lingucomponent/source/lingutil/lingutil.cxx

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2000, 2010 Oracle and/or its affiliates.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * This file is part of OpenOffice.org.
  10  *
  11  * OpenOffice.org is free software: you can redistribute it and/or modify
  12  * it under the terms of the GNU Lesser General Public License version 3
  13  * only, as published by the Free Software Foundation.
  14  *
  15  * OpenOffice.org is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18  * GNU Lesser General Public License version 3 for more details
  19  * (a copy is included in the LICENSE file that accompanied this code).
  20  *
  21  * You should have received a copy of the GNU Lesser General Public License
  22  * version 3 along with OpenOffice.org.  If not, see
  23  * <http://www.openoffice.org/license.html>
  24  * for a copy of the LGPLv3 License.
  25  *
  26  ************************************************************************/
  27
  28 // MARKER(update_precomp.py): autogen include statement, do not remove
  29 #include "precompiled_lingucomponent.hxx"
  30
  31 #if defined(WNT)
  32 #include <tools/prewin.h>
  33 #endif
  34
  35 #if defined(WNT)
  36 #include <Windows.h>
  37 #endif
  38
  39 #if defined(WNT)
  40 #include <tools/postwin.h>
  41 #endif
  42
  43
  44 #include <osl/thread.h>
  45 #include <osl/file.hxx>
  46 #include <tools/debug.hxx>
  47 #include <tools/urlobj.hxx>
  48 #include <i18npool/mslangid.hxx>
  49 #include <unotools/lingucfg.hxx>
  50 #include <unotools/pathoptions.hxx>
  51 #include <rtl/ustring.hxx>
  52 #include <rtl/string.hxx>
  53 #include <rtl/tencinfo.h>
  54 #include <linguistic/misc.hxx>
  55
  56 #include <set>
  57 #include <vector>
  58 #include <string.h>
  59
  60 #include <lingutil.hxx>
  61 #include <dictmgr.hxx>
  62
  63
  64
  65
  66 using ::com::sun::star::lang::Locale;
  67 using namespace ::com::sun::star;
  68
  69 #if 0
  70 //////////////////////////////////////////////////////////////////////
  71
  72 String GetDirectoryPathFromFileURL( const String &rFileURL )
  73 {
  74     // get file URL
  75     INetURLObject aURLObj;
  76     aURLObj.SetSmartProtocol( INET_PROT_FILE );
  77     aURLObj.SetSmartURL( rFileURL );
  78     aURLObj.removeSegment();
  79     DBG_ASSERT( !aURLObj.HasError(), "invalid URL" );
  80     String aRes = aURLObj.GetMainURL( INetURLObject::DECODE_TO_IURI );
  81     return aRes;
  82 }
  83 #endif
  84
  85 #if defined(WNT)
  86 rtl::OString Win_GetShortPathName( const rtl::OUString &rLongPathName )
  87 {
  88     rtl::OString aRes;
  89
  90     sal_Unicode aShortBuffer[1024] = {0};
  91     sal_Int32   nShortBufSize = sizeof( aShortBuffer ) / sizeof( aShortBuffer[0] );
  92
  93     // use the version of 'GetShortPathName' that can deal with Unicode...
  94     sal_Int32 nShortLen = GetShortPathNameW(
  95             reinterpret_cast<LPCWSTR>( rLongPathName.getStr() ),
  96             reinterpret_cast<LPWSTR>( aShortBuffer ),
  97             nShortBufSize );
  98
  99     if (nShortLen < nShortBufSize) // conversion successful?
 100         aRes = rtl::OString( OU2ENC( rtl::OUString( aShortBuffer, nShortLen ), osl_getThreadTextEncoding()) );
 101     else
 102         DBG_ERROR( "Win_GetShortPathName: buffer to short" );
 103
 104     return aRes;
 105 }
 106 #endif //defined(WNT)
 107
 108 //////////////////////////////////////////////////////////////////////
 109
 110 // build list of old style diuctionaries (not as extensions) to use.
 111 // User installed dictionaries (the ones residing in the user paths)
 112 // will get precedence over system installed ones for the same language.
