!table)
 150     {
 151         SetLastError( ERROR_INVALID_PARAMETER );
 152         return FALSE;
 153     }
 154     if (table->info.def_char & 0xff00)
 155     {
 156         cpinfo->DefaultChar[0] = table->info.def_char & 0xff00;
 157         cpinfo->DefaultChar[1] = table->info.def_char & 0x00ff;
 158     }
 159     else
 160     {
 161         cpinfo->DefaultChar[0] = table->info.def_char & 0xff;
 162         cpinfo->DefaultChar[1] = 0;
 163     }
 164     if ((cpinfo->MaxCharSize = table->info.char_size) == 2)
 165         memcpy( cpinfo->LeadByte, table->dbcs.lead_bytes, sizeof(cpinfo->LeadByte) );
 166     else
 167         cpinfo->LeadByte[0] = cpinfo->LeadByte[1] = 0;
 168
 169     return TRUE;
 170 }
 171
 172
 173 /***********************************************************************
 174  *              EnumSystemCodePagesA   (KERNEL32)
 175  */
 176 BOOL WINAPI EnumSystemCodePagesA( CODEPAGE_ENUMPROCA lpfnCodePageEnum, DWORD flags )
 177 {
 178     const union cptable *table;
 179     char buffer[10];
 180     int index = 0;
 181
 182     for (;;)
 183     {
 184         if (!(table = cp_enum_table( index++ ))) break;
 185         sprintf( buffer, "%d", table->info.codepage );
 186         if (!lpfnCodePageEnum( buffer )) break;
 187     }
 188     return TRUE;
 189 }
 190
 191
 192 /***********************************************************************
 193  *              EnumSystemCodePagesW   (KERNEL32)
 194  */
 195 BOOL WINAPI EnumSystemCodePagesW( CODEPAGE_ENUMPROCW lpfnCodePageEnum, DWORD flags )
 196 {
 197     const union cptable *table;
 198     WCHAR buffer[10], *p;
 199     int page, index = 0;
 200
 201     for (;;)
 202     {
 203         if (!(table = cp_enum_table( index++ ))) break;
 204         p = buffer + sizeof(buffer)/sizeof(WCHAR);
 205         *--p = 0;
 206         page = table->info.codepage;
 207         do
 208         {
 209             *--p = '0' + (page % 10);
 210             page /= 10;
 211         } while( page );
 212         if (!lpfnCodePageEnum( p )) break;
 213     }
 214     return TRUE;
 215 }
 216
 217
 218 /***********************************************************************
 219  *              MultiByteToWideChar   (KERNEL32)
 220  *
 221  * PARAMS
 222  *   page [in]    Codepage character set to convert from
 223  *   flags [in]   Character mapping flags
 224  *   src [in]     Source string buffer
 225  *   srclen [in]  Length of source string buffer
 226  *   dst [in]     Destination buffer
 227  *   dstlen [in]  Length of destination buffer
 228  *
 229  * NOTES
 230  *   The returned length includes the null terminator character.
 231  *
 232  * RETURNS
 233  *   Success: If dstlen > 0, number of characters written to destination
 234  *            buffer.  If dstlen == 0, number of characters needed to do
 235  *            conversion.
 236  *   Failure: 0. Occurs if not enough space is available.
