sal/osl/os2/nlsupport.c

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2008 by Sun Microsystems, Inc.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * $RCSfile: nlsupport.c,v $
  10  * $Revision: 1.5 $
  11  *
  12  * This file is part of OpenOffice.org.
  13  *
  14  * OpenOffice.org is free software: you can redistribute it and/or modify
  15  * it under the terms of the GNU Lesser General Public License version 3
  16  * only, as published by the Free Software Foundation.
  17  *
  18  * OpenOffice.org is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21  * GNU Lesser General Public License version 3 for more details
  22  * (a copy is included in the LICENSE file that accompanied this code).
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * version 3 along with OpenOffice.org.  If not, see
  26  * <http://www.openoffice.org/license.html>
  27  * for a copy of the LGPLv3 License.
  28  *
  29  ************************************************************************/
  30
  31 #define INCL_WIN
  32 #include "svpm.h"
  33
  34 #include <osl/nlsupport.h>
  35 #include <osl/diagnose.h>
  36 #include <osl/process.h>
  37 #include <rtl/memory.h>
  38
  39 #include <string.h>
  40
  41 /*****************************************************************************
  42  typedefs
  43  *****************************************************************************/
  44
  45
  46 typedef struct {
  47     const char              *key;
  48     const rtl_TextEncoding   value;
  49 } _pair;
  50
  51
  52 /*****************************************************************************
  53  compare function for binary search
  54  *****************************************************************************/
  55
  56 static int
  57 _pair_compare (const char *key, const _pair *pair)
  58 {
  59     int result = rtl_str_compareIgnoreAsciiCase( key, pair->key );
  60     return result;
  61 }
  62
  63 /*****************************************************************************
  64  binary search on encoding tables
  65  *****************************************************************************/
  66
  67 static const _pair*
  68 _pair_search (const char *key, const _pair *base, unsigned int member )
  69 {
  70     unsigned int lower = 0;
  71     unsigned int upper = member;
  72     unsigned int current;
  73     int comparison;
  74
  75     /* check for validity of input */
  76     if ( (key == NULL) || (base == NULL) || (member == 0) )
  77         return NULL;
  78
  79     /* binary search */
  80     while ( lower < upper )
  81     {
  82         current = (lower + upper) / 2;
  83         comparison = _pair_compare( key, base + current );
  84         if (comparison < 0)
  85             upper = current;
  86         else if (comparison > 0)
  87             lower = current + 1;
  88         else
  89             return base + current;
  90     }
  91
  92     return NULL;
  93 }
  94
  95
  96 /*****************************************************************************
  97  convert rtl_Locale to locale string
  98  *****************************************************************************/
  99
 100 static char * _compose_locale( rtl_Locale * pLocale, char * buffer, size_t n )
 101 {
 102     /* check if a valid locale is specified */
 103     if( pLocale && pLocale->Language && (pLocale->Language->length == 2) )
 104     {
 105         size_t offset = 0;
 106
 107         /* convert language code to ascii */
 108         {
 109             rtl_String *pLanguage = NULL;
 110
 111             rtl_uString2String( &pLanguage,
 112                 pLocale->Language->buffer, pLocale->Language->length,
 113                 RTL_TEXTENCODING_ASCII_US, OUSTRING_TO_OSTRING_CVTFLAGS );
 114
 115             if( pLanguage->length < n )
 116             {
 117                 strcpy( buffer, pLanguage->buffer );
 118                 offset = pLanguage->length;
 119             }
 120
