linguistic/source/convdicxml.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: convdicxml.cxx,v $
  10  * $Revision: 1.9 $
  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_linguistic.hxx"
  33 #include <tools/urlobj.hxx>
  34 #include <tools/debug.hxx>
  35 #include <tools/fsys.hxx>
  36 #include <tools/string.hxx>
  37 #include <i18npool/mslangid.hxx>
  38 #include <tools/stream.hxx>
  39 #include <sfx2/docfile.hxx>
  40 #include <osl/mutex.hxx>
  41 #include <unotools/processfactory.hxx>
  42 #include <ucbhelper/content.hxx>
  43
  44 #include <cppuhelper/factory.hxx>   // helper for factories
  45 #include <com/sun/star/linguistic2/XConversionDictionary.hpp>
  46 #include <com/sun/star/linguistic2/ConversionDictionaryType.hpp>
  47 #include <com/sun/star/linguistic2/ConversionPropertyType.hpp>
  48 #include <com/sun/star/util/XFlushable.hpp>
  49 #include <com/sun/star/lang/Locale.hpp>
  50 #include <com/sun/star/lang/EventObject.hpp>
  51 #ifndef _COM_SUN_STAR_UNO_REFERENCE_HPP_
  52 #include <com/sun/star/uno/Reference.h>
  53 #endif
  54 #include <com/sun/star/registry/XRegistryKey.hpp>
  55 #include <com/sun/star/util/XFlushListener.hpp>
  56 #include <com/sun/star/io/XActiveDataSource.hpp>
  57 #include <com/sun/star/document/XFilter.hpp>
  58 #include <com/sun/star/beans/PropertyValue.hpp>
  59 #include <xmloff/nmspmap.hxx>
  60 #include <xmloff/xmlnmspe.hxx>
  61 #include <unotools/streamwrap.hxx>
  62
  63 #include "convdic.hxx"
  64 #include "convdicxml.hxx"
  65 #include "misc.hxx"
  66 #include "defs.hxx"
  67
  68 using namespace std;
  69 using namespace utl;
  70 using namespace osl;
  71 using namespace rtl;
  72 using namespace com::sun::star;
  73 using namespace com::sun::star::lang;
  74 using namespace com::sun::star::uno;
  75 using namespace com::sun::star::linguistic2;
  76 using namespace linguistic;
  77
  78 #define XML_NAMESPACE_TCD_STRING        "http://openoffice.org/2003/text-conversion-dictionary"
  79 #define CONV_TYPE_HANGUL_HANJA          "Hangul / Hanja"
  80 #define CONV_TYPE_SCHINESE_TCHINESE     "Chinese simplified / Chinese traditional"
  81
  82 ///////////////////////////////////////////////////////////////////////////
  83
  84 static const OUString ConversionTypeToText( sal_Int16 nConversionType )
  85 {
  86     OUString aRes;
  87     if (nConversionType == ConversionDictionaryType::HANGUL_HANJA)
  88         aRes = A2OU( CONV_TYPE_HANGUL_HANJA );
  89     else if (nConversionType == ConversionDictionaryType::SCHINESE_TCHINESE)
  90         aRes = A2OU( CONV_TYPE_SCHINESE_TCHINESE );
  91     return aRes;
  92 }
  93
  94 static sal_Int16 GetConversionTypeFromText( const String &rText )
  95 {
  96     sal_Int16 nRes = -1;
  97     if (rText.EqualsAscii( CONV_TYPE_HANGUL_HANJA ))
  98         nRes = ConversionDictionaryType::HANGUL_HANJA;
  99     else if (rText.EqualsAscii( CONV_TYPE_SCHINESE_TCHINESE ))
 100         nRes = ConversionDictionaryType::SCHINESE_TCHINESE;
 101     return nRes;
 102 }
 103
 104 ///////////////////////////////////////////////////////////////////////////
 105
 106 class ConvDicXMLImportContext :
 107     public SvXMLImportContext
 108 {
 109 public:
 110     ConvDicXMLImportContext(
 111             ConvDicXMLImport &rImport,
 112             sal_uInt16 nPrfx, const OUString& rLName ) :
 113         SvXMLImportContext( rImport, nPrfx, rLName )
 114     {
 115     }
 116
 117     const ConvDicXMLImport & GetConvDicImport() const
 118     {
 119         return (const ConvDicXMLImport &) GetImport();
 120     }
 121
 122     ConvDicXMLImport & GetConvDicImport()
 123     {
 124         return (ConvDicXMLImport &) GetImport();
 125     }
 126
 127     // SvXMLImportContext
 128     virtual void Characters( const OUString &rChars );
 129     virtual SvXMLImportContext * CreateChildContext( sal_uInt16 nPrefix, const OUString& rLocalName, const uno::Reference< xml::sax::XAttributeList > &rxAttrList);
 130 };
 131
 132
