1 |
7606
|
tao
|
/**
|
2 |
8138
|
tao
|
* Copyright: 2013 Regents of the University of California and the
|
3 |
7606
|
tao
|
* National Center for Ecological Analysis and Synthesis
|
4 |
|
|
*
|
5 |
|
|
* This program is free software; you can redistribute it and/or modify
|
6 |
|
|
* it under the terms of the GNU General Public License as published by
|
7 |
|
|
* the Free Software Foundation; either version 2 of the License, or
|
8 |
|
|
* (at your option) any later version.
|
9 |
|
|
*
|
10 |
|
|
* This program is distributed in the hope that it will be useful,
|
11 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
12 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
13 |
|
|
* GNU General Public License for more details.
|
14 |
|
|
*
|
15 |
|
|
* You should have received a copy of the GNU General Public License
|
16 |
|
|
* along with this program; if not, write to the Free Software
|
17 |
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
18 |
|
|
*/
|
19 |
|
|
package edu.ucsb.nceas.metacat.index;
|
20 |
|
|
|
21 |
7691
|
tao
|
import java.io.FileInputStream;
|
22 |
|
|
import java.io.FileNotFoundException;
|
23 |
7876
|
tao
|
import java.io.IOException;
|
24 |
7606
|
tao
|
import java.io.InputStream;
|
25 |
|
|
import java.util.ArrayList;
|
26 |
7876
|
tao
|
import java.util.Collections;
|
27 |
7606
|
tao
|
import java.util.Date;
|
28 |
|
|
import java.util.List;
|
29 |
7788
|
tao
|
import java.util.TimerTask;
|
30 |
7606
|
tao
|
|
31 |
7876
|
tao
|
import javax.xml.parsers.ParserConfigurationException;
|
32 |
|
|
import javax.xml.xpath.XPathExpressionException;
|
33 |
|
|
|
34 |
7606
|
tao
|
import org.apache.commons.logging.Log;
|
35 |
|
|
import org.apache.commons.logging.LogFactory;
|
36 |
|
|
import org.apache.solr.client.solrj.SolrServerException;
|
37 |
|
|
import org.dataone.configuration.Settings;
|
38 |
|
|
import org.dataone.service.exceptions.InvalidRequest;
|
39 |
|
|
import org.dataone.service.exceptions.InvalidToken;
|
40 |
|
|
import org.dataone.service.exceptions.NotAuthorized;
|
41 |
7876
|
tao
|
import org.dataone.service.exceptions.NotFound;
|
42 |
7606
|
tao
|
import org.dataone.service.exceptions.NotImplemented;
|
43 |
|
|
import org.dataone.service.exceptions.ServiceFailure;
|
44 |
7876
|
tao
|
import org.dataone.service.exceptions.UnsupportedType;
|
45 |
7606
|
tao
|
import org.dataone.service.types.v1.Identifier;
|
46 |
7740
|
tao
|
import org.dataone.service.types.v1.ObjectFormatIdentifier;
|
47 |
8826
|
leinfelder
|
import org.dataone.service.types.v2.SystemMetadata;
|
48 |
8023
|
tao
|
import org.dspace.foresite.OREParserException;
|
49 |
7876
|
tao
|
import org.xml.sax.SAXException;
|
50 |
7606
|
tao
|
|
51 |
7691
|
tao
|
import com.hazelcast.core.IMap;
|
52 |
7793
|
tao
|
import com.hazelcast.core.ISet;
|
53 |
7606
|
tao
|
|
54 |
8464
|
leinfelder
|
import edu.ucsb.nceas.metacat.common.index.IndexTask;
|
55 |
7828
|
leinfelder
|
import edu.ucsb.nceas.metacat.common.index.event.IndexEvent;
|
56 |
8864
|
tao
|
import edu.ucsb.nceas.metacat.common.resourcemap.ResourceMapNamespaces;
|
57 |
7802
|
tao
|
import edu.ucsb.nceas.metacat.index.event.EventlogFactory;
|
58 |
7806
|
tao
|
import edu.ucsb.nceas.metacat.index.event.IndexEventLogException;
|
59 |
7691
|
tao
|
|
60 |
7802
|
tao
|
|
61 |
7606
|
tao
|
/**
|
62 |
|
|
* A class represents the object to generate massive solr indexes.
|
63 |
|
|
* This can happen during an update of Metacat (generating index for all existing documents)
|
64 |
|
|
* or regenerate index for those documents
|
65 |
|
|
* failing to build index during the insert or update.
