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  *           IsDBCSLeadByte   (KERNEL.207)
 135  */
 136 BOOL WINAPI IsDBCSLeadByte( BYTE testchar )
 137 {
 138     if (!ansi_cptable) init_codepages();
 139     return is_dbcs_leadbyte( ansi_cptable, testchar );
 140 }
 141
 142
 143 /***********************************************************************
 144  *           GetCPInfo   (KERNEL32.@)
 145  */
 146 BOOL WINAPI GetCPInfo( UINT codepage, LPCPINFO cpinfo )
 147 {
 148     const union cptable *table = get_codepage_table( codepage );
 149
 150     if (!table)
 151     {
 152         SetLastError( ERROR_INVALID_PARAMETER );
 153         return FALSE;
 154     }
 155     if (table->info.def_char & 0xff00)
 156     {
 157         cpinfo->DefaultChar[0] = table->info.def_char & 0xff00;
 158         cpinfo->DefaultChar[1] = table->info.def_char & 0x00ff;
 159     }
 160     else
 161     {
 162         cpinfo->DefaultChar[0] = table->info.def_char & 0xff;
 163         cpinfo->DefaultChar[1] = 0;
 164     }
 165     if ((cpinfo->MaxCharSize = table->info.char_size) == 2)
 166         memcpy( cpinfo->LeadByte, table->dbcs.lead_bytes, sizeof(cpinfo->LeadByte) );
 167     else
 168         cpinfo->LeadByte[0] = cpinfo->LeadByte[1] = 0;
 169
 170     return TRUE;
 171 }
 172
 173
 174 /***********************************************************************
 175  *              EnumSystemCodePagesA   (KERNEL32.@)
 176  */
 177 BOOL WINAPI EnumSystemCodePagesA( CODEPAGE_ENUMPROCA lpfnCodePageEnum, DWORD flags )
 178 {
 179     const union cptable *table;
 180     char buffer[10];
 181     int index = 0;
 182
 183     for (;;)
 184     {
 185         if (!(table = cp_enum_table( index++ ))) break;
 186         sprintf( buffer, "%d", table->info.codepage );
 187         if (!lpfnCodePageEnum( buffer )) break;
 188     }
 189     return TRUE;
 190 }
 191
 192
 193 /***********************************************************************
 194  *              EnumSystemCodePagesW   (KERNEL32.@)
 195  */
 196 BOOL WINAPI EnumSystemCodePagesW( CODEPAGE_ENUMPROCW lpfnCodePageEnum, DWORD flags )
 197 {
 198     const union cptable *table;
 199     WCHAR buffer[10], *p;
 200     int page, index = 0;
 201
 202     for (;;)
 203     {
 204         if (!(table = cp_enum_table( index++ ))) break;
 205         p = buffer + sizeof(buffer)/sizeof(WCHAR);
 206         *--p = 0;
 207         page = table->info.codepage;
 208         do
 209         {
 210             *--p = '0' + (page % 10);
 211             page /= 10;
 212         } while( page );
 213         if (!lpfnCodePageEnum( p )) break;
 214     }
 215     return TRUE;
 216 }
 217
 218
 219 /***********************************************************************
 220  *              MultiByteToWideChar   (KERNEL32.@)
 221  *
 222  * PARAMS
 223  *   page [in]    Codepage character set to convert from
 224  *   flags [in]   Character mapping flags
 225  *   src [in]     Source string buffer
 226  *   srclen [in]  Length of source string buffer
 227  *   dst [in]     Destination buffer
 228  *   dstlen [in]  Length of destination buffer
 229  *
 230  * NOTES
 231  *   The returned length includes the null terminator character.
 232  *
 233  * RETURNS
 234  *   Success: If dstlen > 0, number of characters written to destination
 235  *            buffer.  If dstlen == 0, number of characters needed to do
 236  *            conversion.
 237  *   Failure: 0. Occurs if not enough space is available.