 133 class ConvDicXMLDictionaryContext_Impl :
 134     public ConvDicXMLImportContext
 135 {
 136     INT16               nLanguage;
 137     sal_Int16   nConversionType;
 138
 139 public:
 140     ConvDicXMLDictionaryContext_Impl( ConvDicXMLImport &rImport,
 141             sal_uInt16 nPrefix, const OUString& rLName) :
 142         ConvDicXMLImportContext( rImport, nPrefix, rLName )
 143     {
 144         nLanguage = LANGUAGE_NONE;
 145         nConversionType = -1;
 146     }
 147
 148     // SvXMLImportContext
 149     virtual void StartElement( const ::com::sun::star::uno::Reference< ::com::sun::star::xml::sax::XAttributeList >& xAttrList );
 150     virtual SvXMLImportContext * CreateChildContext( sal_uInt16 nPrefix, const OUString& rLocalName, const uno::Reference< xml::sax::XAttributeList > &rxAttrList );
 151
 152     INT16               GetLanguage() const                     { return nLanguage; }
 153     sal_Int16   GetConversionType() const       { return nConversionType; }
 154 };
 155
 156
 157 class ConvDicXMLEntryTextContext_Impl :
 158     public ConvDicXMLImportContext
 159 {
 160     OUString    aLeftText;
 161     sal_Int16   nPropertyType;  // used for Chinese simplified/traditional conversion
 162     ConvDicXMLDictionaryContext_Impl    &rDicContext;
 163
 164 public:
 165     ConvDicXMLEntryTextContext_Impl(
 166             ConvDicXMLImport &rImport,
 167             sal_uInt16 nPrefix, const OUString& rLName,
 168             ConvDicXMLDictionaryContext_Impl &rParentContext ) :
 169         ConvDicXMLImportContext( rImport, nPrefix, rLName ),
 170         nPropertyType( ConversionPropertyType::NOT_DEFINED ),
 171         rDicContext( rParentContext )
 172     {
 173     }
 174
 175     // SvXMLImportContext
 176     virtual void StartElement( const ::com::sun::star::uno::Reference< ::com::sun::star::xml::sax::XAttributeList >& xAttrList );
 177     virtual SvXMLImportContext * CreateChildContext( sal_uInt16 nPrefix, const OUString& rLocalName, const uno::Reference< xml::sax::XAttributeList > &rxAttrList );
 178
 179     const OUString &    GetLeftText() const     { return aLeftText; }
 180     sal_Int16           GetPropertyType() const { return nPropertyType; }
 181     void                SetPropertyType( sal_Int16 nVal )   { nPropertyType = nVal; }
 182 };
 183
 184
 185 class ConvDicXMLRightTextContext_Impl :
 186     public ConvDicXMLImportContext
 187 {
 188     OUString aRightText;
 189     ConvDicXMLEntryTextContext_Impl &rEntryContext;
 190
 191 public:
 192     ConvDicXMLRightTextContext_Impl(
 193             ConvDicXMLImport &rImport,
 194             sal_uInt16 nPrefix, const OUString& rLName,
 195             ConvDicXMLEntryTextContext_Impl &rParentContext ) :
 196         ConvDicXMLImportContext( rImport, nPrefix, rLName ),
 197         rEntryContext( rParentContext )
 198     {
 199     }
 200
 201     // SvXMLImportContext
 202     virtual void EndElement();
 203     virtual SvXMLImportContext * CreateChildContext( sal_uInt16 nPrefix, const OUString& rLocalName, const uno::Reference< xml::sax::XAttributeList > &rxAttrList );
 204     virtual void Characters( const OUString &rChars );
 205
 206     const OUString &    GetRightText() const    { return aRightText; }
 207     const OUString &    GetLeftText() const             { return rEntryContext.GetLeftText(); }
 208     ConvDic *                   GetDic()                                { return GetConvDicImport().GetDic(); }
 209 };
 210
 211 ///////////////////////////////////////////////////////////////////////////
 212
 213 void ConvDicXMLImportContext::Characters(const OUString & /*rChars*/)
 214 {
 215     /*
 216     Whitespace occurring within the content of token elements is "trimmed"
 217     from the ends (i.e. all whitespace at the beginning and end of the
 218     content is removed), and "collapsed" internally (i.e. each sequence of
 219     1 or more whitespace characters is replaced with one blank character).
