001/*
002 * This library is part of OpenCms -
003 * the Open Source Content Management System
004 *
005 * Copyright (c) Alkacon Software GmbH & Co. KG (http://www.alkacon.com)
006 *
007 * This library is free software; you can redistribute it and/or
008 * modify it under the terms of the GNU Lesser General Public
009 * License as published by the Free Software Foundation; either
010 * version 2.1 of the License, or (at your option) any later version.
011 *
012 * This library is distributed in the hope that it will be useful,
013 * but WITHOUT ANY WARRANTY; without even the implied warranty of
014 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
015 * Lesser General Public License for more details.
016 *
017 * For further information about Alkacon Software GmbH & Co. KG, please see the
018 * company website: http://www.alkacon.com
019 *
020 * For further information about OpenCms, please see the
021 * project website: http://www.opencms.org
022 *
023 * You should have received a copy of the GNU Lesser General Public
024 * License along with this library; if not, write to the Free Software
025 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
026 */
027
028package org.opencms.search.documents;
029
030import org.opencms.file.CmsFile;
031import org.opencms.file.CmsObject;
032import org.opencms.file.CmsResource;
033import org.opencms.main.CmsException;
034import org.opencms.search.CmsIndexException;
035import org.opencms.search.I_CmsSearchIndex;
036import org.opencms.search.extractors.CmsExtractionResult;
037import org.opencms.search.extractors.I_CmsExtractionResult;
038import org.opencms.util.CmsHtmlExtractor;
039import org.opencms.util.CmsStringUtil;
040import org.opencms.xml.page.CmsXmlPage;
041import org.opencms.xml.page.CmsXmlPageFactory;
042
043import java.util.Iterator;
044import java.util.LinkedHashMap;
045import java.util.List;
046import java.util.Locale;
047
048/**
049 * Lucene document factory class to extract index data from a cms resource
050 * of type <code>CmsResourceTypeXmlPage</code>.<p>
051 *
052 * @since 6.0.0
053 */
054public class CmsDocumentXmlPage extends A_CmsVfsDocument {
055
056    /**
057     * Creates a new instance of this lucene document factory.<p>
058     *
059     * @param name name of the documenttype
060     */
061    public CmsDocumentXmlPage(String name) {
062
063        super(name);
064    }
065
066    /**
067     * Returns the raw text content of a given vfs resource of type <code>CmsResourceTypeXmlPage</code>.<p>
068     *
069     * @see org.opencms.search.documents.I_CmsSearchExtractor#extractContent(CmsObject, CmsResource, I_CmsSearchIndex)
070     */
071    public I_CmsExtractionResult extractContent(CmsObject cms, CmsResource resource, I_CmsSearchIndex index)
072    throws CmsException {
073
074        logContentExtraction(resource, index);
075        try {
076            CmsFile file = readFile(cms, resource);
077            CmsXmlPage page = CmsXmlPageFactory.unmarshal(cms, file);
078            Locale locale = index.getLocaleForResource(cms, resource, page.getLocales());
079
080            List<String> elements = page.getNames(locale);
081            StringBuffer content = new StringBuffer();
082            LinkedHashMap<String, String> items = new LinkedHashMap<String, String>();
083            for (Iterator<String> i = elements.iterator(); i.hasNext();) {
084                String elementName = i.next();
085                String value = page.getStringValue(cms, elementName, locale);
086                String extracted = CmsHtmlExtractor.extractText(value, page.getEncoding());
087                if (CmsStringUtil.isNotEmptyOrWhitespaceOnly(extracted)) {
088                    items.put(elementName, extracted);
089                    content.append(extracted);
090                    content.append('\n');
091                }
092            }
093
094            return new CmsExtractionResult(content.toString(), items);
095
096        } catch (Exception e) {
097            throw new CmsIndexException(
098                Messages.get().container(Messages.ERR_TEXT_EXTRACTION_1, resource.getRootPath()),
099                e);
100        }
101    }
102
103    /**
104     * @see org.opencms.search.documents.I_CmsDocumentFactory#isLocaleDependend()
105     */
106    public boolean isLocaleDependend() {
107
108        return true;
109    }
110
111    /**
112     * @see org.opencms.search.documents.I_CmsDocumentFactory#isUsingCache()
113     */
114    public boolean isUsingCache() {
115
116        return true;
117    }
118}