 121             rtl_string_release( pLanguage );
 122         }
 123
 124         /* convert country code to ascii */
 125         if( pLocale->Country && (pLocale->Country->length == 2) )
 126         {
 127             rtl_String *pCountry = NULL;
 128
 129             rtl_uString2String( &pCountry,
 130                 pLocale->Country->buffer, pLocale->Country->length,
 131                 RTL_TEXTENCODING_ASCII_US, OUSTRING_TO_OSTRING_CVTFLAGS );
 132
 133             if( offset + pCountry->length + 1 < n )
 134             {
 135                 strcpy( buffer + offset++, "_" );
 136                 strcpy( buffer + offset, pCountry->buffer );
 137                 offset += pCountry->length;
 138             }
 139
 140             rtl_string_release( pCountry );
 141         }
 142
 143         /* convert variant to ascii - check if there is enough space for the variant string */
 144         if( pLocale->Variant && pLocale->Variant->length &&
 145             ( pLocale->Variant->length < n - 6 ) )
 146         {
 147             rtl_String *pVariant = NULL;
 148
 149             rtl_uString2String( &pVariant,
 150                 pLocale->Variant->buffer, pLocale->Variant->length,
 151                 RTL_TEXTENCODING_ASCII_US, OUSTRING_TO_OSTRING_CVTFLAGS );
 152
 153             if( offset + pVariant->length + 1 < n )
 154             {
 155                 strcpy( buffer + offset, pVariant->buffer );
 156                 offset += pVariant->length;
 157             }
 158
 159             rtl_string_release( pVariant );
 160         }
 161
 162         return buffer;
 163     }
 164
 165     return NULL;
 166 }
 167
 168 /*****************************************************************************
 169  convert locale string to rtl_Locale
 170  *****************************************************************************/
 171
 172 static rtl_Locale * _parse_locale( const char * locale )
 173 {
 174     static sal_Unicode c_locale[2] = { (sal_Unicode) 'C', 0 };
 175
 176     /* check if locale contains a valid string */
 177     if( locale )
 178     {
 179         size_t len = strlen( locale );
 180
 181         if( len >= 2 )
 182         {
 183             rtl_uString * pLanguage = NULL;
 184             rtl_uString * pCountry  = NULL;
 185             rtl_uString * pVariant  = NULL;
 186
 187             size_t offset = 2;
 188
 189             /* convert language code to unicode */
 190             rtl_string2UString( &pLanguage, locale, 2, RTL_TEXTENCODING_ASCII_US, OSTRING_TO_OUSTRING_CVTFLAGS );
 191             OSL_ASSERT(pLanguage != NULL);
 192
 193             /* convert country code to unicode */
 194             if( len >= 5 && '_' == locale[2] )
 195             {
 196                 rtl_string2UString( &pCountry, locale + 3, 2, RTL_TEXTENCODING_ASCII_US, OSTRING_TO_OUSTRING_CVTFLAGS );
 197                 OSL_ASSERT(pCountry != NULL);
 198                 offset = 5;
 199             }
 200
 201             /* convert variant code to unicode - do not rely on "." as delimiter */
 202             if( len > offset ) {
 203                 rtl_string2UString( &pVariant, locale + offset, len - offset, RTL_TEXTENCODING_ASCII_US, OSTRING_TO_OUSTRING_CVTFLAGS );
 204                 OSL_ASSERT(pVariant != NULL);
 205             }
 206
 207             rtl_Locale * ret =  rtl_locale_register( pLanguage->buffer, pCountry ? pCountry->buffer : c_locale + 1, pVariant ? pVariant->buffer : c_locale + 1 );
 208
 209             if (pVariant) rtl_uString_release(pVariant);
 210             if (pCountry) rtl_uString_release(pCountry);
 211             if (pLanguage) rtl_uString_release(pLanguage);
 212
 213             return ret;
 214         }
 215         else
 216             return rtl_locale_register( c_locale, c_locale + 1, c_locale + 1 );
 217     }
 218
 219     return NULL;
 220 }
 221
 222 /*
 223  * This implementation of osl_getTextEncodingFromLocale maps
 224  * from the ISO language codes.