 220     */
 221     //collapsing not done yet!
 222
 223     // warning-free code: since the result is not used there is no need for trimming...
 224     //const OUString &rChars2 = rChars.trim();
 225 }
 226
 227 SvXMLImportContext * ConvDicXMLImportContext::CreateChildContext(
 228         sal_uInt16 nPrefix, const OUString& rLocalName,
 229         const uno::Reference< xml::sax::XAttributeList > & /*rxAttrList*/ )
 230 {
 231     SvXMLImportContext *pContext = 0;
 232     if (nPrefix == XML_NAMESPACE_TCD && rLocalName.equalsAscii( "text-conversion-dictionary" ))
 233         pContext = new ConvDicXMLDictionaryContext_Impl( GetConvDicImport(), nPrefix, rLocalName );
 234     else
 235         pContext = new SvXMLImportContext( GetImport(), nPrefix, rLocalName );
 236     return pContext;
 237 }
 238
 239 ////////////////////////////////////////
 240
 241 void ConvDicXMLDictionaryContext_Impl::StartElement(
 242     const uno::Reference< xml::sax::XAttributeList > &rxAttrList )
 243 {
 244     sal_Int16 nAttrCount = rxAttrList.is() ? rxAttrList->getLength() : 0;
 245     for (sal_Int16 i = 0;  i < nAttrCount;  ++i)
 246     {
 247         OUString aAttrName = rxAttrList->getNameByIndex(i);
 248         OUString aLocalName;
 249         sal_uInt16 nPrefix = GetImport().GetNamespaceMap().
 250                                     GetKeyByAttrName( aAttrName, &aLocalName );
 251         OUString aValue = rxAttrList->getValueByIndex(i);
 252
 253         if (nPrefix == XML_NAMESPACE_TCD && aLocalName.equalsAscii( "lang" ))
 254             nLanguage = MsLangId::convertIsoStringToLanguage( aValue );
 255         else if (nPrefix == XML_NAMESPACE_TCD && aLocalName.equalsAscii( "conversion-type" ))
 256             nConversionType = GetConversionTypeFromText( aValue );
 257     }
 258     GetConvDicImport().SetLanguage( nLanguage );
 259     GetConvDicImport().SetConversionType( nConversionType );
 260
 261     //!! hack to stop the parser from reading the rest of the file  !!
 262     //!! when only the header (language, conversion type) is needed !!
 263 //   if (GetConvDicImport().GetDic() == 0)
 264 //        throw uno::RuntimeException();
 265 }
 266
 267 SvXMLImportContext * ConvDicXMLDictionaryContext_Impl::CreateChildContext(
 268         sal_uInt16 nPrefix, const OUString& rLocalName,
 269         const uno::Reference< xml::sax::XAttributeList > & /*rxAttrList*/ )
 270 {
 271     SvXMLImportContext *pContext = 0;
 272     if (nPrefix == XML_NAMESPACE_TCD  &&  rLocalName.equalsAscii( "entry" ))
 273         pContext = new ConvDicXMLEntryTextContext_Impl( GetConvDicImport(), nPrefix, rLocalName, *this );
 274     else
 275         pContext = new SvXMLImportContext( GetImport(), nPrefix, rLocalName );
 276     return pContext;
 277 }
 278
 279 ////////////////////////////////////////
 280
 281 SvXMLImportContext * ConvDicXMLEntryTextContext_Impl::CreateChildContext(
 282         sal_uInt16 nPrefix, const OUString& rLocalName,
 283         const uno::Reference< xml::sax::XAttributeList > & /*rxAttrList*/ )
 284 {
 285     SvXMLImportContext *pContext = 0;
 286     if (nPrefix == XML_NAMESPACE_TCD  &&  rLocalName.equalsAscii( "right-text" ))
 287         pContext = new ConvDicXMLRightTextContext_Impl( GetConvDicImport(), nPrefix, rLocalName, *this );
 288     else
 289         pContext = new SvXMLImportContext( GetImport(), nPrefix, rLocalName );
 290     return pContext;
 291 }
 292
 293 void ConvDicXMLEntryTextContext_Impl::StartElement(
 294         const uno::Reference< xml::sax::XAttributeList >& rxAttrList )
 295 {
 296     sal_Int16 nAttrCount = rxAttrList.is() ? rxAttrList->getLength() : 0;
 297     for (sal_Int16 i = 0;  i < nAttrCount;  ++i)
 298     {
 299         OUString aAttrName = rxAttrList->getNameByIndex(i);
 300         OUString aLocalName;
 301         sal_uInt16 nPrefix = GetImport().GetNamespaceMap().