 237  *
 238  * ERRORS
 239  *   ERROR_INSUFFICIENT_BUFFER
 240  *   ERROR_INVALID_PARAMETER
 241  *   ERROR_NO_UNICODE_TRANSLATION
 242  *
 243  */
 244 INT WINAPI MultiByteToWideChar( UINT page, DWORD flags, LPCSTR src, INT srclen,
 245                                 LPWSTR dst, INT dstlen )
 246 {
 247     const union cptable *table;
 248     int ret;
 249
 250     if (!src || (!dst && dstlen))
 251     {
 252         SetLastError( ERROR_INVALID_PARAMETER );
 253         return 0;
 254     }
 255
 256     if (srclen == -1) srclen = strlen(src) + 1;
 257
 258     if (flags & MB_USEGLYPHCHARS) FIXME("MB_USEGLYPHCHARS not supported\n");
 259
 260     switch(page)
 261     {
 262     case CP_UTF7:
 263         FIXME("UTF not supported\n");
 264         SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
 265         return 0;
 266     case CP_UTF8:
 267         ret = utf8_mbstowcs( flags, src, srclen, dst, dstlen );
 268         break;
 269     default:
 270         if (!(table = get_codepage_table( page )))
 271         {
 272             SetLastError( ERROR_INVALID_PARAMETER );
 273             return 0;
 274         }
 275         ret = cp_mbstowcs( table, flags, src, srclen, dst, dstlen );
 276         break;
 277     }
 278
 279     if (ret < 0)
 280     {
 281         switch(ret)
 282         {
 283         case -1: SetLastError( ERROR_INSUFFICIENT_BUFFER ); break;
 284         case -2: SetLastError( ERROR_NO_UNICODE_TRANSLATION ); break;
 285         }
 286         ret = 0;
 287     }
 288     return ret;
 289 }
 290
 291
 292 /***********************************************************************
 293  *              WideCharToMultiByte   (KERNEL32)
 294  *
 295  * PARAMS
 296  *   page [in]    Codepage character set to convert to
 297  *   flags [in]   Character mapping flags
 298  *   src [in]     Source string buffer
 299  *   srclen [in]  Length of source string buffer
 300  *   dst [in]     Destination buffer
 301  *   dstlen [in]  Length of destination buffer
 302  *   defchar [in] Default character to use for conversion if no exact
 303  *                  conversion can be made
 304  *   used [out]   Set if default character was used in the conversion
 305  *
 306  * NOTES
 307  *   The returned length includes the null terminator character.
 308  *
 309  * RETURNS
 310  *   Success: If dstlen > 0, number of characters written to destination
 311  *            buffer.  If dstlen == 0, number of characters needed to do
 312  *            conversion.
 313  *   Failure: 0. Occurs if not enough space is available.
 314  *
 315  * ERRORS
 316  *   ERROR_INSUFFICIENT_BUFFER
 317  *   ERROR_INVALID_PARAMETER
 318  */
 319 INT WINAPI WideCharToMultiByte( UINT page, DWORD flags, LPCWSTR src, INT srclen,
 320                                 LPSTR dst, INT dstlen, LPCSTR defchar, BOOL *used )
 321 {
 322     const union cptable *table;
 323     int ret, used_tmp;
 324
 325     if (!src || (!dst && dstlen))
 326     {
 327         SetLastError( ERROR_INVALID_PARAMETER );
 328         return 0;
 329     }
 330
 331     if (srclen == -1) srclen = strlenW(src) + 1;
 332
 333     switch(page)
 334     {
 335     case CP_UTF7:
 336         FIXME("UTF-7 not supported\n");
 337         SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
 338         return 0;
 339     case CP_UTF8:
 340         ret = utf8_wcstombs( src, srclen, dst, dstlen );
 341         break;
 342     default:
 343         if (!(table = get_codepage_table( page )))
 344         {
 345             SetLastError( ERROR_INVALID_PARAMETER );
 346             return 0;
 347         }
 348         ret = cp_wcstombs( table, flags, src, srclen, dst, dstlen,
 349                            defchar, used ? &used_tmp : NULL );
 350         if (used) *used = used_tmp;
 351         break;
 352     }
 353
 354     if (ret == -1)
 355     {
 356         SetLastError( ERROR_INSUFFICIENT_BUFFER );
 357         ret = 0;
 358     }
 359     return ret;
 360 }
 361
 362
 363 /******************************************************************************
 364  *              GetStringTypeW   (KERNEL32)
 365  *
 366  */
 367 BOOL WINAPI GetStringTypeW( DWORD type, LPCWSTR src, INT count, LPWORD chartype )
 368 {
 369     if (count == -1) count = strlenW(src) + 1;
 370     switch(type)
 371     {
 372     case CT_CTYPE1:
 373         while (count--) *chartype++ = get_char_typeW( *src++ ) & 0xfff;
 374         break;
 375     case CT_CTYPE2:
 376         while (count--) *chartype++ = get_char_typeW( *src++ ) >> 12;
 377         break;
 378     case CT_CTYPE3:
 379         FIXME("CT_CTYPE3 not supported.\n");
 380     default:
 381         SetLastError( ERROR_INVALID_PARAMETER );
 382         return FALSE;
 383     }
 384     return TRUE;
 385 }
 386
 387
 388 /******************************************************************************
 389  *              GetStringTypeExW   (KERNEL32)
 390  */
 391 BOOL WINAPI GetStringTypeExW( LCID locale, DWORD type, LPCWSTR src, INT count, LPWORD chartype )
 392 {
 393     /* locale is ignored for Unicode */
 394     return GetStringTypeW( type, src, count, chartype );
 395 }