 225  */
 226
 227 const _pair _full_locale_list[] = {
 228     { "ja_JP.eucJP",  RTL_TEXTENCODING_EUC_JP      },
 229     { "ja_JP.EUC",    RTL_TEXTENCODING_EUC_JP      },
 230     { "ko_KR.EUC",    RTL_TEXTENCODING_EUC_KR      },
 231     { "zh_CN.EUC",    RTL_TEXTENCODING_EUC_CN      },
 232     { "zh_TW.EUC",    RTL_TEXTENCODING_EUC_TW      }
 233 };
 234
 235 const _pair _locale_extension_list[] = {
 236     { "big5",         RTL_TEXTENCODING_BIG5        },
 237     { "big5hk",       RTL_TEXTENCODING_BIG5_HKSCS  },
 238     { "gb18030",      RTL_TEXTENCODING_GB_18030    },
 239     { "euc",          RTL_TEXTENCODING_EUC_JP      },
 240     { "iso8859-1",    RTL_TEXTENCODING_ISO_8859_1  },
 241     { "iso8859-10",   RTL_TEXTENCODING_ISO_8859_10 },
 242     { "iso8859-13",   RTL_TEXTENCODING_ISO_8859_13 },
 243     { "iso8859-14",   RTL_TEXTENCODING_ISO_8859_14 },
 244     { "iso8859-15",   RTL_TEXTENCODING_ISO_8859_15 },
 245     { "iso8859-2",    RTL_TEXTENCODING_ISO_8859_2  },
 246     { "iso8859-3",    RTL_TEXTENCODING_ISO_8859_3  },
 247     { "iso8859-4",    RTL_TEXTENCODING_ISO_8859_4  },
 248     { "iso8859-5",    RTL_TEXTENCODING_ISO_8859_5  },
 249     { "iso8859-6",    RTL_TEXTENCODING_ISO_8859_6  },
 250     { "iso8859-7",    RTL_TEXTENCODING_ISO_8859_7  },
 251     { "iso8859-8",    RTL_TEXTENCODING_ISO_8859_8  },
 252     { "iso8859-9",    RTL_TEXTENCODING_ISO_8859_9  },
 253     { "koi8-r",       RTL_TEXTENCODING_KOI8_R      },
 254     { "koi8-u",       RTL_TEXTENCODING_KOI8_U      },
 255     { "pck",          RTL_TEXTENCODING_MS_932      },
 256 #if (0)
 257     { "sun_eu_greek", RTL_TEXTENCODING_DONTKNOW    },
 258 #endif
 259     { "utf-16",       RTL_TEXTENCODING_UNICODE     },
 260     { "utf-7",        RTL_TEXTENCODING_UTF7        },
 261     { "utf-8",        RTL_TEXTENCODING_UTF8        }
 262 };
 263
 264 const _pair _iso_language_list[] = {
 265     { "af",  RTL_TEXTENCODING_ISO_8859_1 },
 266     { "ar",  RTL_TEXTENCODING_ISO_8859_6 },
 267     { "az",  RTL_TEXTENCODING_ISO_8859_9 },
 268     { "be",  RTL_TEXTENCODING_ISO_8859_5 },
 269     { "bg",  RTL_TEXTENCODING_ISO_8859_5 },
 270     { "ca",  RTL_TEXTENCODING_ISO_8859_1 },
 271     { "cs",  RTL_TEXTENCODING_ISO_8859_2 },
 272     { "da",  RTL_TEXTENCODING_ISO_8859_1 },
 273     { "de",  RTL_TEXTENCODING_ISO_8859_1 },
 274     { "el",  RTL_TEXTENCODING_ISO_8859_7 },
 275     { "en",  RTL_TEXTENCODING_ISO_8859_1 },
 276     { "es",  RTL_TEXTENCODING_ISO_8859_1 },
 277     { "et",  RTL_TEXTENCODING_ISO_8859_4 },
 278     { "eu",  RTL_TEXTENCODING_ISO_8859_1 },
 279     { "fa",  RTL_TEXTENCODING_ISO_8859_6 },
 280     { "fi",  RTL_TEXTENCODING_ISO_8859_1 },
 281     { "fo",  RTL_TEXTENCODING_ISO_8859_1 },
 282     { "fr",  RTL_TEXTENCODING_ISO_8859_1 },
 283     { "gr",  RTL_TEXTENCODING_ISO_8859_7 },
 284     { "he",  RTL_TEXTENCODING_ISO_8859_8 },
 285     { "hi",  RTL_TEXTENCODING_DONTKNOW },
 286     { "hr",  RTL_TEXTENCODING_ISO_8859_2 },
 287     { "hu",  RTL_TEXTENCODING_ISO_8859_2 },
 288     { "hy",  RTL_TEXTENCODING_DONTKNOW },
 289     { "id",  RTL_TEXTENCODING_ISO_8859_1 },
 290     { "is",  RTL_TEXTENCODING_ISO_8859_1 },
 291     { "it",  RTL_TEXTENCODING_ISO_8859_1 },
 292     { "iw",  RTL_TEXTENCODING_ISO_8859_8 },
