1*cdf0e10cSrcweir /************************************************************************* 2*cdf0e10cSrcweir * 3*cdf0e10cSrcweir * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 4*cdf0e10cSrcweir * 5*cdf0e10cSrcweir * Copyright 2000, 2010 Oracle and/or its affiliates. 6*cdf0e10cSrcweir * 7*cdf0e10cSrcweir * OpenOffice.org - a multi-platform office productivity suite 8*cdf0e10cSrcweir * 9*cdf0e10cSrcweir * This file is part of OpenOffice.org. 10*cdf0e10cSrcweir * 11*cdf0e10cSrcweir * OpenOffice.org is free software: you can redistribute it and/or modify 12*cdf0e10cSrcweir * it under the terms of the GNU Lesser General Public License version 3 13*cdf0e10cSrcweir * only, as published by the Free Software Foundation. 14*cdf0e10cSrcweir * 15*cdf0e10cSrcweir * OpenOffice.org is distributed in the hope that it will be useful, 16*cdf0e10cSrcweir * but WITHOUT ANY WARRANTY; without even the implied warranty of 17*cdf0e10cSrcweir * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 18*cdf0e10cSrcweir * GNU Lesser General Public License version 3 for more details 19*cdf0e10cSrcweir * (a copy is included in the LICENSE file that accompanied this code). 20*cdf0e10cSrcweir * 21*cdf0e10cSrcweir * You should have received a copy of the GNU Lesser General Public License 22*cdf0e10cSrcweir * version 3 along with OpenOffice.org. If not, see 23*cdf0e10cSrcweir * <http://www.openoffice.org/license.html> 24*cdf0e10cSrcweir * for a copy of the LGPLv3 License. 25*cdf0e10cSrcweir * 26*cdf0e10cSrcweir ************************************************************************/ 27*cdf0e10cSrcweir 28*cdf0e10cSrcweir // MARKER(update_precomp.py): autogen include statement, do not remove 29*cdf0e10cSrcweir #include "precompiled_shell.hxx" 30*cdf0e10cSrcweir #include "internal/contentreader.hxx" 31*cdf0e10cSrcweir #include "dummytag.hxx" 32*cdf0e10cSrcweir #include "simpletag.hxx" 33*cdf0e10cSrcweir #include "autostyletag.hxx" 34*cdf0e10cSrcweir 35*cdf0e10cSrcweir #include "assert.h" 36*cdf0e10cSrcweir 37*cdf0e10cSrcweir /** constructor. 38*cdf0e10cSrcweir */ 39*cdf0e10cSrcweir CContentReader::CContentReader( const std::string& DocumentName, LocaleSet_t const & DocumentLocale ): 40*cdf0e10cSrcweir CBaseReader( DocumentName ) 41*cdf0e10cSrcweir { 42*cdf0e10cSrcweir try 43*cdf0e10cSrcweir { 44*cdf0e10cSrcweir m_DefaultLocale = DocumentLocale; 45*cdf0e10cSrcweir Initialize( DOC_CONTENT_NAME ); 46*cdf0e10cSrcweir } 47*cdf0e10cSrcweir catch(xml_parser_exception& 48*cdf0e10cSrcweir #if OSL_DEBUG_LEVEL > 0 49*cdf0e10cSrcweir ex 50*cdf0e10cSrcweir #endif 51*cdf0e10cSrcweir ) 52*cdf0e10cSrcweir { 53*cdf0e10cSrcweir ENSURE(false, ex.what()); 54*cdf0e10cSrcweir } 55*cdf0e10cSrcweir catch(...) 