|
66 |
|
|
*
|
67 |
|
|
* @author tao
|
68 |
|
|
*
|
69 |
|
|
*/
|
70 |
8352
|
tao
|
public class IndexGeneratorTimerTask extends TimerTask {
|
71 |
7606
|
tao
|
|
72 |
7740
|
tao
|
private static final int FIRST =0;
|
73 |
|
|
private static final int SECOND =1;
|
74 |
7876
|
tao
|
private static final int THIRD = 2;
|
75 |
|
|
private static final int FOURTH = 3;
|
76 |
7774
|
tao
|
public static final int WAITTIME = 10000;
|
77 |
|
|
public static final int MAXWAITNUMBER = 180;
|
78 |
7606
|
tao
|
private static final String HTTP = "http://";
|
79 |
|
|
private static final String MNAPPENDIX = "/d1/mn";
|
80 |
8864
|
tao
|
//private static final String RESOURCEMAPPROPERYNAME = "index.resourcemap.namespace";
|
81 |
7774
|
tao
|
public static final String WAITIMEPOPERTYNAME = "index.regenerate.start.waitingtime";
|
82 |
|
|
public static final String MAXATTEMPTSPROPERTYNAME = "index.regenerate.start.maxattempts";
|
83 |
7748
|
tao
|
|
84 |
|
|
|
85 |
7606
|
tao
|
private SolrIndex solrIndex = null;
|
86 |
7691
|
tao
|
//private SystemMetadataEventListener systemMetadataListener = null;
|
87 |
|
|
private IMap<Identifier, SystemMetadata> systemMetadataMap;
|
88 |
|
|
private IMap<Identifier, String> objectPathMap;
|
89 |
8464
|
leinfelder
|
private IMap<Identifier, IndexTask> indexQueue;
|
90 |
8352
|
tao
|
private Log log = LogFactory.getLog(IndexGeneratorTimerTask.class);
|
91 |
7793
|
tao
|
//private MNode mNode = null;
|
92 |
7786
|
tao
|
private static List<String> resourceMapNamespaces = null;
|
93 |
7606
|
tao
|
|
94 |
|
|
/**
|
95 |
|
|
* Constructor
|
96 |
|
|
* @param solrIndex
|
97 |
|
|
* @param systemMetadataListener
|
98 |
|
|
*/
|
99 |
8352
|
tao
|
public IndexGeneratorTimerTask(SolrIndex solrIndex) {
|
100 |
7606
|
tao
|
this.solrIndex = solrIndex;
|
101 |
8864
|
tao
|
resourceMapNamespaces = ResourceMapNamespaces.getNamespaces();
|
102 |
7691
|
tao
|
//this.systemMetadataListener = systemMetadataListener;
|
103 |
7793
|
tao
|
//this.mNode = new MNode(buildMNBaseURL());
|
104 |
8292
|
tao
|
|
105 |
7606
|
tao
|
}
|
106 |
|
|
|
107 |
8343
|
tao
|
|
108 |
7606
|
tao
|
|
109 |
|
|
/**
|
110 |
7806
|
tao
|
* Build the index for all documents.
|
111 |
7606
|
tao
|
* @throws SolrServerException
|
112 |
|
|
* @throws ServiceFailure
|
113 |
|
|
* @throws NotImplemented
|
114 |
|
|
* @throws NotAuthorized
|
115 |
|
|
* @throws InvalidToken
|
116 |
|
|
* @throws InvalidRequest
|
117 |
7806
|
tao
|
* @throws IndexEventLogException
|
118 |
|
|
* @throws IllegalAccessException
|
119 |
|
|
* @throws InstantiationException
|
120 |
|
|
* @throws ClassNotFoundException
|
121 |
7876
|
tao
|
* @throws ParserConfigurationException
|
122 |
|
|
* @throws SAXException
|
123 |
|
|
* @throws IOException
|
124 |
|
|
* @throws UnsupportedType
|
125 |
|
|
* @throws NotFound
|
126 |
|
|
* @throws XPathExpressionException
|
127 |
8023
|
tao
|
* @throws OREParserException
|
128 |
7606
|
tao
|
*/
|
129 |
7806
|
tao
|
public void indexAll() throws InvalidRequest, InvalidToken,
|
130 |
8023
|
tao
|
NotAuthorized, NotImplemented, ServiceFailure, SolrServerException, ClassNotFoundException, InstantiationException, IllegalAccessException, IndexEventLogException, XPathExpressionException, NotFound, UnsupportedType, IOException, SAXException, ParserConfigurationException, OREParserException {
|
131 |
7606
|
tao
|
Date since = null;
|
132 |
|
|
Date until = null;
|
133 |
7806
|
tao
|
index(since, until);
|
134 |
7606
|
tao
|
}
|
135 |
|
|
|
136 |
|
|
/**
|
137 |
|
|
* Build the index for the docs which have been modified since the specified date.
|
138 |
|
|
* @param since
|
139 |
|
|
* @throws SolrServerException
|
140 |
|
|
* @throws ServiceFailure
|
141 |
|
|
* @throws NotImplemented
|
142 |
|
|
* @throws NotAuthorized
|
143 |
|
|
* @throws InvalidToken
|
144 |
|
|
* @throws InvalidRequest
|
145 |
7806
|
tao
|
* @throws IndexEventLogException
|
146 |
|
|
* @throws IllegalAccessException
|
147 |
|
|
* @throws InstantiationException
|
148 |
|
|
* @throws ClassNotFoundException
|
149 |
7876
|
tao
|
* @throws ParserConfigurationException
|
150 |
|
|
* @throws SAXException
|
151 |
|
|
* @throws IOException
|
152 |
|
|
* @throws UnsupportedType
|
153 |
|
|
* @throws NotFound
|
154 |
|
|
* @throws XPathExpressionException
|
155 |
8023
|
tao
|
