lingucomponent/source/languageguessing/guesslang.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 #include <iostream>
  32
  33 #include <tools/debug.hxx>
  34
  35 #include <sal/config.h>
  36 #include <cppuhelper/factory.hxx>
  37 #include <cppuhelper/implementationentry.hxx>
  38 #include <cppuhelper/implbase2.hxx>
  39 #include <tools/string.hxx>
  40
  41 #include <simpleguesser.hxx>
  42 #include <guess.hxx>
  43
  44 //#include <cppuhelper/queryinterface.hxx> // helper for queryInterface() impl
  45
  46 //#include <com/sun/star/lang/XMultiServiceFactory.hpp>
  47 #include <com/sun/star/registry/XRegistryKey.hpp>
  48 #include <com/sun/star/lang/XServiceInfo.hpp>
  49 #include <com/sun/star/linguistic2/XLanguageGuessing.hpp>
  50 #include <unotools/pathoptions.hxx>
  51 #include <unotools/localfilehelper.hxx>
  52 #include <osl/thread.h>
  53
  54 using namespace ::rtl;
  55 using namespace ::osl;
  56 using namespace ::cppu;
  57 using namespace ::com::sun::star;
  58 using namespace ::com::sun::star::uno;
  59 using namespace ::com::sun::star::lang;
  60 using namespace ::com::sun::star::linguistic2;
  61
  62 namespace css = ::com::sun::star;
  63
  64 //==================================================================================================
  65
  66 #define A2OU(x) ::rtl::OUString::createFromAscii( x )
  67
  68 #define SERVICENAME     "com.sun.star.linguistic2.LanguageGuessing"
  69
  70 #define IMPLNAME        "com.sun.star.lingu2.LanguageGuessing"
  71
  72 static Sequence< OUString > getSupportedServiceNames_LangGuess_Impl()
  73 {
  74     Sequence<OUString> names(1);
  75     names[0] = A2OU( SERVICENAME );
  76     return names;
  77 }
  78
  79 static OUString getImplementationName_LangGuess_Impl()
  80 {
  81     return A2OU( IMPLNAME );
  82 }
  83
  84 static osl::Mutex &  GetLangGuessMutex()
  85 {
  86     static osl::Mutex aMutex;
  87     return aMutex;
  88 }
  89
  90
  91 class LangGuess_Impl :
  92     public ::cppu::WeakImplHelper2<
  93         XLanguageGuessing,
  94         XServiceInfo >
  95 {
  96     SimpleGuesser   m_aGuesser;
  97     bool            m_bInitialized;
  98     css::uno::Reference< css::uno::XComponentContext >  m_xContext;
  99
 100     LangGuess_Impl( const LangGuess_Impl & ); // not defined
 101     LangGuess_Impl & operator =( const LangGuess_Impl & ); // not defined
 102
 103     virtual ~LangGuess_Impl() {}
 104     void    EnsureInitialized();
 105
 106 public:
 107     explicit LangGuess_Impl(css::uno::Reference< css::uno::XComponentContext > const & rxContext);
 108
 109     // XServiceInfo implementation
 110     virtual OUString SAL_CALL getImplementationName(  ) throw(RuntimeException);
 111     virtual sal_Bool SAL_CALL supportsService( const OUString& ServiceName ) throw(RuntimeException);
 112     virtual Sequence< OUString > SAL_CALL getSupportedServiceNames(  ) throw(RuntimeException);
 113     static Sequence< OUString > SAL_CALL getSupportedServiceNames_Static(  );
 114
 115     // XLanguageGuessing implementation
 116     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);
 117     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);
 118     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);
 119     virtual ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale > SAL_CALL getAvailableLanguages(  ) throw (::com::sun::star::uno::RuntimeException);
 120     virtual ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale > SAL_CALL getEnabledLanguages(  ) throw (::com::sun::star::uno::RuntimeException);
 121     virtual ::com::sun::star::uno::Sequence< ::com::sun::star::lang::Locale > SAL_CALL getDisabledLanguages(  ) throw (::com::sun::star::uno::RuntimeException);
 122
 123     // implementation specific
 124     void SetFingerPrintsDB( const rtl::OUString &fileName ) throw (RuntimeException);
 125
 126     static const OUString & SAL_CALL getImplementationName_Static() throw();
 127
 128 };
 129
 130 //*************************************************************************
 131
 132 LangGuess_Impl::LangGuess_Impl(css::uno::Reference< css::uno::XComponentContext > const & rxContext) :
 133     m_bInitialized( false ),
 134     m_xContext( rxContext )
 135 {
 136 }
 137
 138 //*************************************************************************
 139
 140 void LangGuess_Impl::EnsureInitialized()
 141 {
 142     if (!m_bInitialized)
 143     {
 144         // set this to true at the very start to prevent loops because of
 145         // implicitly called functions below
 146         m_bInitialized = true;
 147
 148         // set default fingerprint path to where those get installed
 149         String aPhysPath;
 150         String aURL( SvtPathOptions().GetFingerprintPath() );
 151         utl::LocalFileHelper::ConvertURLToPhysicalName( aURL, aPhysPath );
 152 #ifdef WNT
 153             aPhysPath += '\\';
 154 #else
 155             aPhysPath += '/';
 156 #endif
 157
 158         SetFingerPrintsDB( aPhysPath );
 159
 160         //
 161         // disable currently not functional languages...
