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.CmsExtractorPdf;
037import org.opencms.search.extractors.I_CmsExtractionResult;
038
039import org.apache.pdfbox.pdmodel.encryption.InvalidPasswordException;
040
041/**
042 * Lucene document factory class to extract index data from a cms resource
043 * containing Adobe pdf data.<p>
044 *
045 * @since 6.0.0
046 */
047public class CmsDocumentPdf extends A_CmsVfsDocument {
048
049    /**
050     * Creates a new instance of this lucene document factory.<p>
051     *
052     * @param name name of the documenttype
053     */
054    public CmsDocumentPdf(String name) {
055
056        super(name);
057    }
058
059    /**
060     * Returns the raw text content of a given vfs resource containing Adobe PDF data.<p>
061     *
062     * @see org.opencms.search.documents.I_CmsSearchExtractor#extractContent(CmsObject, CmsResource, I_CmsSearchIndex)
063     */
064    public I_CmsExtractionResult extractContent(CmsObject cms, CmsResource resource, I_CmsSearchIndex index)
065    throws CmsIndexException, CmsException {
066
067        logContentExtraction(resource, index);
068        CmsFile file = readFile(cms, resource);
069        try {
070            return CmsExtractorPdf.getExtractor().extractText(file.getContents());
071        } catch (Exception e) {
072            if (e.getClass().getSimpleName().equals("EncryptedDocumentException")) {
073                throw new CmsIndexException(
074                    Messages.get().container(Messages.ERR_DECRYPTING_RESOURCE_1, resource.getRootPath()),
075                    e);
076            }
077            if (e instanceof InvalidPasswordException) {
078                // default password "" was wrong.
079                throw new CmsIndexException(
080                    Messages.get().container(Messages.ERR_PWD_PROTECTED_1, resource.getRootPath()),
081                    e);
082            }
083            throw new CmsIndexException(
084                Messages.get().container(Messages.ERR_TEXT_EXTRACTION_1, resource.getRootPath()),
085                e);
086        }
087    }
088
089    /**
090     * @see org.opencms.search.documents.I_CmsDocumentFactory#isLocaleDependend()
091     */
092    public boolean isLocaleDependend() {
093
094        return false;
095    }
096
097    /**
098     * @see org.opencms.search.documents.I_CmsDocumentFactory#isUsingCache()
099     */
100    public boolean isUsingCache() {
101
102        return true;
103    }
104}