lingucomponent/source/languageguessing/guesslang.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: guesslang.cxx,v $
  10  * $Revision: 1.7 $
  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 #include <iostream>
  35
  36 #include <tools/debug.hxx>
  37
  38 #include <sal/config.h>
  39 #include <cppuhelper/factory.hxx>
  40 #include <cppuhelper/implementationentry.hxx>
  41 #include <cppuhelper/implbase2.hxx>
  42 #include <tools/string.hxx>
  43
  44 #include <simpleguesser.hxx>
  45 #include <guess.hxx>
  46
  47 //#include <cppuhelper/queryinterface.hxx> // helper for queryInterface() impl
  48
  49 //#include <com/sun/star/lang/XMultiServiceFactory.hpp>
  50 #include <com/sun/star/registry/XRegistryKey.hpp>
  51 #include <com/sun/star/lang/XServiceInfo.hpp>
  52 #include <com/sun/star/linguistic2/XLanguageGuessing.hpp>
  53 #include <svtools/pathoptions.hxx>
  54 #include <unotools/localfilehelper.hxx>
  55 #include <osl/thread.h>
  56
  57 using namespace ::rtl;
  58 using namespace ::osl;
  59 using namespace ::cppu;
  60 using namespace ::com::sun::star;
  61 using namespace ::com::sun::star::uno;
  62 using namespace ::com::sun::star::lang;
  63 using namespace ::com::sun::star::linguistic2;
  64
  65 namespace css = ::com::sun::star;
  66
  67 //==================================================================================================
  68
  69 #define A2OU(x) ::rtl::OUString::createFromAscii( x )
  70
  71 #define SERVICENAME     "com.sun.star.linguistic2.LanguageGuessing"
  72
  73 #define IMPLNAME        "com.sun.star.lingu2.LanguageGuessing"
  74
  75 static Sequence< OUString > getSupportedServiceNames_LangGuess_Impl()
  76 {
  77     Sequence<OUString> names(1);
  78     names[0] = A2OU( SERVICENAME );
  79     return names;
  80 }
  81
  82 static OUString getImplementationName_LangGuess_Impl()
  83 {
  84     return A2OU( IMPLNAME );
  85 }
  86
  87 static osl::Mutex &  GetLangGuessMutex()
  88 {
  89     static osl::Mutex aMutex;
  90     return aMutex;
  91 }
  92
  93
  94 class LangGuess_Impl :
  95     public ::cppu::WeakImplHelper2<
  96         XLanguageGuessing,
  97         XServiceInfo >
  98 {
  99     SimpleGuesser   m_aGuesser;
 100     bool            m_bInitialized;
 101     css::uno::Reference< css::uno::XComponentContext >  m_xContext;
 102
 103     LangGuess_Impl( const LangGuess_Impl & ); // not defined
 104     LangGuess_Impl & operator =( const LangGuess_Impl & ); // not defined
 105
 106     virtual ~LangGuess_Impl() {}
 107     void    EnsureInitialized();
 108
 109 public:
 110     explicit LangGuess_Impl(css::uno::Reference< css::uno::XComponentContext > const & rxContext);
 111
 112     // XServiceInfo     implementation
 113     virtual OUString SAL_CALL getImplementationName(  ) throw(RuntimeException);
 114     virtual sal_Bool SAL_CALL supportsService( const OUString& ServiceName ) throw(RuntimeException);
 115     virtual Sequence< OUString > SAL_CALL getSupportedServiceNames(  ) throw(RuntimeException);
 116     static Sequence< OUString > SAL_CALL getSupportedServiceNames_Static(  );
 117
 118     // XLanguageGuessing implementation
 119     virtual ::com::sun::star::lang::Locale SAL_CALL guessPrimaryLanguage( const ::rtl::OUString& aText, ::sal_Int32 nStartPos, ::sal_Int32 nLen ) throw (::com::sun::star::lang::IllegalArgumentException, ::com::sun::star::uno::RuntimeException);
 120     virtual void SAL_CALL disableLanguages( const ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale >& aLanguages ) throw (::com::sun::star::lang::IllegalArgumentException, ::com::sun::star::uno::RuntimeException);
