1
|
package eu.dnetlib.msro.workflows.nodes.dedup;
|
2
|
|
3
|
import java.io.StringReader;
|
4
|
import java.util.Iterator;
|
5
|
import java.util.List;
|
6
|
import java.util.Queue;
|
7
|
|
8
|
import com.google.common.collect.Lists;
|
9
|
import com.google.common.collect.Queues;
|
10
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
11
|
import eu.dnetlib.enabling.resultset.client.ResultSetClient;
|
12
|
import eu.dnetlib.enabling.resultset.factory.ResultSetFactory;
|
13
|
import eu.dnetlib.msro.workflows.graph.Arc;
|
14
|
import eu.dnetlib.msro.workflows.nodes.AsyncJobNode;
|
15
|
import eu.dnetlib.msro.workflows.nodes.utils.Similarity;
|
16
|
import eu.dnetlib.msro.workflows.nodes.utils.SimilarityMeshBuilder;
|
17
|
import eu.dnetlib.msro.workflows.procs.Env;
|
18
|
import eu.dnetlib.rmi.common.ResultSet;
|
19
|
import org.antlr.stringtemplate.StringTemplate;
|
20
|
import org.apache.commons.logging.Log;
|
21
|
import org.apache.commons.logging.LogFactory;
|
22
|
import org.dom4j.Document;
|
23
|
import org.dom4j.DocumentException;
|
24
|
import org.dom4j.Node;
|
25
|
import org.dom4j.io.SAXReader;
|
26
|
import org.springframework.beans.factory.annotation.Autowired;
|
27
|
import org.springframework.beans.factory.annotation.Required;
|
28
|
|
29
|
public class BuildSimilarityMeshJobNode extends AsyncJobNode {
|
30
|
|
31
|
private static final Log log = LogFactory.getLog(BuildSimilarityMeshJobNode.class);
|
32
|
|
33
|
/**
|
34
|
* The result set client factory.
|
35
|
*/
|
36
|
@Autowired
|
37
|
private ResultSetClient resultSetClient;
|
38
|
|
39
|
@Autowired
|
40
|
private ResultSetFactory resultSetFactory;
|
41
|
|
42
|
private StringTemplate similarity;
|
43
|
|
44
|
private String inputEprParam;
|
45
|
|
46
|
private String outputEprParam;
|
47
|
|
48
|
@Override
|
49
|
protected String execute(final Env env) throws Exception {
|
50
|
|
51
|
final ResultSet<?> inputRs = env.getAttribute(getInputEprParam(), ResultSet.class);
|
52
|
final Iterator<String> it = resultSetClient.iter(inputRs, String.class).iterator();
|
53
|
|
54
|
final Queue<Object> queue = Queues.newLinkedBlockingQueue();
|
55
|
final SAXReader reader = new SAXReader();
|
56
|
|
57
|
if (it.hasNext()) {
|
58
|
populateQueue(queue, reader, it.next());
|
59
|
}
|
60
|
|
61
|
final ResultSet<String> rsOut = resultSetFactory.createResultSet(() -> {
|
62
|
return new Iterator<String>() {
|
63
|
|
64
|
@Override
|
65
|
public boolean hasNext() {
|
66
|
synchronized (queue) {
|
67
|
return !queue.isEmpty();
|
68
|
}
|
69
|
}
|
70
|
|
71
|
@Override
|
72
|
public String next() {
|
73
|
synchronized (queue) {
|
74
|
final Object o = queue.poll();
|
75
|
while (queue.isEmpty() && it.hasNext()) {
|
76
|
populateQueue(queue, reader, it.next());
|
77
|
}
|
78
|
return buildSimilarity((Similarity) o);
|
79
|
}
|
80
|
}
|
81
|
|
82
|
@Override
|
83
|
public void remove() {
|
84
|
throw new UnsupportedOperationException();
|
85
|
}
|
86
|
};
|
87
|
});
|
88
|
|
89
|
env.setAttribute(getOutputEprParam(), rsOut);
|
90
|
|
91
|
return Arc.DEFAULT_ARC;
|
92
|
}
|
93
|
|
94
|
private void populateQueue(final Queue<Object> q, final SAXReader r, final String xml) {
|
95
|
try {
|
96
|
final Document d = r.read(new StringReader(xml));
|
97
|
final String groupid = d.valueOf("//FIELD[@name='id']");
|
98
|
final List<?> items = d.selectNodes("//FIELD[@name='group']/ITEM");
|
99
|
final String entitytype = d.valueOf("//FIELD[@name='entitytype']");
|
100
|
final List<String> group = Lists.newArrayList();
|
101
|
for (final Object id : items) {
|
102
|
group.add(((Node) id).getText());
|
103
|
}
|
104
|
// compute the full mesh
|
105
|
final Type type = Type.valueOf(entitytype);
|
106
|
|
107
|
final List<Similarity> mesh = SimilarityMeshBuilder.build(type, group);
|
108
|
// total += mesh.size();
|
109
|
if (log.isDebugEnabled()) {
|
110
|
log.debug(String.format("built mesh for group '%s', size %d", groupid, mesh.size()));
|
111
|
}
|
112
|
for (final Similarity s : mesh) {
|
113
|
if (log.isDebugEnabled()) {
|
114
|
log.debug(String.format("adding to queue: %s", s.toString()));
|
115
|
}
|
116
|
q.add(s);
|
117
|
}
|
118
|
} catch (final DocumentException e) {
|
119
|
log.error("invalid document: " + xml);
|
120
|
}
|
121
|
}
|
122
|
|
123
|
private String buildSimilarity(final Similarity s) {
|
124
|
final StringTemplate template = new StringTemplate(getSimilarity().getTemplate());
|
125
|
|
126
|
template.setAttribute("source", s.getPair().getKey());
|
127
|
template.setAttribute("target", s.getPair().getValue());
|
128
|
template.setAttribute("type", s.getType().toString());
|
129
|
|
130
|
final String res = template.toString();
|
131
|
return res;
|
132
|
}
|
133
|
|
134
|
public String getInputEprParam() {
|
135
|
return inputEprParam;
|
136
|
}
|
137
|
|
138
|
public void setInputEprParam(final String inputEprParam) {
|
139
|
this.inputEprParam = inputEprParam;
|
140
|
}
|
141
|
|
142
|
public String getOutputEprParam() {
|
143
|
return outputEprParam;
|
144
|
}
|
145
|
|
146
|
public void setOutputEprParam(final String outputEprParam) {
|
147
|
this.outputEprParam = outputEprParam;
|
148
|
}
|
149
|
|
150
|
public StringTemplate getSimilarity() {
|
151
|
return similarity;
|
152
|
}
|
153
|
|
154
|
@Required
|
155
|
public void setSimilarity(final StringTemplate similarity) {
|
156
|
this.similarity = similarity;
|
157
|
}
|
158
|
|
159
|
}
|