 302                                     GetKeyByAttrName( aAttrName, &aLocalName );
 303         OUString aValue = rxAttrList->getValueByIndex(i);
 304
 305         if (nPrefix == XML_NAMESPACE_TCD && aLocalName.equalsAscii( "left-text" ))
 306             aLeftText = aValue;
 307         if (nPrefix == XML_NAMESPACE_TCD && aLocalName.equalsAscii( "property-type" ))
 308             nPropertyType = (sal_Int16) aValue.toInt32();
 309     }
 310 }
 311
 312 ////////////////////////////////////////
 313
 314 SvXMLImportContext * ConvDicXMLRightTextContext_Impl::CreateChildContext(
 315         sal_uInt16 nPrefix, const OUString& rLocalName,
 316         const uno::Reference< xml::sax::XAttributeList > & /*rxAttrList*/ )
 317 {
 318     // leaf: return default (empty) context
 319     SvXMLImportContext *pContext = new SvXMLImportContext( GetImport(), nPrefix, rLocalName );
 320     return pContext;
 321 }
 322
 323 void ConvDicXMLRightTextContext_Impl::Characters( const OUString &rChars )
 324 {
 325     aRightText += rChars;
 326 }
 327
 328 void ConvDicXMLRightTextContext_Impl::EndElement()
 329 {
 330     ConvDic *pDic = GetDic();
 331     if (pDic)
 332         pDic->AddEntry( GetLeftText(), GetRightText() );
 333 }
 334
 335
 336 ///////////////////////////////////////////////////////////////////////////
 337
 338 sal_Bool ConvDicXMLExport::Export( SfxMedium & /*rMedium*/ )
 339 {
 340     sal_Bool bRet = sal_False;
 341
 342     uno::Reference< document::XExporter > xExporter( this );
 343     uno::Reference< document::XFilter > xFilter( xExporter, UNO_QUERY );
 344     uno::Sequence< beans::PropertyValue > aProps(0);
 345     xFilter->filter( aProps );      // calls exportDoc implicitly
 346
 347     return bRet = bSuccess;
 348 }
 349
 350
 351 sal_uInt32 ConvDicXMLExport::exportDoc( enum ::xmloff::token::XMLTokenEnum /*eClass*/ )
 352 {
 353     _GetNamespaceMap().Add( A2OU( "tcd" ),
 354             A2OU( XML_NAMESPACE_TCD_STRING ), XML_NAMESPACE_TCD );
 355
 356     GetDocHandler()->startDocument();
 357
 358     // Add xmlns line and some other arguments
 359     AddAttribute( _GetNamespaceMap().GetAttrNameByKey( XML_NAMESPACE_TCD ),
 360                   _GetNamespaceMap().GetNameByKey( XML_NAMESPACE_TCD ) );
 361     AddAttributeASCII( XML_NAMESPACE_TCD, "package", "org.openoffice.Office" );
 362
 363     OUString aIsoLang( MsLangId::convertLanguageToIsoString( rDic.nLanguage ) );
 364     AddAttribute( XML_NAMESPACE_TCD, "lang", aIsoLang );
 365     OUString aConvType( ConversionTypeToText( rDic.nConversionType ) );
 366     AddAttribute( XML_NAMESPACE_TCD, "conversion-type", aConvType );
 367
 368     //!! block necessary in order to have SvXMLElementExport d-tor called
 369     //!! before the call to endDocument
 370     {
 371         SvXMLElementExport aRoot( *this, XML_NAMESPACE_TCD, "text-conversion-dictionary", sal_True, sal_True );
 372         _ExportContent();
 373     }
 374
 375     GetDocHandler()->endDocument();
 376
 377     bSuccess = sal_True;
 378     return 0;
 379 }
 380
 381
 382 void ConvDicXMLExport::_ExportContent()
 383 {
 384     // aquire sorted list of all keys
 385     ConvMapKeySet   aKeySet;
 386     ConvMap::iterator aIt;
 387     for (aIt = rDic.aFromLeft.begin();  aIt != rDic.aFromLeft.end();  ++aIt)
 388         aKeySet.insert( (*aIt).first );
 389
 390     ConvMapKeySet::iterator aKeyIt;
 391     for (aKeyIt = aKeySet.begin();  aKeyIt != aKeySet.end();  ++aKeyIt)
 392     {
 393         OUString aLeftText( *aKeyIt );
 394         AddAttribute( XML_NAMESPACE_TCD, "left-text", aLeftText );
 395         if (rDic.pConvPropType.get())   // property-type list available?