 121     virtual void SAL_CALL enableLanguages( const ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale >& aLanguages ) throw (::com::sun::star::lang::IllegalArgumentException, ::com::sun::star::uno::RuntimeException);
 122     virtual ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale > SAL_CALL getAvailableLanguages(  ) throw (::com::sun::star::uno::RuntimeException);
 123     virtual ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale > SAL_CALL getEnabledLanguages(  ) throw (::com::sun::star::uno::RuntimeException);
 124     virtual ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale > SAL_CALL getDisabledLanguages(  ) throw (::com::sun::star::uno::RuntimeException);
 125
 126     // implementation specific
 127     void SetFingerPrintsDB( const rtl::OUString &fileName ) throw (RuntimeException);
 128
 129     static const OUString & SAL_CALL getImplementationName_Static() throw();
 130
 131 };
 132
 133 //*************************************************************************
 134
 135 LangGuess_Impl::LangGuess_Impl(css::uno::Reference< css::uno::XComponentContext > const & rxContext) :
 136     m_bInitialized( false ),
 137     m_xContext( rxContext )
 138 {
 139 }
 140
 141 //*************************************************************************
 142
 143 void LangGuess_Impl::EnsureInitialized()
 144 {
 145     if (!m_bInitialized)
 146     {
 147         // set this to true at the very start to prevent loops because of
 148         // implicitly called functions below
 149         m_bInitialized = true;
 150
 151         // set default fingerprint path to where those get installed
 152         String aPhysPath;
 153         String aURL( SvtPathOptions().GetFingerprintPath() );
 154         utl::LocalFileHelper::ConvertURLToPhysicalName( aURL, aPhysPath );
 155 #ifdef WNT
 156             aPhysPath += '\\';
 157 #else
 158             aPhysPath += '/';
 159 #endif
 160
 161         SetFingerPrintsDB( aPhysPath );
 162
 163         //
 164         // disable currently not functional languages...
 165         //
 166         struct LangCountry
 167         {
 168             const char *pLang;
 169             const char *pCountry;
 170         };
 171         LangCountry aDisable[] =
 172         {
 173             {"gv", ""}, {"sco", ""},                            // no lang-id available yet...
 174 //            {"hy", ""}, {"drt", ""},                          // 0 bytes fingerprints...
 175             {"zh", "CN"}, {"zh", "TW"}, {"ja", ""}, {"ko", ""}, // not yet correct functional...
 176             {"ka", ""}, {"hi", ""}, {"mr", ""}, {"ne", ""},
 177             {"sa", ""}, {"ta", ""}, {"th", ""},
 178             {"qu", ""}, {"yi", ""}
 179         };
 180         sal_Int32 nNum = sizeof(aDisable) / sizeof(aDisable[0]);
 181         Sequence< Locale > aDisableSeq( nNum );
 182         Locale *pDisableSeq = aDisableSeq.getArray();
 183         for (sal_Int32 i = 0;  i < nNum;  ++i)
 184         {
 185             Locale aLocale;
 186             aLocale.Language = OUString::createFromAscii( aDisable[i].pLang );
 187             aLocale.Country  = OUString::createFromAscii( aDisable[i].pCountry );
 188             pDisableSeq[i] = aLocale;
 189         }
 190         disableLanguages( aDisableSeq );
 191         DBG_ASSERT( nNum == getDisabledLanguages().getLength(), "size mismatch" );
 192     }
 193 }
 194
 195 //*************************************************************************
 196
 197 /* TL: currently not part of the API