 113 std::vector< SvtLinguConfigDictionaryEntry > GetOldStyleDics( const char *pDicType )
 114 {
 115     std::vector< SvtLinguConfigDictionaryEntry > aRes;
 116
 117     if (!pDicType)
 118         return aRes;
 119
 120     rtl::OUString aFormatName;
 121     String aDicExtension;
 122 #ifdef SYSTEM_DICTS
 123     rtl::OUString aSystemDir;
 124     rtl::OUString aSystemPrefix;
 125     rtl::OUString aSystemSuffix;
 126 #endif
 127     bool bSpell = false;
 128     bool bHyph  = false;
 129     bool bThes  = false;
 130     if (strcmp( pDicType, "DICT" ) == 0)
 131     {
 132         aFormatName     = A2OU("DICT_SPELL");
 133         aDicExtension   = String::CreateFromAscii( ".dic" );
 134 #ifdef SYSTEM_DICTS
 135         aSystemDir      = A2OU( DICT_SYSTEM_DIR );
 136         aSystemSuffix       = aDicExtension;
 137 #endif
 138         bSpell = true;
 139     }
 140     else if (strcmp( pDicType, "HYPH" ) == 0)
 141     {
 142         aFormatName     = A2OU("DICT_HYPH");
 143         aDicExtension   = String::CreateFromAscii( ".dic" );
 144 #ifdef SYSTEM_DICTS
 145         aSystemDir      = A2OU( HYPH_SYSTEM_DIR );
 146         aSystemPrefix       = A2OU( "hyph_" );
 147         aSystemSuffix       = aDicExtension;
 148 #endif
 149         bHyph = true;
 150     }
 151     else if (strcmp( pDicType, "THES" ) == 0)
 152     {
 153         aFormatName     = A2OU("DICT_THES");
 154         aDicExtension   = String::CreateFromAscii( ".dat" );
 155 #ifdef SYSTEM_DICTS
 156         aSystemDir      = A2OU( THES_SYSTEM_DIR );
 157         aSystemPrefix       = A2OU( "th_" );
 158         aSystemSuffix       = A2OU( "_v2.dat" );
 159 #endif
 160         bThes = true;
 161     }
 162
 163
 164     if (aFormatName.getLength() == 0 || aDicExtension.Len() == 0)
 165         return aRes;
 166
 167     // set of languages to remember the language where it is already
 168     // decided to make use of the dictionary.
 169     std::set< LanguageType > aDicLangInUse;
 170
 171 #ifdef SYSTEM_DICTS
 172    osl::Directory aSystemDicts(aSystemDir);
 173    if (aSystemDicts.open() == osl::FileBase::E_None)
 174    {
 175        osl::DirectoryItem aItem;
 176        osl::FileStatus aFileStatus(FileStatusMask_FileURL);
 177        while (aSystemDicts.getNextItem(aItem) == osl::FileBase::E_None)
 178        {
 179            aItem.getFileStatus(aFileStatus);
 180            rtl::OUString sPath = aFileStatus.getFileURL();
 181            if (sPath.lastIndexOf(aSystemSuffix) == sPath.getLength()-aSystemSuffix.getLength())
 182            {
 183                sal_Int32 nStartIndex = sPath.lastIndexOf(sal_Unicode('/')) + 1;
 184                if (!sPath.match(aSystemPrefix, nStartIndex))
 185                    continue;
 186                rtl::OUString sChunk = sPath.copy(0, sPath.getLength() - aSystemSuffix.getLength());
 187                sal_Int32 nIndex = nStartIndex + aSystemPrefix.getLength();
 188                rtl::OUString sLang = sChunk.getToken( 0, '_', nIndex );
 189                if (!sLang.getLength())
 190                    continue;
 191                rtl::OUString sRegion;
 192                if (nIndex != -1)
 193                    sRegion = sChunk.copy( nIndex, sChunk.getLength() - nIndex );
 194
 195                // Thus we first get the language of the dictionary
 196                LanguageType nLang = MsLangId::convertIsoNamesToLanguage(
 197                   sLang, sRegion );
 198
 199                if (aDicLangInUse.count( nLang ) == 0)
 200                {
 201                    // remember the new language in use
 202                    aDicLangInUse.insert( nLang );
 203
