|
1
|
package eu.dnetlib.msro.workflows.nodes.hostedby;
|
|
2
|
|
|
3
|
import java.io.StringReader;
|
|
4
|
import java.util.function.UnaryOperator;
|
|
5
|
|
|
6
|
import eu.dnetlib.enabling.resultset.factory.ResultSetFactory;
|
|
7
|
import eu.dnetlib.msro.workflows.graph.Arc;
|
|
8
|
import eu.dnetlib.msro.workflows.nodes.SimpleJobNode;
|
|
9
|
import eu.dnetlib.msro.workflows.procs.Env;
|
|
10
|
import eu.dnetlib.rmi.common.ResultSet;
|
|
11
|
import org.dom4j.Document;
|
|
12
|
import org.dom4j.Element;
|
|
13
|
import org.dom4j.io.SAXReader;
|
|
14
|
import org.springframework.beans.factory.annotation.Autowired;
|
|
15
|
|
|
16
|
// TODO: Auto-generated Javadoc
|
|
17
|
|
|
18
|
/**
|
|
19
|
* The Class FindHostedByJonbNode.
|
|
20
|
*/
|
|
21
|
public class FindHostedByJobNode extends SimpleJobNode {
|
|
22
|
|
|
23
|
private final String unknown_repo_id = "openaire____::1256f046-bf1f-4afc-8b47-d0b147148b18";
|
|
24
|
/**
|
|
25
|
* The input epr param.
|
|
26
|
*/
|
|
27
|
private String inputEprParam;
|
|
28
|
/**
|
|
29
|
* The output epr param.
|
|
30
|
*/
|
|
31
|
private String outputEprParam;
|
|
32
|
/**
|
|
33
|
* The counters param.
|
|
34
|
*/
|
|
35
|
private String countersParam;
|
|
36
|
/**
|
|
37
|
* The mapped result set factory.
|
|
38
|
*/
|
|
39
|
@Autowired
|
|
40
|
private ResultSetFactory resultSetFactory;
|
|
41
|
|
|
42
|
/*
|
|
43
|
* (non-Javadoc)
|
|
44
|
*
|
|
45
|
* @see eu.dnetlib.msro.workflows.nodes.SimpleJobNode#execute(com.googlecode.sarasvati.NodeToken)
|
|
46
|
*/
|
|
47
|
@Override
|
|
48
|
protected String execute(final Env env) throws Exception {
|
|
49
|
final ResultSet<?> inputEpr = env.getAttribute(this.inputEprParam, ResultSet.class);
|
|
50
|
final HostedByCounters counters = new HostedByCounters();
|
|
51
|
|
|
52
|
final SAXReader reader = new SAXReader();
|
|
53
|
|
|
54
|
final UnaryOperator<String> hostedByMapFunction = input -> {
|
|
55
|
try {
|
|
56
|
final Document doc = reader.read(new StringReader(input));
|
|
57
|
final Element node = (Element) doc.selectSingleNode("//*[local-name()='hostedBy']");
|
|
58
|
if (node != null) {
|
|
59
|
final String hostedById = node.attributeValue("id");
|
|
60
|
if (!hostedById.equals(FindHostedByJobNode.this.unknown_repo_id)) {
|
|
61
|
counters.increaseCounter(hostedById);
|
|
62
|
}
|
|
63
|
}
|
|
64
|
} catch (final Exception e) {
|
|
65
|
|
|
66
|
}
|
|
67
|
|
|
68
|
return input;
|
|
69
|
};
|
|
70
|
|
|
71
|
final ResultSet<String> epr = this.resultSetFactory.map(inputEpr, String.class,
|
|
72
|
hostedByMapFunction);
|
|
73
|
|
|
74
|
env.setAttribute(this.outputEprParam, epr);
|
|
75
|
env.setAttribute(getCountersParam(), counters);
|
|
76
|
|
|
77
|
return Arc.DEFAULT_ARC;
|
|
78
|
}
|
|
79
|
|
|
80
|
/**
|
|
81
|
* @return the inputEprParam
|
|
82
|
*/
|
|
83
|
public String getInputEprParam() {
|
|
84
|
return this.inputEprParam;
|
|
85
|
}
|
|
86
|
|
|
87
|
/**
|
|
88
|
* @param inputEprParam the inputEprParam to set
|
|
89
|
*/
|
|
90
|
public void setInputEprParam(final String inputEprParam) {
|
|
91
|
this.inputEprParam = inputEprParam;
|
|
92
|
}
|
|
93
|
|
|
94
|
/**
|
|
95
|
* @return the outputEprParam
|
|
96
|
*/
|
|
97
|
public String getOutputEprParam() {
|
|
98
|
return this.outputEprParam;
|
|
99
|
}
|
|
100
|
|
|
101
|
/**
|
|
102
|
* @param outputEprParam the outputEprParam to set
|
|
103
|
*/
|
|
104
|
public void setOutputEprParam(final String outputEprParam) {
|
|
105
|
this.outputEprParam = outputEprParam;
|
|
106
|
}
|
|
107
|
|
|
108
|
/**
|
|
109
|
* @return the countersParam
|
|
110
|
*/
|
|
111
|
public String getCountersParam() {
|
|
112
|
return this.countersParam;
|
|
113
|
}
|
|
114
|
|
|
115
|
/**
|
|
116
|
* @param countersParam the countersParam to set
|
|
117
|
*/
|
|
118
|
public void setCountersParam(final String countersParam) {
|
|
119
|
this.countersParam = countersParam;
|
|
120
|
}
|
|
121
|
|
|
122
|
}
|