56*cdf0e10cSrcweir { 57*cdf0e10cSrcweir ENSURE(false, "Unknown error"); 58*cdf0e10cSrcweir } 59*cdf0e10cSrcweir } 60*cdf0e10cSrcweir 61*cdf0e10cSrcweir CContentReader::CContentReader( void* stream, LocaleSet_t const & DocumentLocale, zlib_filefunc_def* fa ) : 62*cdf0e10cSrcweir CBaseReader( stream, fa ) 63*cdf0e10cSrcweir { 64*cdf0e10cSrcweir try 65*cdf0e10cSrcweir { 66*cdf0e10cSrcweir m_DefaultLocale = DocumentLocale; 67*cdf0e10cSrcweir Initialize( DOC_CONTENT_NAME ); 68*cdf0e10cSrcweir } 69*cdf0e10cSrcweir catch(xml_parser_exception& 70*cdf0e10cSrcweir #if OSL_DEBUG_LEVEL > 0 71*cdf0e10cSrcweir ex 72*cdf0e10cSrcweir #endif 73*cdf0e10cSrcweir ) 74*cdf0e10cSrcweir { 75*cdf0e10cSrcweir ENSURE(false, ex.what()); 76*cdf0e10cSrcweir } 77*cdf0e10cSrcweir catch(...) 78*cdf0e10cSrcweir { 79*cdf0e10cSrcweir ENSURE(false, "Unknown error"); 80*cdf0e10cSrcweir } 81*cdf0e10cSrcweir } 82*cdf0e10cSrcweir 83*cdf0e10cSrcweir 84*cdf0e10cSrcweir /** destructor. 85*cdf0e10cSrcweir */ 86*cdf0e10cSrcweir 87*cdf0e10cSrcweir CContentReader::~CContentReader( void ) 88*cdf0e10cSrcweir { 89*cdf0e10cSrcweir } 90*cdf0e10cSrcweir 91*cdf0e10cSrcweir /*********************** helper functions ***********************/ 92*cdf0e10cSrcweir 93*cdf0e10cSrcweir /** choose an appropriate tag reader 94*cdf0e10cSrcweir */ 95*cdf0e10cSrcweir 96*cdf0e10cSrcweir ITag* CContentReader::chooseTagReader( const std::wstring& tag_name, const XmlTagAttributes_t& XmlAttributes ) 97*cdf0e10cSrcweir { 98*cdf0e10cSrcweir if (( tag_name == CONTENT_TEXT_A )||( tag_name == CONTENT_TEXT_P )|| 99*cdf0e10cSrcweir ( tag_name == CONTENT_TEXT_SPAN ) ||( tag_name == CONTENT_TEXT_H )|| 100*cdf0e10cSrcweir ( tag_name == CONTENT_TEXT_SEQUENCE ) ||( tag_name == CONTENT_TEXT_BOOKMARK_REF )|| 101*cdf0e10cSrcweir ( tag_name == CONTENT_TEXT_INDEX_TITLE_TEMPLATE ) ) 102*cdf0e10cSrcweir return new CSimpleTag(XmlAttributes); 103*cdf0e10cSrcweir else if ( tag_name == CONTENT_STYLE_STYLE ) 104*cdf0e10cSrcweir { 105*cdf0e10cSrcweir // if style:style | style:name is exist,, fill the style field, otherwise do nothing; 106*cdf0e10cSrcweir if ( XmlAttributes.find(CONTENT_STYLE_STYLE_NAME) != XmlAttributes.end()) 107*cdf0e10cSrcweir return new CAutoStyleTag(XmlAttributes); 108*cdf0e10cSrcweir else 109*cdf0e10cSrcweir return new CDummyTag(); 110*cdf0e10cSrcweir } 111*cdf0e10cSrcweir else if ( ( tag_name == CONTENT_STYLE_PROPERTIES ) || ( tag_name == CONTENT_TEXT_STYLE_PROPERTIES ) ) 112*cdf0e10cSrcweir { 113*cdf0e10cSrcweir assert( !m_TagBuilderStack.empty() ); 114*cdf0e10cSrcweir 115*cdf0e10cSrcweir //here we presume that if CONTENT_STYLE_PROPERTIES tag is present, it just follow CONTENT_STYLE_STYLE; 116*cdf0e10cSrcweir ITag* pTagBuilder = m_TagBuilderStack.top(); 117*cdf0e10cSrcweir pTagBuilder->addAttributes( XmlAttributes ); 118*cdf0e10cSrcweir 119*cdf0e10cSrcweir return new CDummyTag(); 120*cdf0e10cSrcweir } 121*cdf0e10cSrcweir else 122*cdf0e10cSrcweir return new CDummyTag(); 123*cdf0e10cSrcweir } 124*cdf0e10cSrcweir 125*cdf0e10cSrcweir /** get style of the current content. 