1f8e2c85aSAndrew Rist /************************************************************** 2cdf0e10cSrcweir * 3f8e2c85aSAndrew Rist * Licensed to the Apache Software Foundation (ASF) under one 4f8e2c85aSAndrew Rist * or more contributor license agreements. See the NOTICE file 5f8e2c85aSAndrew Rist * distributed with this work for additional information 6f8e2c85aSAndrew Rist * regarding copyright ownership. The ASF licenses this file 7f8e2c85aSAndrew Rist * to you under the Apache License, Version 2.0 (the 8f8e2c85aSAndrew Rist * "License"); you may not use this file except in compliance 9f8e2c85aSAndrew Rist * with the License. You may obtain a copy of the License at 10cdf0e10cSrcweir * 11f8e2c85aSAndrew Rist * http://www.apache.org/licenses/LICENSE-2.0 12cdf0e10cSrcweir * 13f8e2c85aSAndrew Rist * Unless required by applicable law or agreed to in writing, 14f8e2c85aSAndrew Rist * software distributed under the License is distributed on an 15f8e2c85aSAndrew Rist * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY 16f8e2c85aSAndrew Rist * KIND, either express or implied. See the License for the 17f8e2c85aSAndrew Rist * specific language governing permissions and limitations 18f8e2c85aSAndrew Rist * under the License. 19cdf0e10cSrcweir * 20f8e2c85aSAndrew Rist *************************************************************/ 21f8e2c85aSAndrew Rist 22f8e2c85aSAndrew Rist 23cdf0e10cSrcweir 24cdf0e10cSrcweir // MARKER(update_precomp.py): autogen include statement, do not remove 25cdf0e10cSrcweir #include "precompiled_shell.hxx" 26cdf0e10cSrcweir #include "internal/metainforeader.hxx" 27cdf0e10cSrcweir #include "dummytag.hxx" 28cdf0e10cSrcweir #include "simpletag.hxx" 29cdf0e10cSrcweir #include "keywordstag.hxx" 30cdf0e10cSrcweir 31cdf0e10cSrcweir #include "assert.h" 32cdf0e10cSrcweir 33cdf0e10cSrcweir /** constructor. 34cdf0e10cSrcweir */ 35cdf0e10cSrcweir CMetaInfoReader::CMetaInfoReader( const std::string& DocumentName ): 36cdf0e10cSrcweir CBaseReader( DocumentName ) 37cdf0e10cSrcweir { 38cdf0e10cSrcweir try 39cdf0e10cSrcweir { 40cdf0e10cSrcweir m_pKeywords_Builder = new CKeywordsTag( ); 41cdf0e10cSrcweir m_pSimple_Builder = new CSimpleTag( ); 42cdf0e10cSrcweir m_pDummy_Builder = new CDummyTag( ); 43cdf0e10cSrcweir 4486e1cf34SPedro Giffuni //retrieve all information that is useful 45cdf0e10cSrcweir m_AllMetaInfo[META_INFO_AUTHOR] = EMPTY_XML_TAG; 46cdf0e10cSrcweir m_AllMetaInfo[META_INFO_TITLE] = EMPTY_XML_TAG; 47cdf0e10cSrcweir m_AllMetaInfo[META_INFO_SUBJECT] = EMPTY_XML_TAG; 48cdf0e10cSrcweir m_AllMetaInfo[META_INFO_KEYWORDS] = EMPTY_XML_TAG; 49cdf0e10cSrcweir m_AllMetaInfo[META_INFO_DESCRIPTION] = EMPTY_XML_TAG; 50cdf0e10cSrcweir m_AllMetaInfo[META_INFO_DOCUMENT_STATISTIC] = EMPTY_XML_TAG; 51cdf0e10cSrcweir 52cdf0e10cSrcweir m_AllMetaInfo[META_INFO_GENERATOR] = EMPTY_XML_TAG; 53cdf0e10cSrcweir m_AllMetaInfo[META_INFO_CREATION] = EMPTY_XML_TAG; 