 198 Sequence< com::sun::star::lang::Locale > SAL_CALL LangGuess_Impl::guessLanguages(
 199         const rtl::OUString   &rText,
 200         sal_Int32       nStartPos,
 201         sal_Int32       nLen )
 202     throw (RuntimeException)
 203 {
 204     Sequence< com::sun::star::lang::Locale > aRes;
 205
 206     OString o = OUStringToOString( rText, RTL_TEXTENCODING_UTF8 );
 207     vector<Guess> gs = m_aGuesser.GuessLanguage(o.pData->buffer);
 208
 209     aRes.realloc(gs.size());
 210
 211     com::sun::star::lang::Locale *pRes = aRes.getArray();
 212
 213 #ifdef DEBUG
 214     std::cout << " We have " << gs.size() << " candidates" << std::endl;
 215 #endif
 216
 217     for(int i = 0; i < gs.size() ; i++ ){
 218         com::sun::star::lang::Locale current_aRes;
 219
 220         current_aRes.Language   = A2OU( gs[i].getLanguage().c_str() );
 221         current_aRes.Country    = A2OU( gs[i].getCountry().c_str() );
 222
 223         pRes[i] = current_aRes;
 224     }
 225
 226     return aRes;
 227 }
 228 */
 229 //*************************************************************************
 230
 231 Locale SAL_CALL LangGuess_Impl::guessPrimaryLanguage(
 232         const ::rtl::OUString& rText,
 233         ::sal_Int32 nStartPos,
 234         ::sal_Int32 nLen )
 235     throw (lang::IllegalArgumentException, uno::RuntimeException)
 236 {
 237     osl::MutexGuard aGuard( GetLangGuessMutex() );
 238
 239     EnsureInitialized();
 240
 241     lang::Locale aRes;
 242     if (nStartPos >=0 && nLen >= 0 && nStartPos + nLen <= rText.getLength())
 243     {
 244         OString o( OUStringToOString( rText.copy(nStartPos, nLen), RTL_TEXTENCODING_UTF8 ) );
 245         Guess g = m_aGuesser.GuessPrimaryLanguage((char*)o.getStr());
 246         aRes.Language   = OUString::createFromAscii(g.GetLanguage().c_str());
 247         aRes.Country    = OUString::createFromAscii(g.GetCountry().c_str());
 248     }
 249     else
 250         throw lang::IllegalArgumentException();
 251
 252     return aRes;
 253 }
 254
 255 //*************************************************************************
 256 #define DEFAULT_CONF_FILE_NAME "fpdb.conf"
 257
 258 void LangGuess_Impl::SetFingerPrintsDB(
 259         const rtl::OUString &filePath )
 260     throw (RuntimeException)
 261 {
 262     //! text encoding for file name / path needs to be in the same encoding the OS uses
 263     OString path = OUStringToOString( filePath, osl_getThreadTextEncoding() );
 264     OString conf_file_name( DEFAULT_CONF_FILE_NAME );
 265     OString conf_file_path(path);
 266     conf_file_path += conf_file_name;
 267
 268     //cout << "Conf file : " << conf_file_path.getStr() << " directory : " << path.getStr() << endl;
 269
 270     m_aGuesser.SetDBPath((const char*)conf_file_path.getStr(), (const char*)path.getStr());
 271 }
 272
 273 //*************************************************************************
 274 uno::Sequence< Locale > SAL_CALL LangGuess_Impl::getAvailableLanguages(  )
 275         throw (uno::RuntimeException)
 276 {
 277     osl::MutexGuard aGuard( GetLangGuessMutex() );
 278
 279     EnsureInitialized();
 280
 281     Sequence< com::sun::star::lang::Locale > aRes;
 282     vector<Guess> gs = m_aGuesser.GetAllManagedLanguages();
 283     aRes.realloc(gs.size());
 284
 285     com::sun::star::lang::Locale *pRes = aRes.getArray();
 286
 287     for(size_t i = 0; i < gs.size() ; i++ ){
 288         com::sun::star::lang::Locale current_aRes;
 289         current_aRes.Language   = A2OU( gs[i].GetLanguage().c_str() );
 290         current_aRes.Country    = A2OU( gs[i].GetCountry().c_str() );