126*cdf0e10cSrcweir */ 127*cdf0e10cSrcweir ::std::wstring CContentReader::getCurrentContentStyle( void ) 128*cdf0e10cSrcweir { 129*cdf0e10cSrcweir assert( !m_TagBuilderStack.empty() ); 130*cdf0e10cSrcweir ITag* pTagBuilder = m_TagBuilderStack.top(); 131*cdf0e10cSrcweir 132*cdf0e10cSrcweir return ( pTagBuilder->getTagAttribute(CONTENT_TEXT_STYLENAME) ); 133*cdf0e10cSrcweir } 134*cdf0e10cSrcweir 135*cdf0e10cSrcweir /** add chunk into Chunk Buffer. 136*cdf0e10cSrcweir */ 137*cdf0e10cSrcweir void CContentReader::addChunk( LocaleSet_t const & Locale, Content_t const & Content ) 138*cdf0e10cSrcweir { 139*cdf0e10cSrcweir if ( Content == EMPTY_STRING ) 140*cdf0e10cSrcweir return; 141*cdf0e10cSrcweir 142*cdf0e10cSrcweir if ( ( ( m_ChunkBuffer.empty() ) || ( m_ChunkBuffer.back().first != Locale ) ) && 143*cdf0e10cSrcweir ( ( Content != SPACE ) && ( Content != LF ) ) ) 144*cdf0e10cSrcweir { 145*cdf0e10cSrcweir // if met a new locale, add a blank new chunk; 146*cdf0e10cSrcweir Chunk_t Chunk; 147*cdf0e10cSrcweir Chunk.first = Locale; 148*cdf0e10cSrcweir Chunk.second = EMPTY_STRING; 149*cdf0e10cSrcweir m_ChunkBuffer.push_back( Chunk ); 150*cdf0e10cSrcweir } 151*cdf0e10cSrcweir 152*cdf0e10cSrcweir if ( !m_ChunkBuffer.empty() ) 153*cdf0e10cSrcweir m_ChunkBuffer.back().second += Content; 154*cdf0e10cSrcweir } 155*cdf0e10cSrcweir 156*cdf0e10cSrcweir /** get a style's locale field. 157*cdf0e10cSrcweir */ 158*cdf0e10cSrcweir 159*cdf0e10cSrcweir LocaleSet_t const & CContentReader::getLocale( const StyleName_t Style ) 160*cdf0e10cSrcweir { 161*cdf0e10cSrcweir if ( m_StyleMap.empty() ) 162*cdf0e10cSrcweir return m_DefaultLocale; 163*cdf0e10cSrcweir 164*cdf0e10cSrcweir StyleLocaleMap_t :: const_iterator style_Iter; 165*cdf0e10cSrcweir 166*cdf0e10cSrcweir if ( ( style_Iter = m_StyleMap.find( Style ) ) == m_StyleMap.end( ) ) 167*cdf0e10cSrcweir return m_DefaultLocale; 168*cdf0e10cSrcweir else 169*cdf0e10cSrcweir return style_Iter->second; 170*cdf0e10cSrcweir 171*cdf0e10cSrcweir } 172*cdf0e10cSrcweir 173*cdf0e10cSrcweir /*********************** event handler functions ***********************/ 174*cdf0e10cSrcweir 175*cdf0e10cSrcweir //------------------------------ 176*cdf0e10cSrcweir // start_element occurs when a tag is start 177*cdf0e10cSrcweir //------------------------------ 178*cdf0e10cSrcweir 179*cdf0e10cSrcweir void CContentReader::start_element( 180*cdf0e10cSrcweir const std::wstring& /*raw_name*/, 181*cdf0e10cSrcweir const std::wstring& local_name, 182*cdf0e10cSrcweir const XmlTagAttributes_t& attributes) 183*cdf0e10cSrcweir { 184*cdf0e10cSrcweir //get appropriate Xml Tag Builder