* @throws OREParserException
|
156 |
7606
|
tao
|
*/
|
157 |
7806
|
tao
|
public void index(Date since) throws InvalidRequest, InvalidToken,
|
158 |
8023
|
tao
|
NotAuthorized, NotImplemented, ServiceFailure, SolrServerException, ClassNotFoundException, InstantiationException, IllegalAccessException, IndexEventLogException, XPathExpressionException, NotFound, UnsupportedType, IOException, SAXException, ParserConfigurationException, OREParserException {
|
159 |
7606
|
tao
|
Date until = null;
|
160 |
7806
|
tao
|
index(since, until);
|
161 |
7606
|
tao
|
}
|
162 |
|
|
|
163 |
|
|
/**
|
164 |
|
|
* Build the index for the docs which have been modified between the specified date.s
|
165 |
|
|
* @param since
|
166 |
|
|
* @param until
|
167 |
|
|
* @throws SolrServerException
|
168 |
|
|
* @throws ServiceFailure
|
169 |
|
|
* @throws NotImplemented
|
170 |
|
|
* @throws NotAuthorized
|
171 |
|
|
* @throws InvalidToken
|
172 |
|
|
* @throws InvalidRequest
|
173 |
7806
|
tao
|
* @throws IndexEventLogException
|
174 |
|
|
* @throws IllegalAccessException
|
175 |
|
|
* @throws InstantiationException
|
176 |
|
|
* @throws ClassNotFoundException
|
177 |
7876
|
tao
|
* @throws ParserConfigurationException
|
178 |
|
|
* @throws SAXException
|
179 |
|
|
* @throws IOException
|
180 |
|
|
* @throws UnsupportedType
|
181 |
|
|
* @throws NotFound
|
182 |
|
|
* @throws XPathExpressionException
|
183 |
8023
|
tao
|
* @throws OREParserException
|
184 |
7606
|
tao
|
*/
|
185 |
7806
|
tao
|
public void index(Date since, Date until) throws SolrServerException, InvalidRequest,
|
186 |
8023
|
tao
|
InvalidToken, NotAuthorized, NotImplemented, ServiceFailure, ClassNotFoundException, InstantiationException, IllegalAccessException, IndexEventLogException, XPathExpressionException, NotFound, UnsupportedType, IOException, SAXException, ParserConfigurationException, OREParserException {
|
187 |
7806
|
tao
|
Date processedDate = null;
|
188 |
7606
|
tao
|
List<String> solrIds = null;
|
189 |
7740
|
tao
|
initSystemMetadataMap();
|
190 |
|
|
initObjectPathMap();
|
191 |
8293
|
tao
|
initIndexQueue();
|
192 |
7806
|
tao
|
List[] metacatIds = getMetacatIds(since, until);
|
193 |
|
|
List<String> otherMetacatIds = metacatIds[FIRST];
|
194 |
|
|
List<String> resourceMapIds = metacatIds[SECOND];
|
195 |
8291
|
tao
|
//List<String> otherDeletedMetacatIds = metacatIds[THIRD];
|
196 |
|
|
//List<String> resourceMapDeletedIds = metacatIds[FOURTH];
|
197 |
7806
|
tao
|
|
198 |
|
|
//figure out the procesedDate by comparing the last element of otherMetacatIds and resourceMapIds.
|
199 |
7876
|
tao
|
List<Long> maxCollection = new ArrayList<Long>();
|
200 |
7806
|
tao
|
Date latestOtherId = null;
|
201 |
|
|
if (otherMetacatIds != null && !otherMetacatIds.isEmpty()) {
|
202 |
|
|
int size = otherMetacatIds.size();
|
203 |
|
|
String id = otherMetacatIds.get(size-1);
|
204 |
|
|
SystemMetadata sysmeta = getSystemMetadata(id);
|
205 |
|
|
latestOtherId = sysmeta.getDateSysMetadataModified();
|
206 |
7876
|
tao
|
maxCollection.add(new Long(latestOtherId.getTime()));
|
207 |
7606
|
tao
|
}
|
208 |
7876
|
tao
|
|
209 |
8291
|
tao
|
/*Date latestDeletedOtherIds = null;
|
210 |
7876
|
tao
|
if (otherDeletedMetacatIds != null && !otherDeletedMetacatIds.isEmpty()) {
|
211 |
|
|
int size = otherDeletedMetacatIds.size();
|
212 |
|
|
String id = otherDeletedMetacatIds.get(size-1);
|
213 |
|
|
SystemMetadata sysmeta = getSystemMetadata(id);
|
214 |
|
|
latestDeletedOtherIds = sysmeta.getDateSysMetadataModified();
|
215 |
|
|
maxCollection.add(new Long(latestDeletedOtherIds.getTime()));
|
216 |
8291
|
tao
|
}*/
|
217 |
7876
|
tao
|
|
218 |
7806
|
tao
|
Date latestResourceId = null;
|
219 |
|
|
if (resourceMapIds != null && !resourceMapIds.isEmpty()) {
|
220 |
|
|
int size = resourceMapIds.size();
|
221 |
|
|
String id = resourceMapIds.get(size-1);
|
222 |
|
|
SystemMetadata sysmeta = getSystemMetadata(id);
|
223 |
|
|
latestResourceId = sysmeta.getDateSysMetadataModified();
|
224 |
7876
|
tao
|
maxCollection.add(new Long(latestResourceId.getTime()));
|
225 |
7806
|
tao
|
}
|
226 |
7876
|
tao
|
|
227 |
8291
|
tao
|
/*Date latestDeletedResourceId = null;
|
228 |
7876
|
tao
|
if(resourceMapDeletedIds != null && !resourceMapDeletedIds.isEmpty()) {
|
229 |
|
|
int size = resourceMapDeletedIds.size();