54cdf0e10cSrcweir m_AllMetaInfo[META_INFO_CREATOR] = EMPTY_XML_TAG; 55cdf0e10cSrcweir m_AllMetaInfo[META_INFO_MODIFIED] = EMPTY_XML_TAG; 56cdf0e10cSrcweir m_AllMetaInfo[META_INFO_LANGUAGE] = EMPTY_XML_TAG; 57cdf0e10cSrcweir m_AllMetaInfo[META_INFO_DOCUMENT_NUMBER] = EMPTY_XML_TAG; 58cdf0e10cSrcweir m_AllMetaInfo[META_INFO_EDITING_TIME] = EMPTY_XML_TAG; 59cdf0e10cSrcweir 60cdf0e10cSrcweir Initialize( META_CONTENT_NAME ); 61cdf0e10cSrcweir } 62cdf0e10cSrcweir catch(xml_parser_exception& 63cdf0e10cSrcweir #if OSL_DEBUG_LEVEL > 0 64cdf0e10cSrcweir ex 65cdf0e10cSrcweir #endif 66cdf0e10cSrcweir ) 67cdf0e10cSrcweir { 68cdf0e10cSrcweir ENSURE(false, ex.what()); 69cdf0e10cSrcweir } 70cdf0e10cSrcweir catch(...) 71cdf0e10cSrcweir { 72cdf0e10cSrcweir ENSURE(false, "Unknown error"); 73cdf0e10cSrcweir } 74cdf0e10cSrcweir } 75cdf0e10cSrcweir 76cdf0e10cSrcweir CMetaInfoReader::CMetaInfoReader( void* stream, zlib_filefunc_def* fa) : 77cdf0e10cSrcweir CBaseReader( stream, fa) 78cdf0e10cSrcweir { 79cdf0e10cSrcweir try 80cdf0e10cSrcweir { 81cdf0e10cSrcweir m_pKeywords_Builder = new CKeywordsTag( ); 82cdf0e10cSrcweir m_pSimple_Builder = new CSimpleTag( ); 83cdf0e10cSrcweir m_pDummy_Builder = new CDummyTag( ); 84cdf0e10cSrcweir 8586e1cf34SPedro Giffuni //retrieve all information that is useful 86cdf0e10cSrcweir m_AllMetaInfo[META_INFO_AUTHOR] = EMPTY_XML_TAG; 87cdf0e10cSrcweir m_AllMetaInfo[META_INFO_TITLE] = EMPTY_XML_TAG; 88cdf0e10cSrcweir m_AllMetaInfo[META_INFO_SUBJECT] = EMPTY_XML_TAG; 89cdf0e10cSrcweir m_AllMetaInfo[META_INFO_KEYWORDS] = EMPTY_XML_TAG; 90cdf0e10cSrcweir m_AllMetaInfo[META_INFO_DESCRIPTION] = EMPTY_XML_TAG; 91cdf0e10cSrcweir m_AllMetaInfo[META_INFO_DOCUMENT_STATISTIC] = EMPTY_XML_TAG; 92cdf0e10cSrcweir 93cdf0e10cSrcweir m_AllMetaInfo[META_INFO_GENERATOR] = EMPTY_XML_TAG; 94cdf0e10cSrcweir m_AllMetaInfo[META_INFO_CREATION] = EMPTY_XML_TAG; 95cdf0e10cSrcweir m_AllMetaInfo[META_INFO_CREATOR] = EMPTY_XML_TAG; 96cdf0e10cSrcweir m_AllMetaInfo[META_INFO_MODIFIED] = EMPTY_XML_TAG; 97cdf0e10cSrcweir m_AllMetaInfo[META_INFO_LANGUAGE] = EMPTY_XML_TAG; 98cdf0e10cSrcweir m_AllMetaInfo[META_INFO_DOCUMENT_NUMBER] = EMPTY_XML_TAG; 99cdf0e10cSrcweir m_AllMetaInfo[META_INFO_EDITING_TIME] = EMPTY_XML_TAG; 100cdf0e10cSrcweir 101cdf0e10cSrcweir Initialize( META_CONTENT_NAME ); 102cdf0e10cSrcweir } 103cdf0e10cSrcweir catch(xml_parser_exception& 104cdf0e10cSrcweir #if OSL_DEBUG_LEVEL > 0 105cdf0e10cSrcweir ex 106cdf0e10cSrcweir #endif 107cdf0e10cSrcweir ) 108cdf0e10cSrcweir { 109cdf0e10cSrcweir ENSURE(false, ex.what()); 110cdf0e10cSrcweir } 111cdf0e10cSrcweir catch(...) 112cdf0e10cSrcweir { 113cdf0e10cSrcweir ENSURE(false, "Unknown error"); 114cdf0e10cSrcweir } 115cdf0e10cSrcweir 116cdf0e10cSrcweir } 117cdf0e10cSrcweir 118cdf0e10cSrcweir /** destructor. 