 162         //
 163         struct LangCountry
 164         {
 165             const char *pLang;
 166             const char *pCountry;
 167         };
 168         LangCountry aDisable[] =
 169         {
 170             {"gv", ""}, {"sco", ""},                            // no lang-id available yet...
 171 //            {"hy", ""}, {"drt", ""},                          // 0 bytes fingerprints...
 172             {"zh", "CN"}, {"zh", "TW"}, {"ja", ""}, {"ko", ""}, // not yet correct functional...
 173             {"ka", ""}, {"hi", ""}, {"mr", ""}, {"ne", ""},
 174             {"sa", ""}, {"ta", ""}, {"th", ""},
 175             {"qu", ""}, {"yi", ""}
 176         };
 177         sal_Int32 nNum = sizeof(aDisable) / sizeof(aDisable[0]);
 178         Sequence< Locale > aDisableSeq( nNum );
 179         Locale *pDisableSeq = aDisableSeq.getArray();
 180         for (sal_Int32 i = 0;  i < nNum;  ++i)
 181         {
 182             Locale aLocale;
 183             aLocale.Language = OUString::createFromAscii( aDisable[i].pLang );
 184             aLocale.Country  = OUString::createFromAscii( aDisable[i].pCountry );
 185             pDisableSeq[i] = aLocale;
 186         }
 187         disableLanguages( aDisableSeq );
 188         DBG_ASSERT( nNum == getDisabledLanguages().getLength(), "size mismatch" );
 189     }
 190 }
 191
 192 //*************************************************************************
 193
 194 /* TL: currently not part of the API
 195 Sequence< com::sun::star::lang::Locale > SAL_CALL LangGuess_Impl::guessLanguages(
 196         const rtl::OUString   &rText,
 197         sal_Int32       nStartPos,
 198         sal_Int32       nLen )
 199     throw (RuntimeException)
 200 {
 201     Sequence< com::sun::star::lang::Locale > aRes;
 202
 203     OString o = OUStringToOString( rText, RTL_TEXTENCODING_UTF8 );
 204     vector<Guess> gs = m_aGuesser.GuessLanguage(o.pData->buffer);
 205
 206     aRes.realloc(gs.size());
 207
 208     com::sun::star::lang::Locale *pRes = aRes.getArray();
 209
 210 #ifdef DEBUG
 211     std::cout << " We have " << gs.size() << " candidates" << std::endl;
 212 #endif
 213
 214     for(int i = 0; i < gs.size() ; i++ ){
 215         com::sun::star::lang::Locale current_aRes;
 216
 217         current_aRes.Language   = A2OU( gs[i].getLanguage().c_str() );
 218         current_aRes.Country    = A2OU( gs[i].getCountry().c_str() );
 219
 220         pRes[i] = current_aRes;
 221     }
 222
 223     return aRes;
 224 }
 225 */
 226 //*************************************************************************
 227
 228 Locale SAL_CALL LangGuess_Impl::guessPrimaryLanguage(
 229         const ::rtl::OUString& rText,
 230         ::sal_Int32 nStartPos,
 231         ::sal_Int32 nLen )
 232     throw (lang::IllegalArgumentException, uno::RuntimeException)
 233 {
 234     osl::MutexGuard aGuard( GetLangGuessMutex() );
 235
 236     EnsureInitialized();
 237
 238     lang::Locale aRes;
 239     if (nStartPos >=0 && nLen >= 0 && nStartPos + nLen <= rText.getLength())
 240     {
 241         OString o( OUStringToOString( rText.copy(nStartPos, nLen), RTL_TEXTENCODING_UTF8 ) );
 242         Guess g = m_aGuesser.GuessPrimaryLanguage((char*)o.getStr());
 243         aRes.Language   = OUString::createFromAscii(g.GetLanguage().c_str());
 244         aRes.Country    = OUString::createFromAscii(g.GetCountry().c_str());
 245     }
 246     else
 247         throw lang::IllegalArgumentException();
 248
 249     return aRes;
 250 }
 251
 252 //*************************************************************************
 253 #define DEFAULT_CONF_FILE_NAME "fpdb.conf"
 254
 255 void LangGuess_Impl::SetFingerPrintsDB(
 256         const rtl::OUString &filePath )
 257     throw (RuntimeException)
 258 {
 259     //! text encoding for file name / path needs to be in the same encoding the OS uses
 260     OString path = OUStringToOString( filePath, osl_getThreadTextEncoding() );
 261     OString conf_file_name( DEFAULT_CONF_FILE_NAME );