|
230 |
|
|
String id = resourceMapDeletedIds.get(size-1);
|
231 |
|
|
SystemMetadata sysmeta = getSystemMetadata(id);
|
232 |
|
|
latestDeletedResourceId = sysmeta.getDateSysMetadataModified();
|
233 |
|
|
maxCollection.add(new Long(latestDeletedResourceId.getTime()));
|
234 |
8291
|
tao
|
}*/
|
235 |
7876
|
tao
|
|
236 |
|
|
if(!maxCollection.isEmpty()) {
|
237 |
|
|
Long max = Collections.max(maxCollection);
|
238 |
|
|
processedDate = new Date(max.longValue());
|
239 |
|
|
}
|
240 |
|
|
/*if(latestOtherId != null && latestResourceId != null && latestOtherId.getTime() > latestResourceId.getTime()) {
|
241 |
7806
|
tao
|
processedDate = latestOtherId;
|
242 |
|
|
} else if (latestOtherId != null && latestResourceId != null && latestOtherId.getTime() <= latestResourceId.getTime()) {
|
243 |
|
|
processedDate = latestResourceId;
|
244 |
|
|
} else if (latestOtherId == null && latestResourceId != null) {
|
245 |
|
|
processedDate = latestResourceId;
|
246 |
|
|
} else if (latestOtherId != null && latestResourceId == null) {
|
247 |
|
|
processedDate = latestOtherId;
|
248 |
7876
|
tao
|
}*/
|
249 |
7806
|
tao
|
|
250 |
7876
|
tao
|
|
251 |
7806
|
tao
|
//add the failedPids
|
252 |
7815
|
leinfelder
|
List<IndexEvent> failedEvents = EventlogFactory.createIndexEventLog().getEvents(null, null, null, null);
|
253 |
8293
|
tao
|
List<String> failedOtherIds = new ArrayList<String>();
|
254 |
|
|
List<String> failedResourceMapIds = new ArrayList<String>();
|
255 |
7815
|
leinfelder
|
if(failedEvents != null) {
|
256 |
|
|
for(IndexEvent event : failedEvents) {
|
257 |
|
|
String id = event.getIdentifier().getValue();
|
258 |
7806
|
tao
|
SystemMetadata sysmeta = getSystemMetadata(id);
|
259 |
7857
|
tao
|
if(sysmeta != null) {
|
260 |
7806
|
tao
|
ObjectFormatIdentifier formatId =sysmeta.getFormatId();
|
261 |
|
|
if(formatId != null && formatId.getValue() != null && resourceMapNamespaces != null && isResourceMap(formatId)) {
|
262 |
8293
|
tao
|
failedResourceMapIds.add(id);
|
263 |
7806
|
tao
|
} else {
|
264 |
8293
|
tao
|
failedOtherIds.add(id);
|
265 |
7806
|
tao
|
}
|
266 |
|
|
}
|
267 |
|
|
}
|
268 |
|
|
}
|
269 |
8293
|
tao
|
//indexFailedIds(failedOtherIds);
|
270 |
|
|
//indexFailedIds(failedResourceMapIds);
|
271 |
7806
|
tao
|
|
272 |
8293
|
tao
|
index(failedOtherIds);
|
273 |
|
|
index(failedResourceMapIds);
|
274 |
|
|
|
275 |
7857
|
tao
|
/*if(!failedOtherIds.isEmpty()) {
|
276 |
7806
|
tao
|
failedOtherIds.addAll(otherMetacatIds);
|
277 |
|
|
} else {
|
278 |
|
|
failedOtherIds = otherMetacatIds;
|
279 |
|
|
}
|
280 |
|
|
|
281 |
|
|
if(!failedResourceMapIds.isEmpty()) {
|
282 |
|
|
failedResourceMapIds.addAll(resourceMapIds);
|
283 |
|
|
} else {
|
284 |
|
|
failedResourceMapIds = resourceMapIds;
|
285 |
7857
|
tao
|
}*/
|
286 |
8293
|
tao
|
//log.info("the ids in index_event for reindex ( except the resourcemap)=====================================\n "+failedOtherIds);
|
287 |
|
|
//log.info("the resourcemap ids in index_event for reindex =====================================\n "+failedResourceMapIds);
|
288 |
7857
|
tao
|
log.info("the metacat ids (except the resource map ids)-----------------------------"+otherMetacatIds);
|
289 |
8027
|
tao
|
//logFile(otherMetacatIds, "ids-for-timed-indexing-log");
|
290 |
8291
|
tao
|
//log.info("the deleted metacat ids (except the resource map ids)-----------------------------"+otherDeletedMetacatIds);
|
291 |
7857
|
tao
|
log.info("the metacat resroucemap ids -----------------------------"+resourceMapIds);
|
292 |
8027
|
tao
|
//logFile(resourceMapIds, "ids-for-timed-indexing-log");
|
293 |
8291
|
tao
|
//log.info("the deleted metacat resroucemap ids -----------------------------"+resourceMapDeletedIds);
|
294 |
7857
|
tao
|
index(otherMetacatIds);
|
295 |
8291
|
tao
|
//removeIndex(otherDeletedMetacatIds);
|
296 |
7857
|
tao
|
index(resourceMapIds);
|
297 |
8291
|
tao
|
//removeIndex(resourceMapDeletedIds);
|
298 |
7740
|
tao
|
|
299 |
7806
|
tao
|
//record the timed index.
|
300 |
|
|
if(processedDate != null) {
|
301 |
|
|
EventlogFactory.createIndexEventLog().setLastProcessDate(processedDate);
|
302 |
|
|
}
|
303 |
7734
|
tao
|
|
304 |
7740
|
tao
|
}
|
305 |
|
|
|
306 |
|
|
/*
|
307 |
7924
|
tao
|
* Write the docids which will be indexed into a file.