 396         {
 397             sal_Int16 nPropertyType = -1;
 398             PropTypeMap::iterator aIt2 = rDic.pConvPropType->find( aLeftText );
 399             if (aIt2 != rDic.pConvPropType->end())
 400                 nPropertyType = (*aIt2).second;
 401             DBG_ASSERT( nPropertyType, "property-type not found" );
 402             if (nPropertyType == -1)
 403                 nPropertyType = ConversionPropertyType::NOT_DEFINED;
 404             AddAttribute( XML_NAMESPACE_TCD, "property-type", OUString::valueOf( (sal_Int32) nPropertyType ) );
 405         }
 406         SvXMLElementExport aEntryMain( *this, XML_NAMESPACE_TCD,
 407                 "entry" , sal_True, sal_True );
 408
 409         pair< ConvMap::iterator, ConvMap::iterator > aRange =
 410                 rDic.aFromLeft.equal_range( *aKeyIt );
 411         for (aIt = aRange.first;  aIt != aRange.second;  ++aIt)
 412         {
 413             DBG_ASSERT( *aKeyIt == (*aIt).first, "key <-> entry mismatch" );
 414             OUString aRightText( (*aIt).second );
 415             SvXMLElementExport aEntryRightText( *this, XML_NAMESPACE_TCD,
 416                     "right-text" , sal_True, sal_False );
 417             Characters( aRightText );
 418         }
 419     }
 420 }
 421
 422 ::rtl::OUString SAL_CALL ConvDicXMLExport::getImplementationName()
 423     throw( uno::RuntimeException )
 424 {
 425     return A2OU( "com.sun.star.lingu2.ConvDicXMLExport" );
 426 }
 427
 428 ///////////////////////////////////////////////////////////////////////////
 429
 430 void SAL_CALL ConvDicXMLImport::startDocument(void)
 431     throw( xml::sax::SAXException, uno::RuntimeException )
 432 {
 433     // register namespace at first possible opportunity
 434     GetNamespaceMap().Add( A2OU( "tcd" ),
 435             A2OU( XML_NAMESPACE_TCD_STRING ), XML_NAMESPACE_TCD );
 436     SvXMLImport::startDocument();
 437 }
 438
 439 void SAL_CALL ConvDicXMLImport::endDocument(void)
 440     throw( xml::sax::SAXException, uno::RuntimeException )
 441 {
 442     SvXMLImport::endDocument();
 443 }
 444
 445 SvXMLImportContext * ConvDicXMLImport::CreateContext(
 446         sal_uInt16 nPrefix,
 447         const rtl::OUString &rLocalName,
 448         const uno::Reference < xml::sax::XAttributeList > & /*rxAttrList*/ )
 449 {
 450     SvXMLImportContext *pContext = 0;
 451     if (nPrefix == XML_NAMESPACE_TCD && rLocalName.equalsAscii( "text-conversion-dictionary" ))
 452         pContext = new ConvDicXMLDictionaryContext_Impl( *this, nPrefix, rLocalName );
 453     else
 454         pContext = new SvXMLImportContext( *this, nPrefix, rLocalName );
 455     return pContext;
 456 }
 457
 458
 459 OUString SAL_CALL ConvDicXMLImport::getImplementationName()
 460     throw( uno::RuntimeException )
 461 {
 462     return A2OU( "com.sun.star.lingu2.ConvDicXMLImport" );
 463 }
 464
 465 ///////////////////////////////////////////////////////////////////////////
 466