1
|
// $Id$
|
2
|
/**
|
3
|
* Copyright (C) 2015 EDIT
|
4
|
* European Distributed Institute of Taxonomy
|
5
|
* http://www.e-taxonomy.eu
|
6
|
*
|
7
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
8
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
9
|
*/
|
10
|
package org.cybertaxonomy.utis.store;
|
11
|
|
12
|
import java.io.File;
|
13
|
import java.io.FileWriter;
|
14
|
import java.io.IOException;
|
15
|
import java.util.Date;
|
16
|
|
17
|
import org.apache.commons.httpclient.util.DateParseException;
|
18
|
import org.apache.commons.httpclient.util.DateUtil;
|
19
|
import org.apache.commons.io.FileUtils;
|
20
|
import org.apache.lucene.index.CorruptIndexException;
|
21
|
import org.apache.lucene.index.IndexReader;
|
22
|
import org.apache.lucene.store.FSDirectory;
|
23
|
import org.openrdf.repository.sail.SailRepository;
|
24
|
import org.openrdf.sail.Sail;
|
25
|
import org.openrdf.sail.SailException;
|
26
|
|
27
|
import com.tinkerpop.blueprints.Graph;
|
28
|
import com.tinkerpop.blueprints.impls.neo4j2.Neo4j2Graph;
|
29
|
import com.tinkerpop.blueprints.oupls.sail.GraphSail;
|
30
|
import com.tinkerpop.blueprints.oupls.sail.SailLoader;
|
31
|
|
32
|
/**
|
33
|
* @author a.kohlbecker
|
34
|
* @date Oct 19, 2015
|
35
|
*
|
36
|
*/
|
37
|
public class Neo4jStore extends Store{
|
38
|
|
39
|
private Neo4j2Graph graph;
|
40
|
private Sail sail;
|
41
|
private SailRepository sailRepo;
|
42
|
private final static String STORE_TYPE = "neo4j";
|
43
|
|
44
|
// January 1, 1970, 00:00:00 GMT as starting time
|
45
|
private Date lastModified = new Date(0);
|
46
|
// private Date lastModified = new GregorianCalendar(2016, 10, 10).getTime(); // new Date(0);
|
47
|
|
48
|
|
49
|
|
50
|
public Neo4jStore(String storeName) throws Exception {
|
51
|
super(storeName);
|
52
|
}
|
53
|
|
54
|
|
55
|
/**
|
56
|
* {@inheritDoc}
|
57
|
* @throws SailException
|
58
|
*/
|
59
|
@Override
|
60
|
protected void initStoreEngine() throws Exception {
|
61
|
|
62
|
graph = new Neo4j2Graph(storeLocation.toString());
|
63
|
sail = new GraphSail<Neo4j2Graph>(graph);
|
64
|
sail.initialize();
|
65
|
sailRepo = new SailRepository(sail);
|
66
|
|
67
|
logger.info("Using Neo4jGraph store at " + storeLocation.toString());
|
68
|
logger.info("Neo4jGraph has " + sizeInfo());
|
69
|
}
|
70
|
|
71
|
/**
|
72
|
* {@inheritDoc}
|
73
|
*/
|
74
|
@Override
|
75
|
protected void stopStoreEngine() throws Exception {
|
76
|
sailRepo.shutDown();
|
77
|
sail.shutDown(); // should be none by the above command already
|
78
|
}
|
79
|
|
80
|
/**
|
81
|
* @throws Exception
|
82
|
*
|
83
|
*/
|
84
|
@Override
|
85
|
protected void load(File rdfFile) throws Exception {
|
86
|
|
87
|
SailLoader loader = new SailLoader(sail);
|
88
|
// loader.setBufferSize(100000); // TODO optimize?
