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.extractors;
029
030import org.opencms.main.CmsLog;
031import org.opencms.main.OpenCms;
032import org.opencms.util.CmsHtmlExtractor;
033import org.opencms.util.CmsStringUtil;
034
035import java.io.InputStream;
036
037import org.apache.commons.logging.Log;
038
039/**
040 * Extracts the text from an HTML document.<p>
041 *
042 * @since 6.0.0
043 */
044public final class CmsExtractorHtml extends A_CmsTextExtractor {
045
046    /** Static member instance of the extractor. */
047    private static final CmsExtractorHtml INSTANCE = new CmsExtractorHtml();
048
049    /** The log object for this class. */
050    private static final Log LOG = CmsLog.getLog(CmsExtractorHtml.class);
051
052    /**
053     * Hide the public constructor.<p>
054     */
055    private CmsExtractorHtml() {
056
057        // noop
058    }
059
060    /**
061     * Returns an instance of this text extractor.<p>
062     *
063     * @return an instance of this text extractor
064     */
065    public static I_CmsTextExtractor getExtractor() {
066
067        return INSTANCE;
068    }
069
070    /**
071     * @see org.opencms.search.extractors.I_CmsTextExtractor#extractText(java.io.InputStream, java.lang.String)
072     */
073    @Override
074    public I_CmsExtractionResult extractText(InputStream in, String encoding) throws Exception {
075
076        String result = "";
077        try {
078            if (CmsStringUtil.isEmpty(encoding)) {
079                encoding = OpenCms.getSystemInfo().getDefaultEncoding();
080            }
081            result = CmsHtmlExtractor.extractText(in, encoding);
082            result = removeControlChars(result);
083        } catch (Exception e) {
084            if (LOG.isErrorEnabled()) {
085                LOG.error(Messages.get().container(Messages.LOG_EXTRACT_TEXT_ERROR_0), e);
086            }
087        }
088        return new CmsExtractionResult(result);
089    }
090}