 238  *
 239  * ERRORS
 240  *   ERROR_INSUFFICIENT_BUFFER
 241  *   ERROR_INVALID_PARAMETER
 242  *   ERROR_NO_UNICODE_TRANSLATION
 243  *
 244  */
 245 INT WINAPI MultiByteToWideChar( UINT page, DWORD flags, LPCSTR src, INT srclen,
 246                                 LPWSTR dst, INT dstlen )
 247 {
 248     const union cptable *table;
 249     int ret;
 250
 251     if (!src || (!dst && dstlen))
 252     {
 253         SetLastError( ERROR_INVALID_PARAMETER );
 254         return 0;
 255     }
 256
 257     if (srclen == -1) srclen = strlen(src) + 1;
 258
 259     if (flags & MB_USEGLYPHCHARS) FIXME("MB_USEGLYPHCHARS not supported\n");
 260
 261     switch(page)
 262     {
 263     case CP_UTF7:
 264         FIXME("UTF not supported\n");
 265         SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
 266         return 0;
 267     case CP_UTF8:
 268         ret = utf8_mbstowcs( flags, src, srclen, dst, dstlen );
 269         break;
 270     default:
 271         if (!(table = get_codepage_table( page )))
 272         {
 273             SetLastError( ERROR_INVALID_PARAMETER );
 274             return 0;
 275         }
 276         ret = cp_mbstowcs( table, flags, src, srclen, dst, dstlen );
 277         break;
 278     }
 279
 280     if (ret < 0)
 281     {
 282         switch(ret)
 283         {
 284         case -1: SetLastError( ERROR_INSUFFICIENT_BUFFER ); break;
 285         case -2: SetLastError( ERROR_NO_UNICODE_TRANSLATION ); break;
 286         }
 287         ret = 0;
 288     }
 289     return ret;
 290 }
 291
 292
 293 /***********************************************************************
 294  *              WideCharToMultiByte   (KERNEL32.@)
 295  *
 296  * PARAMS
 297  *   page [in]    Codepage character set to convert to
 298  *   flags [in]   Character mapping flags
 299  *   src [in]     Source string buffer
 300  *   srclen [in]  Length of source string buffer
 301  *   dst [in]     Destination buffer
 302  *   dstlen [in]  Length of destination buffer
 303  *   defchar [in] Default character to use for conversion if no exact
 304  *                  conversion can be made
 305  *   used [out]   Set if default character was used in the conversion
 306  *
 307  * NOTES
 308  *   The returned length includes the null terminator character.
 309  *
 310  * RETURNS
 311  *   Success: If dstlen > 0, number of characters written to destination
 312  *            buffer.  If dstlen == 0, number of characters needed to do
 313  *            conversion.
 314  *   Failure: 0. Occurs if not enough space is available.
 315  *
 316  * ERRORS
 317  *   ERROR_INSUFFICIENT_BUFFER
 318  *   ERROR_INVALID_PARAMETER
 319  */
 320 INT WINAPI WideCharToMultiByte( UINT page, DWORD flags, LPCWSTR src, INT srclen,
 321                                 LPSTR dst, INT dstlen, LPCSTR defchar, BOOL *used )
 322 {
 323     const union cptable *table;
 324     int ret, used_tmp;
 325
 326     if (!src || (!dst && dstlen))
 327     {
 328         SetLastError( ERROR_INVALID_PARAMETER );
 329         return 0;
 330     }
 331
 332     if (srclen == -1) srclen = strlenW(src) + 1;
 333
 334     switch(page)
 335     {
 336     case CP_UTF7:
 337         FIXME("UTF-7 not supported\n");
 338         SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
 339         return 0;
 340     case CP_UTF8:
 341         ret = utf8_wcstombs( src, srclen, dst, dstlen );
 342         break;
 343     default:
 344         if (!(table = get_codepage_table( page )))
 345         {
 346             SetLastError( ERROR_INVALID_PARAMETER );
 347             return 0;
 348         }
 349         ret = cp_wcstombs( table, flags, src, srclen, dst, dstlen,
 350                            defchar, used ? &used_tmp : NULL );
 351         if (used) *used = used_tmp;
 352         break;
 353     }
 354
 355     if (ret == -1)
 356     {
 357         SetLastError( ERROR_INSUFFICIENT_BUFFER );
 358         ret = 0;
 359     }
 360     return ret;
 361 }
 362
 363
 364 /******************************************************************************
 365  *              GetStringTypeW   (KERNEL32.@)
 366  *
 367  */
 368 BOOL WINAPI GetStringTypeW( DWORD type, LPCWSTR src, INT count, LPWORD chartype )
 369 {
 370     if (count == -1) count = strlenW(src) + 1;
 371     switch(type)
 372     {
 373     case CT_CTYPE1:
 374         while (count--) *chartype++ = get_char_typeW( *src++ ) & 0xfff;
 375         break;
 376     case CT_CTYPE2:
 377         while (count--) *chartype++ = get_char_typeW( *src++ ) >> 12;
 378         break;
 379     case CT_CTYPE3:
 380         FIXME("CT_CTYPE3 not supported.\n");
 381     default:
 382         SetLastError( ERROR_INVALID_PARAMETER );
 383         return FALSE;
 384     }
 385     return TRUE;
 386 }
 387
 388
 389 /******************************************************************************
 390  *              GetStringTypeExW   (KERNEL32.@)
 391  */
 392 BOOL WINAPI GetStringTypeExW( LCID locale, DWORD type, LPCWSTR src, INT count, LPWORD chartype )
 393 {
 394     /* locale is ignored for Unicode */
 395     return GetStringTypeW( type, src, count, chartype );
 396 }