Index rebuild is done synchronously. Added some logging messages for monitoring
[pithos] / src / gr / ebs / gss / server / ejb / indexer / IndexerMDBean.java
1 /*
2  * Copyright 2007, 2008, 2009 Electronic Business Systems Ltd.
3  *
4  * This file is part of GSS.
5  *
6  * GSS is free software: you can redistribute it and/or modify
7  * it under the terms of the GNU General Public License as published by
8  * the Free Software Foundation, either version 3 of the License, or
9  * (at your option) any later version.
10  *
11  * GSS is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
14  * GNU General Public License for more details.
15  *
16  * You should have received a copy of the GNU General Public License
17  * along with GSS.  If not, see <http://www.gnu.org/licenses/>.
18  */
19 package gr.ebs.gss.server.ejb.indexer;
20
21 import static gr.ebs.gss.server.configuration.GSSConfigurationFactory.getConfiguration;
22 import gr.ebs.gss.client.exceptions.ObjectNotFoundException;
23 import gr.ebs.gss.server.configuration.GSSConfigurationFactory;
24 import gr.ebs.gss.server.domain.FileBody;
25 import gr.ebs.gss.server.domain.FileHeader;
26 import gr.ebs.gss.server.domain.FileTag;
27 import gr.ebs.gss.server.ejb.ExternalAPI;
28 import gr.ebs.gss.server.ejb.GSSDAO;
29
30 import java.io.File;
31 import java.io.IOException;
32 import java.io.UnsupportedEncodingException;
33 import java.net.MalformedURLException;
34 import java.util.StringTokenizer;
35
36 import javax.ejb.ActivationConfigProperty;
37 import javax.ejb.EJB;
38 import javax.ejb.EJBException;
39 import javax.ejb.MessageDriven;
40 import javax.ejb.TransactionAttribute;
41 import javax.ejb.TransactionAttributeType;
42 import javax.jms.JMSException;
43 import javax.jms.MapMessage;
44 import javax.jms.Message;
45 import javax.jms.MessageListener;
46
47 import org.apache.commons.configuration.Configuration;
48 import org.apache.commons.configuration.ConfigurationException;
49 import org.apache.commons.httpclient.HttpClient;
50 import org.apache.commons.httpclient.HttpException;
51 import org.apache.commons.httpclient.methods.PostMethod;
52 import org.apache.commons.httpclient.methods.StringRequestEntity;
53 import org.apache.commons.logging.Log;
54 import org.apache.commons.logging.LogFactory;
55 import org.apache.solr.client.solrj.SolrServerException;
56 import org.apache.solr.client.solrj.impl.CommonsHttpSolrServer;
57 import org.apache.solr.client.solrj.request.AbstractUpdateRequest;
58 import org.apache.solr.client.solrj.request.ContentStreamUpdateRequest;
59 import org.apache.solr.common.SolrException;
60 import org.apache.solr.common.SolrInputDocument;
61 import org.jboss.ejb3.annotation.ResourceAdapter;
62
63 /**
64  * Message driven bean that accepts messages whenever a document is created,
65  * modified or deleted and adds/removes the item from the search index.
66  */
67 @MessageDriven(activationConfig={@ActivationConfigProperty(propertyName="destinationType", propertyValue="javax.jms.Queue"),
68                                                                         @ActivationConfigProperty(propertyName="destination", propertyValue="queue/gss-indexingQueue")})
69 @ResourceAdapter("hornetq-ra.rar")
70 public class IndexerMDBean implements MessageListener {
71         /**
72          * The logger
73          */
74         private static final Log logger = LogFactory.getLog(IndexerMDBean.class);
75
76         /**
77          * EJB offering access to the JPA entity manager
78          */
79         @EJB ExternalAPI service;
80
81         /**
82          * Decides to add or drop an item from the index depending on the message
83          * received
84          *
85          * It currently uses the patched solr API for rich documents. This API does not
86          * allow indexing time field boosting. For this reason we have to use the dismax search API (instead of the
87          * standard) that allows for search time field boosting
88          *
89          * @param msg
90          * @see javax.jms.MessageListener#onMessage(javax.jms.Message)
91          */
92         @Override
93         public void onMessage(Message msg) {
94                 Long id = null;
95                 try {
96                         MapMessage map = (MapMessage) msg;
97                         id = (Long) map.getObject("id");
98                         boolean delete = map.getBoolean("delete");
99                         Configuration config = GSSConfigurationFactory.getConfiguration();
100             CommonsHttpSolrServer solr = new CommonsHttpSolrServer(getConfiguration().getString("solr.url"));
101             if (delete) {
102                                 sendDelete(solr, id);
103                 solr.commit();
104                         } else {
105                                 service.postFileToSolr(solr, id);
106                 solr.commit();
107                         }       
108                 }
109                 catch (JMSException e) {
110                         throw new EJBException("Error processing file ID " + id, e);
111                 }
112                 catch (IOException e) {
113                         throw new EJBException("Error processing file ID " + id, e);
114                 }
115                 catch (SolrServerException e) {
116                         throw new EJBException(e);
117                 }
118         }
119
120
121         /**
122          * Sends a delete command to solr. The id is the Long id of the indexed document
123          * 
124          * @param solr
125          * @param id
126          * @throws SolrServerException
127          * @throws IOException
128          */
129         private void sendDelete(CommonsHttpSolrServer solr, Long id)    throws SolrServerException, IOException {
130                 solr.deleteById(id.toString());
131         }
132 }