119cdf0e10cSrcweir */ 120cdf0e10cSrcweir 121cdf0e10cSrcweir CMetaInfoReader::~CMetaInfoReader( void ) 122cdf0e10cSrcweir { 123cdf0e10cSrcweir delete m_pKeywords_Builder; 124cdf0e10cSrcweir delete m_pSimple_Builder; 125cdf0e10cSrcweir delete m_pDummy_Builder; 126cdf0e10cSrcweir } 127cdf0e10cSrcweir 128cdf0e10cSrcweir 129cdf0e10cSrcweir /*********************** export functions ***********************/ 130cdf0e10cSrcweir 131cdf0e10cSrcweir /** check if the Tag is in the target meta.xml file. 132cdf0e10cSrcweir 133cdf0e10cSrcweir @param TagName 134*8cc42d09Smseidel the name of the tag that will be retrieve. 135cdf0e10cSrcweir */ 136cdf0e10cSrcweir bool CMetaInfoReader::hasTag( std::wstring TagName ) const 137cdf0e10cSrcweir { 138cdf0e10cSrcweir return ( m_AllMetaInfo.find(TagName) != m_AllMetaInfo.end()); 139cdf0e10cSrcweir } 140cdf0e10cSrcweir 141cdf0e10cSrcweir /** Get a specific tag content, compound tags will be returned as comma separated list. 142cdf0e10cSrcweir 143cdf0e10cSrcweir @param TagName 144*8cc42d09Smseidel the name of the tag that will be retrieve. 145cdf0e10cSrcweir */ 146cdf0e10cSrcweir std::wstring CMetaInfoReader::getTagData( const std::wstring& TagName) 147cdf0e10cSrcweir { 148cdf0e10cSrcweir if( hasTag( TagName ) ) 149cdf0e10cSrcweir return m_AllMetaInfo[TagName].first; 150cdf0e10cSrcweir else 151cdf0e10cSrcweir return EMPTY_STRING; 152cdf0e10cSrcweir } 153cdf0e10cSrcweir 154cdf0e10cSrcweir /** check if the a tag has the specific attribute. 155cdf0e10cSrcweir 156cdf0e10cSrcweir @param TagName 157cdf0e10cSrcweir the name of the tag. 158cdf0e10cSrcweir @param AttributeName 159cdf0e10cSrcweir the name of the attribute. 160cdf0e10cSrcweir */ 161cdf0e10cSrcweir bool CMetaInfoReader::hasTagAttribute( const std::wstring TagName, std::wstring AttributeName) 162cdf0e10cSrcweir { 163cdf0e10cSrcweir return ( m_AllMetaInfo[TagName].second.find( AttributeName) != m_AllMetaInfo[TagName].second.end() ); 164cdf0e10cSrcweir } 165cdf0e10cSrcweir 166cdf0e10cSrcweir /** Get a specific attribute content. 167cdf0e10cSrcweir 168cdf0e10cSrcweir @param TagName 169cdf0e10cSrcweir the name of the tag. 170cdf0e10cSrcweir @param AttributeName 171cdf0e10cSrcweir the name of the attribute. 172cdf0e10cSrcweir */ 173cdf0e10cSrcweir std::wstring CMetaInfoReader::getTagAttribute( const std::wstring TagName, std::wstring AttributeName) 174cdf0e10cSrcweir { 175cdf0e10cSrcweir if ( hasTagAttribute( TagName, AttributeName ) ) 176cdf0e10cSrcweir return m_AllMetaInfo[ TagName ].second[AttributeName]; 177cdf0e10cSrcweir else 178cdf0e10cSrcweir return EMPTY_STRING; 179cdf0e10cSrcweir } 180cdf0e10cSrcweir 181cdf0e10cSrcweir /** helper function for getDefaultLocale(). 182cdf0e10cSrcweir */ 183cdf0e10cSrcweir const LocaleSet_t EN_US_LOCALE( ::std::make_pair( ::std::wstring( L"en" ), ::std::wstring( L"US" ))); 184cdf0e10cSrcweir 185cdf0e10cSrcweir bool isValidLocale ( ::std::wstring Locale ) 186cdf0e10cSrcweir { 187cdf0e10cSrcweir return ( Locale.length() == 5 ); 188cdf0e10cSrcweir } 189cdf0e10cSrcweir 190cdf0e10cSrcweir /** Get the default language of the whole document, if no such field, en-US is returned. 