 262     OString conf_file_path(path);
 263     conf_file_path += conf_file_name;
 264
 265     //cout << "Conf file : " << conf_file_path.getStr() << " directory : " << path.getStr() << endl;
 266
 267     m_aGuesser.SetDBPath((const char*)conf_file_path.getStr(), (const char*)path.getStr());
 268 }
 269
 270 //*************************************************************************
 271 uno::Sequence< Locale > SAL_CALL LangGuess_Impl::getAvailableLanguages(  )
 272         throw (uno::RuntimeException)
 273 {
 274     osl::MutexGuard aGuard( GetLangGuessMutex() );
 275
 276     EnsureInitialized();
 277
 278     Sequence< com::sun::star::lang::Locale > aRes;
 279     vector<Guess> gs = m_aGuesser.GetAllManagedLanguages();
 280     aRes.realloc(gs.size());
 281
 282     com::sun::star::lang::Locale *pRes = aRes.getArray();
 283
 284     for(size_t i = 0; i < gs.size() ; i++ ){
 285         com::sun::star::lang::Locale current_aRes;
 286         current_aRes.Language   = A2OU( gs[i].GetLanguage().c_str() );
 287         current_aRes.Country    = A2OU( gs[i].GetCountry().c_str() );
 288         pRes[i] = current_aRes;
 289     }
 290
 291     return aRes;
 292 }
 293
 294 //*************************************************************************
 295 uno::Sequence< Locale > SAL_CALL LangGuess_Impl::getEnabledLanguages(  )
 296         throw (uno::RuntimeException)
 297 {
 298     osl::MutexGuard aGuard( GetLangGuessMutex() );
 299
 300     EnsureInitialized();
 301
 302     Sequence< com::sun::star::lang::Locale > aRes;
 303     vector<Guess> gs = m_aGuesser.GetAvailableLanguages();
 304     aRes.realloc(gs.size());
 305
 306     com::sun::star::lang::Locale *pRes = aRes.getArray();
 307
 308     for(size_t i = 0; i < gs.size() ; i++ ){
 309         com::sun::star::lang::Locale current_aRes;
 310         current_aRes.Language   = A2OU( gs[i].GetLanguage().c_str() );
 311         current_aRes.Country    = A2OU( gs[i].GetCountry().c_str() );
 312         pRes[i] = current_aRes;
 313     }
 314
 315     return aRes;
 316 }
 317
 318 //*************************************************************************
 319 uno::Sequence< Locale > SAL_CALL LangGuess_Impl::getDisabledLanguages(  )
 320         throw (uno::RuntimeException)
 321 {
 322     osl::MutexGuard aGuard( GetLangGuessMutex() );
 323
 324     EnsureInitialized();
 325
 326     Sequence< com::sun::star::lang::Locale > aRes;
 327     vector<Guess> gs = m_aGuesser.GetUnavailableLanguages();
 328     aRes.realloc(gs.size());
 329
 330     com::sun::star::lang::Locale *pRes = aRes.getArray();
 331
 332     for(size_t i = 0; i < gs.size() ; i++ ){
 333         com::sun::star::lang::Locale current_aRes;
 334         current_aRes.Language   = A2OU( gs[i].GetLanguage().c_str() );
 335         current_aRes.Country    = A2OU( gs[i].GetCountry().c_str() );
 336         pRes[i] = current_aRes;
 337     }
 338
 339     return aRes;
 340 }
 341
 342 //*************************************************************************
 343 void SAL_CALL LangGuess_Impl::disableLanguages(
 344         const uno::Sequence< Locale >& rLanguages )
 345     throw (lang::IllegalArgumentException, uno::RuntimeException)
 346 {
 347     osl::MutexGuard aGuard( GetLangGuessMutex() );
 348
 349     EnsureInitialized();
 350
 351     sal_Int32 nLanguages = rLanguages.getLength();
 352     const Locale *pLanguages = rLanguages.getConstArray();
 353
 354     for (sal_Int32 i = 0;  i < nLanguages;  ++i)
 355     {
 356         string language;
 357
 358         OString l = OUStringToOString( pLanguages[i].Language, RTL_TEXTENCODING_ASCII_US );
 359         OString c = OUStringToOString( pLanguages[i].Country, RTL_TEXTENCODING_ASCII_US );
 360
 361         language += l.getStr();
 362         language += "-";
 363         language += c.getStr();
 364         m_aGuesser.DisableLanguage(language);
 365     }
 366 }
 367
 368 //*************************************************************************
 369 void SAL_CALL LangGuess_Impl::enableLanguages(