 293     { "ja",  RTL_TEXTENCODING_EUC_JP },
 294     { "ka",  RTL_TEXTENCODING_DONTKNOW },
 295     { "kk",  RTL_TEXTENCODING_ISO_8859_5 },
 296     { "ko",  RTL_TEXTENCODING_EUC_KR },
 297     { "lt",  RTL_TEXTENCODING_ISO_8859_4 },
 298     { "lv",  RTL_TEXTENCODING_ISO_8859_4 },
 299     { "mk",  RTL_TEXTENCODING_ISO_8859_5 },
 300     { "mr",  RTL_TEXTENCODING_DONTKNOW },
 301     { "ms",  RTL_TEXTENCODING_ISO_8859_1 },
 302     { "nl",  RTL_TEXTENCODING_ISO_8859_1 },
 303     { "no",  RTL_TEXTENCODING_ISO_8859_1 },
 304     { "pl",  RTL_TEXTENCODING_ISO_8859_2 },
 305     { "pt",  RTL_TEXTENCODING_ISO_8859_1 },
 306     { "ro",  RTL_TEXTENCODING_ISO_8859_2 },
 307     { "ru",  RTL_TEXTENCODING_ISO_8859_5 },
 308     { "sa",  RTL_TEXTENCODING_DONTKNOW },
 309     { "sk",  RTL_TEXTENCODING_ISO_8859_2 },
 310     { "sl",  RTL_TEXTENCODING_ISO_8859_2 },
 311     { "sq",  RTL_TEXTENCODING_ISO_8859_2 },
 312     { "sv",  RTL_TEXTENCODING_ISO_8859_1 },
 313     { "sw",  RTL_TEXTENCODING_ISO_8859_1 },
 314     { "ta",  RTL_TEXTENCODING_DONTKNOW },
 315     { "th",  RTL_TEXTENCODING_DONTKNOW },
 316     { "tr",  RTL_TEXTENCODING_ISO_8859_9 },
 317     { "tt",  RTL_TEXTENCODING_ISO_8859_5 },
 318     { "uk",  RTL_TEXTENCODING_ISO_8859_5 },
 319     { "ur",  RTL_TEXTENCODING_ISO_8859_6 },
 320     { "uz",  RTL_TEXTENCODING_ISO_8859_9 },
 321     { "vi",  RTL_TEXTENCODING_DONTKNOW },
 322     { "zh",  RTL_TEXTENCODING_BIG5 }
 323 };
 324
 325 /*****************************************************************************
 326  return the text encoding corresponding to the given locale
 327  *****************************************************************************/
 328
 329 rtl_TextEncoding osl_getTextEncodingFromLocale( rtl_Locale * pLocale )
 330 {
 331     const _pair *language = 0;
 332     char locale_buf[64] = "";
 333     char *cp;
 334
 335     WinMessageBox(HWND_DESKTOP,HWND_DESKTOP,
 336         "Please contact technical support and report above informations.\n\n",
 337         "Critical error: osl_getTextEncodingFromLocale",
 338         0, MB_ERROR | MB_OK | MB_MOVEABLE);
 339
 340     /* default to process locale if pLocale == NULL */
 341     if( NULL == pLocale )
 342         osl_getProcessLocale( &pLocale );
 343
 344     /* convert rtl_Locale to locale string */
 345     if( _compose_locale( pLocale, locale_buf, 64 ) )
 346     {
 347         /* check special handling list (EUC) first */
 348         const unsigned int members = sizeof( _full_locale_list ) / sizeof( _pair );
 349         language = _pair_search( locale_buf, _full_locale_list, members);
 350
 351         if( NULL == language )
 352         {
 353             /*
 354              *  check if there is a charset qualifier at the end of the given locale string
 355              *  e.g. de.ISO8859-15 or de.ISO8859-15@euro which strongly indicates what
 356              *  charset to use
 357              */
 358             cp = strrchr( locale_buf, '.' );
 359
 360             if( NULL != cp )
 361             {
 362                 const unsigned int members = sizeof( _locale_extension_list ) / sizeof( _pair );
 363                 language = _pair_search( cp + 1, _locale_extension_list, members);
 364             }
 365         }
 366