191cdf0e10cSrcweir */ 192cdf0e10cSrcweir LocaleSet_t CMetaInfoReader::getDefaultLocale() 193cdf0e10cSrcweir { 194cdf0e10cSrcweir if (hasTag(META_INFO_LANGUAGE)) 195cdf0e10cSrcweir { 196cdf0e10cSrcweir ::std::wstring locale = m_AllMetaInfo[META_INFO_LANGUAGE].first; 197cdf0e10cSrcweir return isValidLocale(locale) ? ::std::make_pair(locale.substr( 0,2), locale.substr( 3,2)) : EN_US_LOCALE; 198cdf0e10cSrcweir } 199cdf0e10cSrcweir else 200cdf0e10cSrcweir return EN_US_LOCALE; 201cdf0e10cSrcweir } 202cdf0e10cSrcweir 203cdf0e10cSrcweir /*********************** tag related functions ***********************/ 204cdf0e10cSrcweir 205cdf0e10cSrcweir /** choose an appropriate tag reader 206cdf0e10cSrcweir */ 207cdf0e10cSrcweir 208cdf0e10cSrcweir ITag* CMetaInfoReader::chooseTagReader( const std::wstring& tag_name, const XmlTagAttributes_t& XmlAttributes ) 209cdf0e10cSrcweir { 210cdf0e10cSrcweir if ( tag_name == META_INFO_KEYWORD ) 211cdf0e10cSrcweir { 212cdf0e10cSrcweir m_AllMetaInfo[META_INFO_KEYWORDS].second = XmlAttributes; 213cdf0e10cSrcweir return m_pKeywords_Builder; 214cdf0e10cSrcweir } 215cdf0e10cSrcweir if (( tag_name == META_INFO_TITLE )||( tag_name == META_INFO_AUTHOR )||( tag_name == META_INFO_SUBJECT )||( tag_name == META_INFO_DESCRIPTION )|| 216cdf0e10cSrcweir ( tag_name == META_INFO_DOCUMENT_STATISTIC )||( tag_name == META_INFO_GENERATOR )||( tag_name == META_INFO_CREATION )||( tag_name == META_INFO_CREATOR )|| 217cdf0e10cSrcweir ( tag_name == META_INFO_MODIFIED )||( tag_name == META_INFO_LANGUAGE )||( tag_name == META_INFO_DOCUMENT_NUMBER )||( tag_name == META_INFO_EDITING_TIME ) ) 218cdf0e10cSrcweir { 219cdf0e10cSrcweir m_AllMetaInfo[tag_name].second = XmlAttributes; 220cdf0e10cSrcweir return m_pSimple_Builder; 221cdf0e10cSrcweir } 222cdf0e10cSrcweir else 223cdf0e10cSrcweir return m_pDummy_Builder; 224cdf0e10cSrcweir 225cdf0e10cSrcweir } 226cdf0e10cSrcweir 227cdf0e10cSrcweir //------------------------------ 228cdf0e10cSrcweir // save the received content into structure. 229cdf0e10cSrcweir //------------------------------ 230cdf0e10cSrcweir void CMetaInfoReader::saveTagContent( const std::wstring& tag_name ) 231cdf0e10cSrcweir { 232cdf0e10cSrcweir ITag* pTagBuilder; 233cdf0e10cSrcweir if ( tag_name == META_INFO_KEYWORDS ) 234cdf0e10cSrcweir pTagBuilder = m_pKeywords_Builder; 235cdf0e10cSrcweir else if ( tag_name == META_INFO_KEYWORD ) 236cdf0e10cSrcweir { 237cdf0e10cSrcweir // added for support for OASIS xml file format. 