|
308 |
|
|
*/
|
309 |
8084
|
tao
|
/*private void logFile(List<String> ids, String fileName) {
|
310 |
7924
|
tao
|
if(ids != null) {
|
311 |
7931
|
tao
|
try {
|
312 |
|
|
String tempDir = System.getProperty("java.io.tmpdir");
|
313 |
|
|
log.info("the temp dir is ===================== "+tempDir);
|
314 |
8017
|
tao
|
File idsForIndex = new File(tempDir, fileName);
|
315 |
7931
|
tao
|
if(!idsForIndex.exists()) {
|
316 |
|
|
idsForIndex.createNewFile();
|
317 |
8017
|
tao
|
}
|
318 |
|
|
|
319 |
7931
|
tao
|
Date date = Calendar.getInstance().getTime();
|
320 |
|
|
SimpleDateFormat format = new SimpleDateFormat("yyyy.MM.dd G 'at' HH:mm:ss z");
|
321 |
|
|
String dateStr = format.format(date);
|
322 |
|
|
List<String> dateList = new ArrayList<String>();
|
323 |
|
|
dateList.add(dateStr);
|
324 |
|
|
Boolean append = true;
|
325 |
|
|
FileUtils.writeLines(idsForIndex, dateList, append);//write time string
|
326 |
|
|
FileUtils.writeLines(idsForIndex, ids, append);
|
327 |
|
|
} catch (Exception e) {
|
328 |
|
|
log.warn("IndexGenerator.logFile - Couldn't log the ids which will be indexed since - "+e.getMessage());
|
329 |
7924
|
tao
|
}
|
330 |
7931
|
tao
|
|
331 |
7924
|
tao
|
}
|
332 |
8084
|
tao
|
}*/
|
333 |
7924
|
tao
|
/*
|
334 |
7806
|
tao
|
* Doing index
|
335 |
7740
|
tao
|
*/
|
336 |
7806
|
tao
|
private void index(List<String> metacatIds) {
|
337 |
7688
|
tao
|
if(metacatIds != null) {
|
338 |
7606
|
tao
|
for(String metacatId : metacatIds) {
|
339 |
|
|
if(metacatId != null) {
|
340 |
8293
|
tao
|
generateIndex(metacatId);
|
341 |
7606
|
tao
|
}
|
342 |
|
|
}
|
343 |
7688
|
tao
|
}
|
344 |
7606
|
tao
|
}
|
345 |
|
|
|
346 |
7857
|
tao
|
/*
|
347 |
|
|
* Index those ids which failed in the process (We got them from the EventLog)
|
348 |
|
|
*/
|
349 |
8293
|
tao
|
/*private void indexFailedIds(List<IndexEvent> events) {
|
350 |
7857
|
tao
|
if(events != null) {
|
351 |
|
|
for(IndexEvent event : events) {
|
352 |
|
|
if(event != null) {
|
353 |
|
|
Identifier identifier = event.getIdentifier();
|
354 |
|
|
if(identifier != null) {
|
355 |
|
|
String id = identifier.getValue();
|
356 |
|
|
if(id != null) {
|
357 |
|
|
Event action = event.getAction();
|
358 |
8291
|
tao
|
//if (action != null && action.equals(Event.CREATE)) {
|
359 |
7857
|
tao
|
try {
|
360 |
|
|
generateIndex(id);
|
361 |
|
|
EventlogFactory.createIndexEventLog().remove(identifier);
|
362 |
|
|
} catch (Exception e) {
|
363 |
|
|
log.error("IndexGenerator.indexFailedIds - Metacat Index couldn't generate the index for the id - "+id+" because "+e.getMessage());
|
364 |
|
|
}
|
365 |
8293
|
tao
|
|
366 |
7857
|
tao
|
}
|
367 |
|
|
}
|
368 |
|
|
}
|
369 |
|
|
}
|
370 |
|
|
}
|
371 |
8293
|
tao
|
}*/
|
372 |
7857
|
tao
|
|
373 |
7613
|
tao
|
public void run() {
|
374 |
8343
|
tao
|
|
375 |
7613
|
tao
|
try {
|
376 |
7806
|
tao
|
Date since = EventlogFactory.createIndexEventLog().getLastProcessDate();
|
377 |
|
|
index(since);
|
378 |
7613
|
tao
|
} catch (InvalidRequest e) {
|
379 |
|
|
// TODO Auto-generated catch block
|
380 |
|
|
//e.printStackTrace();
|
381 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
382 |
|
|
} catch (InvalidToken e) {
|
383 |
|
|
// TODO Auto-generated catch block
|
384 |
|
|
//e.printStackTrace();
|
385 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
386 |
|
|
} catch (NotAuthorized e) {
|
387 |
|
|
// TODO Auto-generated catch block
|
388 |
|
|
//e.printStackTrace();
|
389 |
|
|
} catch (NotImplemented e) {
|
390 |
|
|
// TODO Auto-generated catch block
|
391 |
|
|
//e.printStackTrace();
|
392 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
393 |
|
|
} catch (ServiceFailure e) {
|
394 |
|
|
// TODO Auto-generated catch block
|
395 |
|
|
//e.printStackTrace();
|
396 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
397 |
|
|
} catch (SolrServerException e) {
|
398 |
|
|
// TODO Auto-generated catch block
|
399 |
|
|
//e.printStackTrace();
|
400 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
401 |
7691
|
tao
|
} catch (FileNotFoundException e) {
|
402 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
403 |
8343
|
tao
|
} catch (ClassNotFoundException e) {
|
404 |
7806
|
tao
|
// TODO Auto-generated catch block
|
405 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
406 |
|
|
} catch (InstantiationException e) {
|
407 |
|
|
// TODO Auto-generated catch block
|
408 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
409 |
|
|
} catch (IllegalAccessException e) {
|
410 |
|
|
// TODO Auto-generated catch block
|
411 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
412 |
|
|
} catch (IndexEventLogException e) {
|
413 |
|
|
// TODO Auto-generated catch block
|
414 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
415 |
7876
|
tao
|
} catch (XPathExpressionException e) {
|
416 |
|
|
// TODO Auto-generated catch block
|
417 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
418 |
|
|
} catch (NotFound e) {
|
419 |
|
|
// TODO Auto-generated catch block
|
420 |
|
|
e.printStackTrace();
|
421 |
|
|
} catch (UnsupportedType e) {
|
422 |
|
|
// TODO Auto-generated catch block
|
423 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
424 |
|
|
} catch (IOException e) {
|
425 |
|
|
// TODO Auto-generated catch block
|
426 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
427 |
|
|
} catch (SAXException e) {
|
428 |
|
|
// TODO Auto-generated catch block
|
429 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
430 |
|
|
} catch (ParserConfigurationException e) {
|
431 |
|
|
// TODO Auto-generated catch block
|
432 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
433 |
8023
|
tao
|
} catch (OREParserException e) {
|
434 |
|
|
// TODO Auto-generated catch block
|
435 |
|
|
log.error("IndexGenerator.run - Metadata-Index couldn't generate indexes for those documents which haven't been indexed : "+e.getMessage());
|
436 |
7806
|
tao
|
}
|
437 |
7613
|
tao
|
}
|
438 |
|
|
|
439 |
8293
|
tao
|
|
440 |
7606
|
tao
|
|
441 |
|
|
/*
|
442 |
7806
|
tao
|
* Get an array of the list of ids of the metacat which has the systemmetadata modification in the range.