 291         pRes[i] = current_aRes;
 292     }
 293
 294     return aRes;
 295 }
 296
 297 //*************************************************************************
 298 uno::Sequence< Locale > SAL_CALL LangGuess_Impl::getEnabledLanguages(  )
 299         throw (uno::RuntimeException)
 300 {
 301     osl::MutexGuard aGuard( GetLangGuessMutex() );
 302
 303     EnsureInitialized();
 304
 305     Sequence< com::sun::star::lang::Locale > aRes;
 306     vector<Guess> gs = m_aGuesser.GetAvailableLanguages();
 307     aRes.realloc(gs.size());
 308
 309     com::sun::star::lang::Locale *pRes = aRes.getArray();
 310
 311     for(size_t i = 0; i < gs.size() ; i++ ){
 312         com::sun::star::lang::Locale current_aRes;
 313         current_aRes.Language   = A2OU( gs[i].GetLanguage().c_str() );
 314         current_aRes.Country    = A2OU( gs[i].GetCountry().c_str() );
 315         pRes[i] = current_aRes;
 316     }
 317
 318     return aRes;
 319 }
 320
 321 //*************************************************************************
 322 uno::Sequence< Locale > SAL_CALL LangGuess_Impl::getDisabledLanguages(  )
 323         throw (uno::RuntimeException)
 324 {
 325     osl::MutexGuard aGuard( GetLangGuessMutex() );
 326
 327     EnsureInitialized();
 328
 329     Sequence< com::sun::star::lang::Locale > aRes;
 330     vector<Guess> gs = m_aGuesser.GetUnavailableLanguages();
 331     aRes.realloc(gs.size());
 332
 333     com::sun::star::lang::Locale *pRes = aRes.getArray();
 334
 335     for(size_t i = 0; i < gs.size() ; i++ ){
 336         com::sun::star::lang::Locale current_aRes;
 337         current_aRes.Language   = A2OU( gs[i].GetLanguage().c_str() );
 338         current_aRes.Country    = A2OU( gs[i].GetCountry().c_str() );
 339         pRes[i] = current_aRes;
 340     }
 341
 342     return aRes;
 343 }
 344
 345 //*************************************************************************
 346 void SAL_CALL LangGuess_Impl::disableLanguages(
 347         const uno::Sequence< Locale >& rLanguages )
 348     throw (lang::IllegalArgumentException, uno::RuntimeException)
 349 {
 350     osl::MutexGuard aGuard( GetLangGuessMutex() );
 351
 352     EnsureInitialized();
 353
 354     sal_Int32 nLanguages = rLanguages.getLength();
 355     const Locale *pLanguages = rLanguages.getConstArray();
 356
 357     for (sal_Int32 i = 0;  i < nLanguages;  ++i)
 358     {
 359         string language;
 360
 361         OString l = OUStringToOString( pLanguages[i].Language, RTL_TEXTENCODING_ASCII_US );
 362         OString c = OUStringToOString( pLanguages[i].Country, RTL_TEXTENCODING_ASCII_US );
 363
 364         language += l.getStr();
 365         language += "-";
 366         language += c.getStr();
 367         m_aGuesser.DisableLanguage(language);
 368     }
 369 }
 370
 371 //*************************************************************************
 372 void SAL_CALL LangGuess_Impl::enableLanguages(
 373         const uno::Sequence< Locale >& rLanguages )
 374     throw (lang::IllegalArgumentException, uno::RuntimeException)
 375 {
 376     osl::MutexGuard aGuard( GetLangGuessMutex() );
 377
 378     EnsureInitialized();
 379
 380     sal_Int32 nLanguages = rLanguages.getLength();
 381     const Locale *pLanguages = rLanguages.getConstArray();
 382
 383     for (sal_Int32 i = 0;  i < nLanguages;  ++i)
 384     {
 385         string language;
 386
 387         OString l = OUStringToOString( pLanguages[i].Language, RTL_TEXTENCODING_ASCII_US );
 388         OString c = OUStringToOString( pLanguages[i].Country, RTL_TEXTENCODING_ASCII_US );
 389