 367         /* use iso language code to determine the charset */
 368         if( NULL == language )
 369         {
 370             const unsigned int members = sizeof( _iso_language_list ) / sizeof( _pair );
 371
 372             /* iso lang codes have 2 charaters */
 373             locale_buf[2] = '\0';
 374
 375             language = _pair_search( locale_buf, _iso_language_list, members);
 376         }
 377     }
 378
 379     /* a matching item in our list provides a mapping from codeset to
 380      * rtl-codeset */
 381     if ( language != NULL )
 382         return language->value;
 383
 384     return RTL_TEXTENCODING_DONTKNOW;
 385 }
 386
 387 /*****************************************************************************
 388  return the current process locale
 389  *****************************************************************************/
 390
 391 void _imp_getProcessLocale( rtl_Locale ** ppLocale )
 392 {
 393     /* simulate behavior off setlocale */
 394     char * locale = getenv( "LC_ALL" );
 395
 396     if( NULL == locale )
 397         locale = getenv( "LC_CTYPE" );
 398
 399     if( NULL == locale )
 400         locale = getenv( "LANG" );
 401
 402     if( NULL == locale )
 403         locale = "C";
 404
 405     *ppLocale = _parse_locale( locale );
 406 }
 407
 408 /*****************************************************************************
 409  set the current process locale
 410  *****************************************************************************/
 411
 412 int _imp_setProcessLocale( rtl_Locale * pLocale )
 413 {
 414 #ifdef IRIX
 415     char env_buf[80];
 416 #endif
 417     char locale_buf[64];
 418
 419     /* convert rtl_Locale to locale string */
 420     if( NULL != _compose_locale( pLocale, locale_buf, 64 ) )
 421     {
 422         /* only change env vars that exist already */
 423         if( getenv( "LC_ALL" ) ) {
 424 #if defined( IRIX )
 425             snprintf(env_buf, sizeof(env_buf), "LC_ALL=%s", locale_buf);
 426             env_buf[sizeof(env_buf)] = '\0';
 427             putenv(env_buf);
 428 #elif defined( FREEBSD ) || defined( NETBSD ) || defined( MACOSX ) || defined( __EMX__ )
 429             setenv( "LC_ALL", locale_buf, 1);
 430 #else
 431             setenv( "LC_ALL", locale_buf );
 432 #endif
 433         }
 434
 435         if( getenv( "LC_CTYPE" ) ) {
 436 #if defined( IRIX )
 437             snprintf(env_buf, sizeof(env_buf), "LC_CTYPE=%s", locale_buf);
 438             env_buf[sizeof(env_buf)] = '\0';
 439             putenv(env_buf);
 440 #elif defined( FREEBSD ) || defined( NETBSD ) || defined( MACOSX ) || defined( __EMX__ )
 441             setenv("LC_CTYPE", locale_buf, 1 );
 442 #else
 443             setenv( "LC_CTYPE", locale_buf );
 444 #endif
 445         }
 446
 447         if( getenv( "LANG" ) ) {
 448 #if defined( IRIX )
 449             snprintf(env_buf, sizeof(env_buf), "LANG=%s", locale_buf);
 450             env_buf[sizeof(env_buf)] = '\0';
 451             putenv(env_buf);
 452 #elif defined( FREEBSD ) || defined( NETBSD ) || defined( MACOSX ) || defined( __EMX__ )
 453             setenv("LC_CTYPE", locale_buf, 1 );
 454 #else
 455             setenv( "LANG", locale_buf );
 456 #endif
 457         }
 458     }
 459
 460     return 0;
 461 }
 462
 463