|
443 |
|
|
*
|
444 |
|
|
* If since and util are null, it will return all of them.
|
445 |
7740
|
tao
|
* The first element of the list is the ids except the resource map. The second elements of the list is the ids of the resource map.
|
446 |
7806
|
tao
|
* The reason to split them is when we index the resource map, we need the index of the documents in the resource map ready.
|
447 |
|
|
* The last element in the each list has the latest SystemMetadata modification date. But they are not sorted.
|
448 |
7606
|
tao
|
*/
|
449 |
7793
|
tao
|
private List[] getMetacatIds(Date since, Date until) throws InvalidRequest,
|
450 |
|
|
InvalidToken, NotAuthorized, NotImplemented, ServiceFailure, FileNotFoundException {
|
451 |
8017
|
tao
|
String fileName = "ids-from-hazelcast";
|
452 |
7740
|
tao
|
List<String> resourceMapIds = new ArrayList();
|
453 |
8291
|
tao
|
//List<String> resourceMapDeletedIds = new ArrayList();
|
454 |
7740
|
tao
|
List<String> otherIds = new ArrayList();
|
455 |
8291
|
tao
|
//List<String> otherDeletedIds = new ArrayList();
|
456 |
|
|
List[] ids = new List[2];
|
457 |
7740
|
tao
|
ids[FIRST]= otherIds;
|
458 |
|
|
ids[SECOND] = resourceMapIds;
|
459 |
8291
|
tao
|
//ids[THIRD] = otherDeletedIds;
|
460 |
|
|
//ids[FOURTH] = resourceMapDeletedIds;
|
461 |
7793
|
tao
|
ISet<Identifier> metacatIds = DistributedMapsFactory.getIdentifiersSet();
|
462 |
7806
|
tao
|
Date otherPreviousDate = null;
|
463 |
7876
|
tao
|
Date otherDeletedPreviousDate = null;
|
464 |
7806
|
tao
|
Date resourceMapPreviousDate = null;
|
465 |
7876
|
tao
|
Date resourceMapDeletedPreviousDate = null;
|
466 |
7793
|
tao
|
if(metacatIds != null) {
|
467 |
|
|
for(Identifier identifier : metacatIds) {
|
468 |
|
|
if(identifier != null && identifier.getValue() != null && !identifier.getValue().equals("")) {
|
469 |
8017
|
tao
|
List<String> idLog = new ArrayList<String>();
|
470 |
|
|
idLog.add(identifier.getValue());
|
471 |
8027
|
tao
|
//logFile(idLog, fileName);
|
472 |
7793
|
tao
|
SystemMetadata sysmeta = getSystemMetadata(identifier.getValue());
|
473 |
7876
|
tao
|
if(sysmeta != null) {
|
474 |
7793
|
tao
|
ObjectFormatIdentifier formatId =sysmeta.getFormatId();
|
475 |
|
|
//System.out.println("the object format id is "+formatId.getValue());
|
476 |
|
|
//System.out.println("the ============ resourcMapNamespaces"+resourceMapNamespaces);
|
477 |
|
|
boolean correctTimeRange = false;
|
478 |
|
|
Date sysDate = sysmeta.getDateSysMetadataModified();
|
479 |
|
|
if(since == null && until == null) {
|
480 |
|
|
correctTimeRange = true;
|
481 |
|
|
} else if (since != null && until == null) {
|
482 |
7858
|
tao
|
if(sysDate.getTime() > since.getTime()) {
|
483 |
7793
|
tao
|
correctTimeRange = true;
|
484 |
|
|
}
|
485 |
|
|
} else if (since == null && until != null) {
|
486 |
7858
|
tao
|
if(sysDate.getTime() < until.getTime()) {
|
487 |
7793
|
tao
|
correctTimeRange = true;
|
488 |
|
|
}
|
489 |
|
|
} else if (since != null && until != null) {
|
490 |
7858
|
tao
|
if(sysDate.getTime() > since.getTime() && sysDate.getTime() < until.getTime()) {
|
491 |
7793
|
tao
|
correctTimeRange = true;
|
492 |
|
|
}
|
493 |
|
|
}
|
494 |
|
|
if(correctTimeRange && formatId != null && formatId.getValue() != null && resourceMapNamespaces != null && isResourceMap(formatId)) {
|
495 |
7806
|
tao
|
//for the resource map
|
496 |
8291
|
tao
|
/*if(sysmeta.getArchived() || sysmeta.getObsoletedBy() != null) {
|
497 |
7876
|
tao
|
//archived ids
|
498 |
|
|
if(!resourceMapDeletedIds.isEmpty()) {
|
499 |
|
|
if(sysDate.getTime() > resourceMapDeletedPreviousDate.getTime()) {
|
500 |
|
|
resourceMapDeletedIds.add(identifier.getValue());//append to the end of the list if current is later than the previous one
|
501 |
|
|
resourceMapDeletedPreviousDate = sysDate;//reset resourceMapPreviousDate to the bigger one
|
502 |
|
|
} else {
|
503 |
|
|
int size = resourceMapDeletedIds.size();//
|
504 |
|
|
resourceMapDeletedIds.add(size -1, identifier.getValue());//keep the previous one at the end of the list.
|
505 |
|
|
}
|
506 |
7806
|
tao
|
} else {
|
507 |
7876
|
tao
|
resourceMapDeletedIds.add(identifier.getValue());
|
508 |
|
|
resourceMapDeletedPreviousDate = sysDate;//init resourcemapPreviousDate
|
509 |
7806
|
tao
|
}
|
510 |
8291
|
tao
|
} else {*/
|
511 |
|
|
// for all ids
|
512 |
7876
|
tao
|
if(!resourceMapIds.isEmpty()) {
|
513 |
|
|
if(sysDate.getTime() > resourceMapPreviousDate.getTime()) {
|
514 |
|
|
resourceMapIds.add(identifier.getValue());//append to the end of the list if current is later than the previous one
|
515 |
|
|
resourceMapPreviousDate = sysDate;//reset resourceMapPreviousDate to the bigger one
|
516 |
|
|
} else {
|
517 |
|
|
int size = resourceMapIds.size();//
|
518 |
|
|
resourceMapIds.add(size -1, identifier.getValue());//keep the previous one at the end of the list.