238cdf0e10cSrcweir m_AllMetaInfo[META_INFO_KEYWORDS].first =m_pKeywords_Builder->getTagContent( ); 239cdf0e10cSrcweir return; 240cdf0e10cSrcweir } 241cdf0e10cSrcweir else if (( tag_name == META_INFO_TITLE )||( tag_name == META_INFO_AUTHOR )||( tag_name == META_INFO_SUBJECT )||( tag_name == META_INFO_DESCRIPTION )|| 242cdf0e10cSrcweir ( tag_name == META_INFO_DOCUMENT_STATISTIC )||( tag_name == META_INFO_GENERATOR )||( tag_name == META_INFO_CREATION )||( tag_name == META_INFO_CREATOR )|| 243cdf0e10cSrcweir ( tag_name == META_INFO_MODIFIED )||( tag_name == META_INFO_LANGUAGE )||( tag_name == META_INFO_DOCUMENT_NUMBER )||( tag_name == META_INFO_EDITING_TIME ) ) 244cdf0e10cSrcweir pTagBuilder = m_pSimple_Builder; 245cdf0e10cSrcweir else 246cdf0e10cSrcweir pTagBuilder = m_pDummy_Builder; 247cdf0e10cSrcweir 248cdf0e10cSrcweir m_AllMetaInfo[tag_name].first =pTagBuilder->getTagContent( ); 249cdf0e10cSrcweir } 250cdf0e10cSrcweir 251cdf0e10cSrcweir 252cdf0e10cSrcweir /*********************** event handler functions ***********************/ 253cdf0e10cSrcweir 254cdf0e10cSrcweir //------------------------------ 255cdf0e10cSrcweir // start_element occurs when a tag is start 256cdf0e10cSrcweir //------------------------------ 257cdf0e10cSrcweir 258cdf0e10cSrcweir void CMetaInfoReader::start_element( 259cdf0e10cSrcweir const std::wstring& /*raw_name*/, 260cdf0e10cSrcweir const std::wstring& local_name, 261cdf0e10cSrcweir const XmlTagAttributes_t& attributes) 262cdf0e10cSrcweir { 263cdf0e10cSrcweir //get appropriate Xml Tag Builder using MetaInfoBuilderFactory; 264cdf0e10cSrcweir ITag* pTagBuilder = chooseTagReader( local_name,attributes ); 265cdf0e10cSrcweir assert( pTagBuilder!= NULL ); 266cdf0e10cSrcweir pTagBuilder->startTag( ); 267cdf0e10cSrcweir m_TagBuilderStack.push( pTagBuilder ); 268cdf0e10cSrcweir 269cdf0e10cSrcweir } 270cdf0e10cSrcweir 271cdf0e10cSrcweir //------------------------------ 272cdf0e10cSrcweir // end_element occurs when a tag is closed 273cdf0e10cSrcweir //------------------------------ 274cdf0e10cSrcweir 275cdf0e10cSrcweir void CMetaInfoReader::end_element(const std::wstring& /*raw_name*/, const std::wstring& local_name) 276cdf0e10cSrcweir { 277cdf0e10cSrcweir assert( !m_TagBuilderStack.empty() ); 278cdf0e10cSrcweir ITag* pTagBuilder = m_TagBuilderStack.top(); 279cdf0e10cSrcweir m_TagBuilderStack.pop(); 280cdf0e10cSrcweir pTagBuilder->endTag(); 281cdf0e10cSrcweir 282cdf0e10cSrcweir saveTagContent( local_name ); 283cdf0e10cSrcweir 284cdf0e10cSrcweir } 285cdf0e10cSrcweir 286cdf0e10cSrcweir //------------------------------ 287cdf0e10cSrcweir // characters occurs when receiving characters 288cdf0e10cSrcweir //------------------------------ 289cdf0e10cSrcweir 290cdf0e10cSrcweir void CMetaInfoReader::characters( const std::wstring& character ) 291cdf0e10cSrcweir { 292cdf0e10cSrcweir if ( character.length() > 0 && !HasOnlySpaces( character ) ) 293cdf0e10cSrcweir { 294cdf0e10cSrcweir ITag* pTagBuilder = m_TagBuilderStack.top(); 295cdf0e10cSrcweir pTagBuilder->addCharacters( character ); 296cdf0e10cSrcweir } 297cdf0e10cSrcweir } 298cdf0e10cSrcweir 299