lingucomponent/source/lingutil/lingutil.cxx

   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: lingutil.cxx,v $
  10  * $Revision: 1.4 $
  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 // MARKER(update_precomp.py): autogen include statement, do not remove
  32 #include "precompiled_lingucomponent.hxx"
  33
  34 #if defined(WNT)
  35 #include <tools/prewin.h>
  36 #endif
  37
  38 #if defined(WNT)
  39 #include <Windows.h>
  40 #endif
  41
  42 #if defined(WNT)
  43 #include <tools/postwin.h>
  44 #endif
  45
  46
  47 #include <osl/thread.h>
  48 #include <osl/file.hxx>
  49 #include <tools/debug.hxx>
  50 #include <tools/urlobj.hxx>
  51 #include <i18npool/mslangid.hxx>
  52 #include <svtools/lingucfg.hxx>
  53 #include <svtools/pathoptions.hxx>
  54 #include <rtl/ustring.hxx>
  55 #include <rtl/string.hxx>
  56 #include <linguistic/misc.hxx>
  57
  58 #include <set>
  59 #include <vector>
  60
  61 #include <lingutil.hxx>
  62 #include <dictmgr.hxx>
  63
  64
  65 using ::com::sun::star::lang::Locale;
  66 using namespace ::com::sun::star;
  67
  68 //////////////////////////////////////////////////////////////////////
  69
  70 String GetDirectoryPathFromFileURL( const String &rFileURL )
  71 {
  72     // get file URL
  73     INetURLObject aURLObj;
  74     aURLObj.SetSmartProtocol( INET_PROT_FILE );
  75     aURLObj.SetSmartURL( rFileURL );
  76     aURLObj.removeSegment();
  77     DBG_ASSERT( !aURLObj.HasError(), "invalid URL" );
  78     String aRes = aURLObj.GetMainURL( INetURLObject::DECODE_TO_IURI );
  79     return aRes;
  80 }
  81
  82
  83 rtl::OString Win_GetShortPathName( const rtl::OUString &rLongPathName )
  84 {
  85     (void) rLongPathName;
  86     rtl::OString aRes;
  87
  88 #if defined(WNT)
  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 #else
 104     DBG_ERROR( "Win_GetShortPathName: functions should nor be called in non-Windows builds" );
 105 #endif //defined(WNT)
 106
 107     return aRes;
 108 }
 109
 110 //////////////////////////////////////////////////////////////////////
 111
 112 // build list of old style diuctionaries (not as extensions) to use.
 113 // User installed dictionaries (the ones residing in the user paths)
 114 // will get precedence over system installed ones for the same language.
 115 std::vector< SvtLinguConfigDictionaryEntry > GetOldStyleDics( const char *pDicType )
 116 {
 117     std::vector< SvtLinguConfigDictionaryEntry > aRes;
 118
 119         if (!pDicType)
 120                 return aRes;
 121
 122         rtl::OUString aFormatName;
 123         String aDicExtension;
 124 #ifdef SYSTEM_DICTS
 125         rtl::OUString aSystemDir;
 126         rtl::OUString aSystemPrefix;
 127         rtl::OUString aSystemSuffix;
 128 #endif
 129         bool bSpell = false;
 130         bool bHyph  = false;
 131         bool bThes  = false;
 132     if (strcmp( pDicType, "DICT" ) == 0)
 133         {
 134                 aFormatName             = A2OU("DICT_SPELL");
 135                 aDicExtension   = String::CreateFromAscii( ".dic" );
 136 #ifdef SYSTEM_DICTS
 137                 aSystemDir              = A2OU( DICT_SYSTEM_DIR );
 138                 aSystemSuffix           = aDicExtension;
 139 #endif
 140                 bSpell = true;
 141         }
 142     else if (strcmp( pDicType, "HYPH" ) == 0)
 143         {
 144                 aFormatName             = A2OU("DICT_HYPH");
 145                 aDicExtension   = String::CreateFromAscii( ".dic" );
 146 #ifdef SYSTEM_DICTS
 147                 aSystemDir              = A2OU( HYPH_SYSTEM_DIR );
 148                 aSystemPrefix           = A2OU( "hyph_" );
 149                 aSystemSuffix           = aDicExtension;
 150 #endif
 151                 bHyph = true;
 152         }
 153     else if (strcmp( pDicType, "THES" ) == 0)
 154         {
 155                 aFormatName             = A2OU("DICT_THES");
 156                 aDicExtension   = String::CreateFromAscii( ".dat" );
 157 #ifdef SYSTEM_DICTS
 158                 aSystemDir              = A2OU( THES_SYSTEM_DIR );
 159                 aSystemPrefix           = A2OU( "th_" );
 160                 aSystemSuffix           = A2OU( "_v2.dat" );
 161 #endif
 162                 bThes = true;
 163         }
 164
 165
 166         if (aFormatName.getLength() == 0 || aDicExtension.Len() == 0)
 167                 return aRes;
 168
 169         // set of languages to remember the language where it is already
 170         // decided to make use of the dictionary.