|
519 |
|
|
}
|
520 |
|
|
} else {
|
521 |
|
|
resourceMapIds.add(identifier.getValue());
|
522 |
|
|
resourceMapPreviousDate = sysDate;//init resourcemapPreviousDate
|
523 |
|
|
}
|
524 |
8291
|
tao
|
//}
|
525 |
7838
|
tao
|
} else if (correctTimeRange) {
|
526 |
8291
|
tao
|
/*if(sysmeta.getArchived() || sysmeta.getObsoletedBy() != null) {
|
527 |
7876
|
tao
|
//for the archived ids
|
528 |
|
|
if(!otherDeletedIds.isEmpty()) {
|
529 |
|
|
if(sysDate.getTime() > otherDeletedPreviousDate.getTime()) {
|
530 |
|
|
otherDeletedIds.add(identifier.getValue());
|
531 |
|
|
otherDeletedPreviousDate = sysDate;//reset otherDeletedPreviousDate to the bigger one
|
532 |
|
|
} else {
|
533 |
|
|
int size = otherDeletedIds.size();
|
534 |
|
|
otherDeletedIds.add(size-1, identifier.getValue());
|
535 |
|
|
}
|
536 |
7740
|
tao
|
} else {
|
537 |
7876
|
tao
|
otherDeletedIds.add(identifier.getValue());
|
538 |
|
|
otherDeletedPreviousDate = sysDate;//init otherDeletedPreviousDate
|
539 |
7740
|
tao
|
}
|
540 |
8291
|
tao
|
} else {*/
|
541 |
|
|
//for all ids
|
542 |
7876
|
tao
|
if(!otherIds.isEmpty()) {
|
543 |
|
|
if(sysDate.getTime() > otherPreviousDate.getTime()) {
|
544 |
|
|
otherIds.add(identifier.getValue());
|
545 |
|
|
otherPreviousDate = sysDate;//reset otherPreviousDate to the bigger one
|
546 |
|
|
} else {
|
547 |
|
|
int size = otherIds.size();
|
548 |
|
|
otherIds.add(size-1, identifier.getValue());
|
549 |
|
|
}
|
550 |
|
|
} else {
|
551 |
|
|
otherIds.add(identifier.getValue());
|
552 |
|
|
otherPreviousDate = sysDate;//init otherPreviousDate
|
553 |
|
|
}
|
554 |
8291
|
tao
|
//}
|
555 |
7607
|
tao
|
}
|
556 |
7806
|
tao
|
|
557 |
7607
|
tao
|
}
|
558 |
|
|
}
|
559 |
|
|
}
|
560 |
7806
|
tao
|
}
|
561 |
7606
|
tao
|
return ids;
|
562 |
|
|
}
|
563 |
|
|
|
564 |
|
|
/*
|
565 |
7740
|
tao
|
* If the specified ObjectFormatIdentifier is a resrouce map namespace.
|
566 |
|
|
*/
|
567 |
8864
|
tao
|
public static boolean isResourceMap(ObjectFormatIdentifier formatId) {
|
568 |
|
|
return ResourceMapNamespaces.isResourceMap(formatId);
|
569 |
7740
|
tao
|
}
|
570 |
|
|
|
571 |
7793
|
tao
|
|
572 |
7606
|
tao
|
|
573 |
|
|
/*
|
574 |
|
|
* Generate index for the id.
|
575 |
|
|
*/
|
576 |
8293
|
tao
|
private void generateIndex(String id) {
|
577 |
8343
|
tao
|
//if id is null and sysmeta will be null. If sysmeta is null, it will be caught in solrIndex.update
|
578 |
|
|
SystemMetadata sysmeta = getSystemMetadata(id);
|
579 |
|
|
Identifier pid = new Identifier();
|
580 |
|
|
pid.setValue(id);
|
581 |
|
|
solrIndex.update(pid, sysmeta);
|
582 |
|
|
|
583 |
7606
|
tao
|
}
|
584 |
7691
|
tao
|
|
585 |
|
|
/*
|
586 |
7876
|
tao
|
* Remove the solr index for the list of ids
|
587 |
|
|
*/
|
588 |
8291
|
tao
|
/*private void removeIndex(List<String> ids) {
|
589 |
7876
|
tao
|
if(ids!= null) {
|
590 |
|
|
for(String id :ids) {
|
591 |
8035
|
tao
|
try {
|
592 |
|
|
removeIndex(id);
|
593 |
|
|
} catch (Exception e) {
|
594 |
|
|
IndexEvent event = new IndexEvent();
|
595 |
|
|
Identifier pid = new Identifier();
|
596 |
|
|
pid.setValue(id);
|
597 |
|
|
event.setIdentifier(pid);
|
598 |
|
|
event.setDate(Calendar.getInstance().getTime());
|
599 |
|
|
event.setAction(Event.DELETE);
|
600 |
|
|
String error = "IndexGenerator.index - Metacat Index couldn't remove the index for the id - "+id+" because "+e.getMessage();
|
601 |
|
|
event.setDescription(error);
|
602 |
|
|
try {
|
603 |
|
|
EventlogFactory.createIndexEventLog().write(event);
|
604 |
|
|
} catch (Exception ee) {
|
605 |
|
|
log.error("SolrIndex.insertToIndex - IndexEventLog can't log the index deleting event :"+ee.getMessage());
|
606 |
|
|
}
|
607 |
|
|
log.error(error);
|
608 |
|
|
}
|
609 |
|
|
|
610 |
7876
|
tao
|
}
|
611 |
|
|
}
|
612 |
8291
|
tao
|
}*/
|
613 |
7876
|
tao
|
|
614 |
|
|
/*
|
615 |
7857
|
tao
|
* Remove the index for the id
|
616 |
|
|
*/
|
617 |
8291
|
tao
|
/*private void removeIndex(String id) throws ServiceFailure, XPathExpressionException, NotImplemented, NotFound, UnsupportedType, IOException, SolrServerException, SAXException, ParserConfigurationException, OREParserException {