 390         language += l.getStr();
 391         language += "-";
 392         language += c.getStr();
 393         m_aGuesser.EnableLanguage(language);
 394     }
 395 }
 396
 397 //*************************************************************************
 398 OUString SAL_CALL LangGuess_Impl::getImplementationName(  )
 399     throw(RuntimeException)
 400 {
 401     osl::MutexGuard aGuard( GetLangGuessMutex() );
 402     return A2OU( IMPLNAME );
 403 }
 404
 405 //*************************************************************************
 406 sal_Bool SAL_CALL LangGuess_Impl::supportsService( const OUString& ServiceName )
 407     throw(RuntimeException)
 408 {
 409     osl::MutexGuard aGuard( GetLangGuessMutex() );
 410     Sequence< OUString > aSNL = getSupportedServiceNames();
 411     const OUString * pArray = aSNL.getArray();
 412     for( sal_Int32 i = 0; i < aSNL.getLength(); i++ )
 413         if( pArray[i] == ServiceName )
 414             return sal_True;
 415     return sal_False;
 416 }
 417
 418 //*************************************************************************
 419 Sequence<OUString> SAL_CALL LangGuess_Impl::getSupportedServiceNames(  )
 420     throw(RuntimeException)
 421 {
 422     osl::MutexGuard aGuard( GetLangGuessMutex() );
 423     return getSupportedServiceNames_Static();
 424 }
 425
 426 //*************************************************************************
 427 Sequence<OUString> SAL_CALL LangGuess_Impl::getSupportedServiceNames_Static(  )
 428 {
 429     OUString aName( A2OU( SERVICENAME ) );
 430     return Sequence< OUString >( &aName, 1 );
 431 }
 432
 433 //*************************************************************************
 434
 435
 436 /**
 437  * Function to create a new component instance; is needed by factory helper implementation.
 438  * @param xMgr service manager to if the components needs other component instances
 439  */
 440 Reference< XInterface > SAL_CALL LangGuess_Impl_create(
 441     Reference< XComponentContext > const & xContext )
 442     SAL_THROW( () )
 443 {
 444     return static_cast< ::cppu::OWeakObject * >( new LangGuess_Impl(xContext) );
 445 }
 446
 447 //##################################################################################################
 448 //#### EXPORTED ### functions to allow for registration and creation of the UNO component
 449 //##################################################################################################
 450
 451 static struct ::cppu::ImplementationEntry s_component_entries [] =
 452 {
 453     {
 454         LangGuess_Impl_create, getImplementationName_LangGuess_Impl,
 455         getSupportedServiceNames_LangGuess_Impl,
 456         ::cppu::createSingleComponentFactory,
 457         0, 0
 458     },
 459     { 0, 0, 0, 0, 0, 0 }
 460 };
 461
 462 extern "C"
 463 {
 464
 465 void SAL_CALL component_getImplementationEnvironment(
 466     sal_Char const ** ppEnvTypeName, uno_Environment ** /*ppEnv*/ )
 467 {
 468     *ppEnvTypeName = CPPU_CURRENT_LANGUAGE_BINDING_NAME;
 469 }
 470
 471 sal_Bool SAL_CALL component_writeInfo(
 472     XMultiServiceFactory * xMgr, registry::XRegistryKey * xRegistry )
 473 {
 474     return ::cppu::component_writeInfoHelper(
 475         xMgr, xRegistry, s_component_entries );
 476 }
 477
 478 void * SAL_CALL component_getFactory(
 479     sal_Char const * implName, lang::XMultiServiceFactory * xMgr,
 480     registry::XRegistryKey * xRegistry )
 481 {
 482     return ::cppu::component_getFactoryHelper(
 483         implName, xMgr, xRegistry, s_component_entries );
 484 }
 485
 486 }
 487