using MetaInfoBuilderFactory; 185*cdf0e10cSrcweir ITag* pTagBuilder = chooseTagReader( local_name,attributes ); 186*cdf0e10cSrcweir assert( pTagBuilder != NULL ); 187*cdf0e10cSrcweir pTagBuilder->startTag( ); 188*cdf0e10cSrcweir m_TagBuilderStack.push( pTagBuilder ); 189*cdf0e10cSrcweir 190*cdf0e10cSrcweir } 191*cdf0e10cSrcweir 192*cdf0e10cSrcweir //------------------------------ 193*cdf0e10cSrcweir // end_element occurs when a tag is closed 194*cdf0e10cSrcweir //------------------------------ 195*cdf0e10cSrcweir 196*cdf0e10cSrcweir void CContentReader::end_element(const std::wstring& /*raw_name*/, const std::wstring& local_name) 197*cdf0e10cSrcweir { 198*cdf0e10cSrcweir assert( !m_TagBuilderStack.empty() ); 199*cdf0e10cSrcweir ITag* pTagBuilder = m_TagBuilderStack.top(); 200*cdf0e10cSrcweir 201*cdf0e10cSrcweir if ( local_name == CONTENT_STYLE_STYLE ) 202*cdf0e10cSrcweir { 203*cdf0e10cSrcweir StyleLocalePair_t StyleLocalePair = static_cast<CAutoStyleTag * >( pTagBuilder)->getStyleLocalePair(); 204*cdf0e10cSrcweir if ( ( static_cast<CAutoStyleTag * >( pTagBuilder)->isFull() ) && ( StyleLocalePair.second != m_DefaultLocale ) ) 205*cdf0e10cSrcweir m_StyleMap.insert( StyleLocalePair ); 206*cdf0e10cSrcweir } 207*cdf0e10cSrcweir if (( local_name == CONTENT_TEXT_A )||( local_name == CONTENT_TEXT_SPAN ) || 208*cdf0e10cSrcweir ( local_name == CONTENT_TEXT_SEQUENCE )||( local_name == CONTENT_TEXT_BOOKMARK_REF )) 209*cdf0e10cSrcweir addChunk( getLocale( getCurrentContentStyle() ), ::std::wstring( SPACE ) ); 210*cdf0e10cSrcweir if ((( local_name == CONTENT_TEXT_P )||( local_name == CONTENT_TEXT_H ) || 211*cdf0e10cSrcweir ( local_name == CONTENT_TEXT_INDEX_TITLE_TEMPLATE ) )&& 212*cdf0e10cSrcweir ( EMPTY_STRING != pTagBuilder->getTagContent( ) ) ) 213*cdf0e10cSrcweir addChunk( getLocale( getCurrentContentStyle() ), ::std::wstring( LF ) ); 214*cdf0e10cSrcweir 215*cdf0e10cSrcweir m_TagBuilderStack.pop(); 216*cdf0e10cSrcweir pTagBuilder->endTag(); 217*cdf0e10cSrcweir delete pTagBuilder; 218*cdf0e10cSrcweir 219*cdf0e10cSrcweir } 220*cdf0e10cSrcweir 221*cdf0e10cSrcweir //------------------------------ 222*cdf0e10cSrcweir // characters occurs when receiving characters 223*cdf0e10cSrcweir //------------------------------ 224*cdf0e10cSrcweir 225*cdf0e10cSrcweir void CContentReader::characters( const std::wstring& character ) 226*cdf0e10cSrcweir { 227*cdf0e10cSrcweir if ( character.length() > 0 && !HasOnlySpaces( character ) ) 228*cdf0e10cSrcweir { 229*cdf0e10cSrcweir addChunk( getLocale( getCurrentContentStyle() ), ::std::wstring( character ) ); 230*cdf0e10cSrcweir 231*cdf0e10cSrcweir ITag* pTagBuilder = m_TagBuilderStack.top(); 232*cdf0e10cSrcweir pTagBuilder->addCharacters( character ); 233*cdf0e10cSrcweir } 234*cdf0e10cSrcweir } 235*cdf0e10cSrcweir 236