|
618 |
7857
|
tao
|
if(id != null) {
|
619 |
8287
|
tao
|
//solrIndex.remove(id);
|
620 |
7857
|
tao
|
}
|
621 |
8291
|
tao
|
}*/
|
622 |
7857
|
tao
|
|
623 |
|
|
/*
|
624 |
7691
|
tao
|
* Initialize the system metadata map
|
625 |
|
|
*/
|
626 |
|
|
private void initSystemMetadataMap() throws FileNotFoundException, ServiceFailure{
|
627 |
|
|
int times = 0;
|
628 |
|
|
if(systemMetadataMap == null) {
|
629 |
7774
|
tao
|
systemMetadataMap = DistributedMapsFactory.getSystemMetadataMap();
|
630 |
7691
|
tao
|
}
|
631 |
|
|
}
|
632 |
|
|
|
633 |
|
|
/*
|
634 |
|
|
* We should call this method after calling initSystemMetadataMap since this method doesn't have the mechanism to wait the readiness of the hazelcast service
|
635 |
|
|
*/
|
636 |
|
|
private void initObjectPathMap() throws FileNotFoundException, ServiceFailure {
|
637 |
|
|
if(objectPathMap == null) {
|
638 |
|
|
objectPathMap = DistributedMapsFactory.getObjectPathMap();
|
639 |
|
|
}
|
640 |
|
|
}
|
641 |
8293
|
tao
|
|
642 |
|
|
|
643 |
|
|
|
644 |
|
|
/*
|
645 |
|
|
* Initialize the index queue
|
646 |
|
|
*/
|
647 |
|
|
private void initIndexQueue() throws FileNotFoundException, ServiceFailure {
|
648 |
|
|
if(indexQueue == null) {
|
649 |
|
|
indexQueue = DistributedMapsFactory.getIndexQueue();
|
650 |
|
|
}
|
651 |
|
|
}
|
652 |
7691
|
tao
|
/**
|
653 |
|
|
* Get an InputStream as the data object for the specific pid.
|
654 |
|
|
* @param pid
|
655 |
|
|
* @return
|
656 |
|
|
* @throws FileNotFoundException
|
657 |
|
|
*/
|
658 |
|
|
private InputStream getDataObject(String pid) throws FileNotFoundException {
|
659 |
|
|
Identifier identifier = new Identifier();
|
660 |
|
|
identifier.setValue(pid);
|
661 |
|
|
String objectPath = objectPathMap.get(identifier);
|
662 |
|
|
InputStream data = null;
|
663 |
|
|
data = new FileInputStream(objectPath);
|
664 |
|
|
return data;
|
665 |
|
|
|
666 |
|
|
}
|
667 |
|
|
|
668 |
|
|
/**
|
669 |
|
|
* Get the SystemMetadata for the specified id from the distributed Map.
|
670 |
|
|
* The null maybe is returned if there is no system metadata found.
|
671 |
|
|
* @param id the specified id.
|
672 |
|
|
* @return the SystemMetadata associated with the id.
|
673 |
|
|
*/
|
674 |
|
|
private SystemMetadata getSystemMetadata(String id) {
|
675 |
|
|
SystemMetadata metadata = null;
|
676 |
|
|
if(systemMetadataMap != null && id != null) {
|
677 |
|
|
Identifier identifier = new Identifier();
|
678 |
|
|
identifier.setValue(id);
|
679 |
|
|
metadata = systemMetadataMap.get(identifier);
|
680 |
|
|
}
|
681 |
|
|
return metadata;
|
682 |
|
|
}
|
683 |
|
|
|
684 |
|
|
/**
|
685 |
|
|
* Get the obsoletes chain of the specified id. The returned list doesn't include
|
686 |
|
|
* the specified id itself. The newer version has the lower index number in the list.
|
687 |
|
|
* Empty list will be returned if there is no document to be obsoleted by this id.
|
688 |
|
|
* @param id
|
689 |
|
|
* @return
|
690 |
|
|
*/
|
691 |
|
|
private List<String> getObsoletes(String id) {
|
692 |
|
|
List<String> obsoletes = new ArrayList<String>();
|
693 |
|
|
while (id != null) {
|
694 |
|
|
SystemMetadata metadata = getSystemMetadata(id);
|
695 |
|
|
id = null;//set it to be null in order to stop the while loop if the id can't be assinged to a new value in the following code.
|
696 |
|
|
if(metadata != null) {
|
697 |
|
|
Identifier identifier = metadata.getObsoletes();
|
698 |
|
|
if(identifier != null && identifier.getValue() != null && !identifier.getValue().trim().equals("")) {
|
699 |
|
|
obsoletes.add(identifier.getValue());
|
700 |
|
|
id = identifier.getValue();
|
701 |
|
|
}
|
702 |
|
|
}
|
703 |
|
|
}
|
704 |
|
|
return obsoletes;
|
705 |
|
|
}
|
706 |
7788
|
tao
|
|
707 |
|
|
/**
|
708 |
|
|
* Overwrite and do nothing
|
709 |
|
|
*/
|
710 |
|
|
public boolean cancel() {
|
711 |
|
|
return true;
|
712 |
|
|
}
|
713 |
7691
|
tao
|
|
714 |
7606
|
tao
|
}
|