 171         std::set< LanguageType > aDicLangInUse;
 172
 173 #ifdef SYSTEM_DICTS
 174    osl::Directory aSystemDicts(aSystemDir);
 175    if (aSystemDicts.open() == osl::FileBase::E_None)
 176    {
 177        osl::DirectoryItem aItem;
 178        osl::FileStatus aFileStatus(FileStatusMask_FileURL);
 179        while (aSystemDicts.getNextItem(aItem) == osl::FileBase::E_None)
 180        {
 181            aItem.getFileStatus(aFileStatus);
 182            rtl::OUString sPath = aFileStatus.getFileURL();
 183            if (sPath.lastIndexOf(aSystemSuffix) == sPath.getLength()-aSystemSuffix.getLength())
 184            {
 185                sal_Int32 nStartIndex = sPath.lastIndexOf(sal_Unicode('/')) + 1;
 186                if (!sPath.match(aSystemPrefix, nStartIndex))
 187                    continue;
 188                rtl::OUString sChunk = sPath.copy(0, sPath.getLength() - aSystemSuffix.getLength());
 189                sal_Int32 nIndex = nStartIndex + aSystemPrefix.getLength();
 190                rtl::OUString sLang = sChunk.getToken( 0, '_', nIndex );
 191                if (!sLang.getLength())
 192                    continue;
 193                rtl::OUString sRegion;
 194                if (nIndex != -1);
 195                    sRegion = sChunk.copy( nIndex, sChunk.getLength() - nIndex );
 196
 197                // Thus we first get the language of the dictionary
 198                LanguageType nLang = MsLangId::convertIsoNamesToLanguage(
 199                   sLang, sRegion );
 200
 201                if (aDicLangInUse.count( nLang ) == 0)
 202                {
 203                    // remember the new language in use
 204                    aDicLangInUse.insert( nLang );
 205
 206                    // add the dictionary to the resulting vector
 207                    SvtLinguConfigDictionaryEntry aDicEntry;
 208                    aDicEntry.aLocations.realloc(1);
 209                    aDicEntry.aLocaleNames.realloc(1);
 210                    rtl::OUString aLocaleName( MsLangId::convertLanguageToIsoString( nLang ) );
 211                    aDicEntry.aLocations[0] = sPath;
 212                    aDicEntry.aFormatName = aFormatName;
 213                    aDicEntry.aLocaleNames[0] = aLocaleName;
 214                    aRes.push_back( aDicEntry );
 215                }
 216            }
 217        }
 218     }
 219
 220 #endif
 221
 222     return aRes;
 223 }
 224
 225
 226 void MergeNewStyleDicsAndOldStyleDics(
 227         std::list< SvtLinguConfigDictionaryEntry > &rNewStyleDics,
 228         const std::vector< SvtLinguConfigDictionaryEntry > &rOldStyleDics )
 229 {
 230         // get list of languages supported by new style dictionaries
 231         std::set< LanguageType > aNewStyleLanguages;
 232         std::list< SvtLinguConfigDictionaryEntry >::const_iterator aIt;
 233         for (aIt = rNewStyleDics.begin() ;  aIt != rNewStyleDics.end();  ++aIt)
 234         {
 235                 const uno::Sequence< rtl::OUString > aLocaleNames( aIt->aLocaleNames );
 236                 sal_Int32 nLocaleNames = aLocaleNames.getLength();
 237                 for (sal_Int32 k = 0;  k < nLocaleNames; ++k)
 238                 {
 239                         LanguageType nLang = MsLangId::convertIsoStringToLanguage( aLocaleNames[k] );
 240                         aNewStyleLanguages.insert( nLang );
 241                 }
 242         }
 243
 244         // now check all old style dictionaries if they will add a not yet
 245         // added language. If so add them to the resulting vector
 246         std::vector< SvtLinguConfigDictionaryEntry >::const_iterator aIt2;
 247         for (aIt2 = rOldStyleDics.begin();  aIt2 != rOldStyleDics.end();  ++aIt2)
 248         {
 249                 sal_Int32 nOldStyleDics = aIt2->aLocaleNames.getLength();
 250
 251                 // old style dics should only have one language listed...
 252                 DBG_ASSERT( nOldStyleDics, "old style dictionary with more then one language found!");
 253                 if (nOldStyleDics > 0)
 254                 {
 255                         LanguageType nLang = MsLangId::convertIsoStringToLanguage( aIt2->aLocaleNames[0] );
 256
 257             if (nLang == LANGUAGE_DONTKNOW || nLang == LANGUAGE_NONE)
 258             {
 259                 DBG_ERROR( "old style dictionary with invalid language found!" );
 260                 continue;
 261             }
 262
 263                         // language not yet added?
 264                         if (aNewStyleLanguages.count( nLang ) == 0)
 265                                 rNewStyleDics.push_back( *aIt2 );
 266                 }
 267                 else
 268                 {
 269                         DBG_ERROR( "old style dictionary with no language found!" );
 270                 }
 271         }
 272 }
 273
 274 //////////////////////////////////////////////////////////////////////
 275