|
89
|
logger.info("loading RDF/XML into Neo4jGraph store");
|
90
|
loader.load(rdfFile);
|
91
|
logger.info("loading RDF/XML done");
|
92
|
logger.info("Neo4jGraph has " + sizeInfo());
|
93
|
|
94
|
logger.info("rdf loaded into Neo4jGraph store at " + storeLocation);
|
95
|
}
|
96
|
|
97
|
|
98
|
public long countEdges() {
|
99
|
|
100
|
String indexName = graph.getRawGraph().index().getRelationshipAutoIndexer().getAutoIndex().getName();
|
101
|
return countIndexDocuments("relationship" + File.separator + indexName);
|
102
|
}
|
103
|
|
104
|
public long countVertexes() {
|
105
|
|
106
|
String indexName = graph.getRawGraph().index().getNodeAutoIndexer().getAutoIndex().getName();
|
107
|
return countIndexDocuments("node" + File.separator + indexName);
|
108
|
}
|
109
|
|
110
|
/**
|
111
|
* @param nodeAutoIndexName
|
112
|
* @return
|
113
|
*/
|
114
|
private long countIndexDocuments(String nodeAutoIndexName) {
|
115
|
File luceneFolder = new File(storeLocation, "index" + File.separator + "lucene");
|
116
|
File indexFolder = new File(luceneFolder, nodeAutoIndexName);
|
117
|
int cnt = 0;
|
118
|
IndexReader reader = null;
|
119
|
try {
|
120
|
reader = IndexReader.open(FSDirectory.open(indexFolder));
|
121
|
cnt = reader.numDocs();
|
122
|
} catch (CorruptIndexException e) {
|
123
|
logger.warn("CorruptIndexException", e);
|
124
|
} catch (IOException e) {
|
125
|
logger.warn("Lucene index can not be read, this is ok as long there store location is empty. Original error: " + e.getMessage());
|
126
|
} finally {
|
127
|
if(reader != null) {
|
128
|
try {
|
129
|
reader.close();
|
130
|
} catch (IOException e) {
|
131
|
/* IGNORE */
|
132
|
}
|
133
|
}
|
134
|
}
|
135
|
return cnt;
|
136
|
}
|
137
|
|
138
|
public String sizeInfo() {
|
139
|
return countEdges() + " edges, " + countVertexes() + " vertexes";
|
140
|
}
|
141
|
|
142
|
public Graph graph() {
|
143
|
return graph;
|
144
|
}
|
145
|
|
146
|
/**
|
147
|
* @return the sailRepo
|
148
|
*/
|
149
|
public SailRepository getSailRepo() {
|
150
|
return sailRepo;
|
151
|
}
|
152
|
|
153
|
/**
|
154
|
* @return the lastModified
|
155
|
*/
|
156
|
public Date getLastModified() {
|
157
|
|
158
|
try {
|
159
|
String lastModifiedString = FileUtils.readFileToString(lastModifiedFile());
|
160
|
lastModified = DateUtil.parseDate(lastModifiedString);
|
161
|
} catch (IOException e) {
|
162
|
logger.info("Could not read " + lastModifiedFile().getAbsolutePath());
|
163
|
} catch (DateParseException e) {
|
164
|
throw new RuntimeException("Error while parsing date in " + lastModifiedFile().getAbsolutePath(), e);
|
165
|
}
|
166
|
|
167
|
return lastModified;
|
168
|
}
|
169
|
|
170
|
/**
|
171
|
* @param lastModified the lastModified to set
|
172
|
* @throws IOException
|
173
|
*/
|
174
|
public void setLastModified(Date lastModified) throws IOException {
|
175
|
|
176
|
File lastModifiedFile = lastModifiedFile();
|
177
|
File updateLogFile = updateLogFile();
|
178
|
|
179
|
// write the timestamp as only content into the last modified file
|
180
|
FileUtils.write(lastModifiedFile, DateUtil.formatDate(lastModified));
|
181
|
|
182
|
// append a new line to the log file
|
183
|
FileWriter logFileWriter = new FileWriter(updateLogFile);
|
184
|
logFileWriter.append("\n").append(DateUtil.formatDate(lastModified));
|
185
|
logFileWriter.close();
|
186
|
|
187
|
// getLastModified() reads the date from the file
|
188
|
this.lastModified = getLastModified();
|
189
|
// test if file has been written correctly
|
190
|
if(lastModified.compareTo(lastModified) != 0){
|
191
|
logger.error("Sote timestamps differ - please check!");
|
192
|
}
|
193
|
logger.info(DateUtil.formatDate(lastModified) + " written to " + lastModifiedFile.getAbsolutePath());
|
194
|
}
|
195
|
|
196
|
/**
|
197
|
* @return
|
198
|
*/
|
199
|
private File lastModifiedFile() {
|
200
|
return new File(storeLocation, "LAST_IMPORT_DATE");
|
201
|
}
|
202
|
|
203
|
private File updateLogFile() {
|
204
|
return new File(storeLocation, "UPDATE.log");
|
205
|
}
|
206
|
|
207
|
/**
|
208
|
* {@inheritDoc}
|
209
|
*/
|
210
|
@Override
|
211
|
protected String storeType() {
|
212
|
return STORE_TYPE;
|
213
|
}
|
214
|
|
215
|
|
216
|
/**
|
217
|
* {@inheritDoc}
|
218
|
*/
|
219
|
@Override
|
220
|
protected String dataFileExtension() {
|
221
|
return "rdf";
|
222
|
}
|
223
|
|
224
|
|
225
|
}
|