 370         const uno::Sequence< Locale >& rLanguages )
 371     throw (lang::IllegalArgumentException, uno::RuntimeException)
 372 {
 373     osl::MutexGuard aGuard( GetLangGuessMutex() );
 374
 375     EnsureInitialized();
 376
 377     sal_Int32 nLanguages = rLanguages.getLength();
 378     const Locale *pLanguages = rLanguages.getConstArray();
 379
 380     for (sal_Int32 i = 0;  i < nLanguages;  ++i)
 381     {
 382         string language;
 383
 384         OString l = OUStringToOString( pLanguages[i].Language, RTL_TEXTENCODING_ASCII_US );
 385         OString c = OUStringToOString( pLanguages[i].Country, RTL_TEXTENCODING_ASCII_US );
 386
 387         language += l.getStr();
 388         language += "-";
 389         language += c.getStr();
 390         m_aGuesser.EnableLanguage(language);
 391     }
 392 }
 393
 394 //*************************************************************************
 395 OUString SAL_CALL LangGuess_Impl::getImplementationName(  )
 396     throw(RuntimeException)
 397 {
 398     osl::MutexGuard aGuard( GetLangGuessMutex() );
 399     return A2OU( IMPLNAME );
 400 }
 401
 402 //*************************************************************************
 403 sal_Bool SAL_CALL LangGuess_Impl::supportsService( const OUString& ServiceName )
 404     throw(RuntimeException)
 405 {
 406     osl::MutexGuard aGuard( GetLangGuessMutex() );
 407     Sequence< OUString > aSNL = getSupportedServiceNames();
 408     const OUString * pArray = aSNL.getArray();
 409     for( sal_Int32 i = 0; i < aSNL.getLength(); i++ )
 410         if( pArray[i] == ServiceName )
 411             return sal_True;
 412     return sal_False;
 413 }
 414
 415 //*************************************************************************
 416 Sequence<OUString> SAL_CALL LangGuess_Impl::getSupportedServiceNames(  )
 417     throw(RuntimeException)
 418 {
 419     osl::MutexGuard aGuard( GetLangGuessMutex() );
 420     return getSupportedServiceNames_Static();
 421 }
 422
 423 //*************************************************************************
 424 Sequence<OUString> SAL_CALL LangGuess_Impl::getSupportedServiceNames_Static(  )
 425 {
 426     OUString aName( A2OU( SERVICENAME ) );
 427     return Sequence< OUString >( &aName, 1 );
 428 }
 429
 430 //*************************************************************************
 431
 432
 433 /**
 434  * Function to create a new component instance; is needed by factory helper implementation.
 435  * @param xMgr service manager to if the components needs other component instances
 436  */
 437 Reference< XInterface > SAL_CALL LangGuess_Impl_create(
 438     Reference< XComponentContext > const & xContext )
 439     SAL_THROW( () )
 440 {
 441     return static_cast< ::cppu::OWeakObject * >( new LangGuess_Impl(xContext) );
 442 }
 443
 444 //##################################################################################################
 445 //#### EXPORTED ### functions to allow for registration and creation of the UNO component
 446 //##################################################################################################
 447
 448 static struct ::cppu::ImplementationEntry s_component_entries [] =
 449 {
 450     {
 451         LangGuess_Impl_create, getImplementationName_LangGuess_Impl,
 452         getSupportedServiceNames_LangGuess_Impl,
 453         ::cppu::createSingleComponentFactory,
 454         0, 0
 455     },
 456     { 0, 0, 0, 0, 0, 0 }
 457 };
 458
 459 extern "C"
 460 {
 461
 462 void SAL_CALL component_getImplementationEnvironment(
 463     sal_Char const ** ppEnvTypeName, uno_Environment ** /*ppEnv*/ )
 464 {
 465     *ppEnvTypeName = CPPU_CURRENT_LANGUAGE_BINDING_NAME;
 466 }
 467
 468 void * SAL_CALL component_getFactory(
 469     sal_Char const * implName, lang::XMultiServiceFactory * xMgr,
 470     registry::XRegistryKey * xRegistry )
 471 {
 472     return ::cppu::component_getFactoryHelper(
 473         implName, xMgr, xRegistry, s_component_entries );
 474 }
 475
 476 }
 477