 204                    // add the dictionary to the resulting vector
 205                    SvtLinguConfigDictionaryEntry aDicEntry;
 206                    aDicEntry.aLocations.realloc(1);
 207                    aDicEntry.aLocaleNames.realloc(1);
 208                    rtl::OUString aLocaleName( MsLangId::convertLanguageToIsoString( nLang ) );
 209                    aDicEntry.aLocations[0] = sPath;
 210                    aDicEntry.aFormatName = aFormatName;
 211                    aDicEntry.aLocaleNames[0] = aLocaleName;
 212                    aRes.push_back( aDicEntry );
 213                }
 214            }
 215        }
 216     }
 217
 218 #endif
 219
 220     return aRes;
 221 }
 222
 223
 224 void MergeNewStyleDicsAndOldStyleDics(
 225     std::list< SvtLinguConfigDictionaryEntry > &rNewStyleDics,
 226     const std::vector< SvtLinguConfigDictionaryEntry > &rOldStyleDics )
 227 {
 228     // get list of languages supported by new style dictionaries
 229     std::set< LanguageType > aNewStyleLanguages;
 230     std::list< SvtLinguConfigDictionaryEntry >::const_iterator aIt;
 231     for (aIt = rNewStyleDics.begin() ;  aIt != rNewStyleDics.end();  ++aIt)
 232     {
 233         const uno::Sequence< rtl::OUString > aLocaleNames( aIt->aLocaleNames );
 234         sal_Int32 nLocaleNames = aLocaleNames.getLength();
 235         for (sal_Int32 k = 0;  k < nLocaleNames; ++k)
 236         {
 237             LanguageType nLang = MsLangId::convertIsoStringToLanguage( aLocaleNames[k] );
 238             aNewStyleLanguages.insert( nLang );
 239         }
 240     }
 241
 242     // now check all old style dictionaries if they will add a not yet
 243     // added language. If so add them to the resulting vector
 244     std::vector< SvtLinguConfigDictionaryEntry >::const_iterator aIt2;
 245     for (aIt2 = rOldStyleDics.begin();  aIt2 != rOldStyleDics.end();  ++aIt2)
 246     {
 247         sal_Int32 nOldStyleDics = aIt2->aLocaleNames.getLength();
 248
 249         // old style dics should only have one language listed...
 250         DBG_ASSERT( nOldStyleDics, "old style dictionary with more then one language found!");
 251         if (nOldStyleDics > 0)
 252         {
 253             LanguageType nLang = MsLangId::convertIsoStringToLanguage( aIt2->aLocaleNames[0] );
 254
 255             if (nLang == LANGUAGE_DONTKNOW || nLang == LANGUAGE_NONE)
 256             {
 257                 DBG_ERROR( "old style dictionary with invalid language found!" );
 258                 continue;
 259             }
 260
 261             // language not yet added?
 262             if (aNewStyleLanguages.count( nLang ) == 0)
 263                 rNewStyleDics.push_back( *aIt2 );
 264         }
 265         else
 266         {
 267             DBG_ERROR( "old style dictionary with no language found!" );
 268         }
 269     }
 270 }
 271
 272
 273 rtl_TextEncoding getTextEncodingFromCharset(const sal_Char* pCharset)
 274 {
 275     // default result: used to indicate that we failed to get the proper encoding
 276     rtl_TextEncoding eRet = RTL_TEXTENCODING_DONTKNOW;
 277
 278     if (pCharset)
 279     {
 280         eRet = rtl_getTextEncodingFromMimeCharset(pCharset);
 281         if (eRet == RTL_TEXTENCODING_DONTKNOW)
 282             eRet = rtl_getTextEncodingFromUnixCharset(pCharset);
 283         if (eRet == RTL_TEXTENCODING_DONTKNOW)
 284         {
 285             if (strcmp("ISCII-DEVANAGARI", pCharset) == 0)
 286                 eRet = RTL_TEXTENCODING_ISCII_DEVANAGARI;
 287         }
 288     }
 289     return eRet;
 290 }
 291
 292 //////////////////////////////////////////////////////////////////////
 293