1
|
package eu.dnetlib.data.mapreduce.dedup;
|
2
|
|
3
|
import java.util.List;
|
4
|
import java.util.UUID;
|
5
|
|
6
|
import org.junit.Before;
|
7
|
import org.junit.Test;
|
8
|
|
9
|
import com.google.common.base.Predicate;
|
10
|
import com.google.common.collect.Iterables;
|
11
|
import com.google.common.collect.Lists;
|
12
|
|
13
|
import eu.dnetlib.data.mapreduce.util.OafEntityMerger;
|
14
|
import eu.dnetlib.data.mapreduce.util.OafTest;
|
15
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
16
|
import eu.dnetlib.data.proto.KindProtos.Kind;
|
17
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
18
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
19
|
import eu.dnetlib.data.proto.OafProtos.OafEntity.Builder;
|
20
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
21
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
22
|
|
23
|
public class OafMergeTest {
|
24
|
|
25
|
private List<Oaf> oafList;
|
26
|
|
27
|
private OafEntityMerger merger;
|
28
|
|
29
|
@Before
|
30
|
public void setUp() throws Exception {
|
31
|
|
32
|
merger = new OafEntityMerger();
|
33
|
oafList = Lists.newArrayList();
|
34
|
oafList.add(getOaf("0.1").setEntity(
|
35
|
getEntity("id_1", null, "pid_1").setResult(
|
36
|
Result.newBuilder().setMetadata(
|
37
|
Result.Metadata.newBuilder().setDateofacceptance(OafTest.sf("2012-01-01"))
|
38
|
.addTitle(OafTest.getStructuredproperty("vvvv Title", "main title", "dnet:dataCite_title"))
|
39
|
.setResulttype(OafTest.getQualifier("publication", "dnet:result_typologies"))))).build());
|
40
|
oafList.add(getOaf("0.1").setEntity(
|
41
|
getEntity("id_2", "originalId_2", "pid_2").setResult(
|
42
|
Result.newBuilder().setMetadata(
|
43
|
Result.Metadata.newBuilder().setDateofacceptance(OafTest.sf(""))
|
44
|
.addTitle(OafTest.getStructuredproperty("aaaa Title", "main title", "dnet:dataCite_title"))
|
45
|
.setResulttype(OafTest.getQualifier("publication", "dnet:result_typologies"))))).build());
|
46
|
oafList.add(getOaf("0.2").setEntity(
|
47
|
getEntity("id_3", "originalId_2", "pid_2").setResult(
|
48
|
Result.newBuilder().setMetadata(
|
49
|
Result.Metadata.newBuilder().addTitle(OafTest.getStructuredproperty("cccc Title", "sub title", "dnet:dataCite_title"))
|
50
|
.setResulttype(OafTest.getQualifier("publication", "dnet:result_typologies"))))).build());
|
51
|
|
52
|
oafList.add(getOaf("0.3").setEntity(
|
53
|
getEntity("id_$", null, "pid_3").setResult(
|
54
|
Result.newBuilder().setMetadata(
|
55
|
Result.Metadata.newBuilder().setPublisher(OafTest.sf("AMER CHEMICAL SOCXXXXXXXXXXXXXXXXX"))
|
56
|
.setResulttype(OafTest.getQualifier("publication", "dnet:result_typologies"))))).build());
|
57
|
oafList.add(getOaf("0.5").setEntity(
|
58
|
getEntity("id_5", null, null).setResult(
|
59
|
Result.newBuilder().setMetadata(
|
60
|
Result.Metadata.newBuilder().addTitle(OafTest.getStructuredproperty("hhhh title", "main title", "dnet:dataCite_title"))
|
61
|
.setPublisher(OafTest.sf("AMER CHEMICAL SOC X"))
|
62
|
.setResulttype(OafTest.getQualifier("publication", "dnet:result_typologies")).setStoragedate(OafTest.sf("2012-11-18"))
|
63
|
.setLanguage(OafTest.getQualifier("eng", "dnet:languages")).addDescription(OafTest.sf("original description")))))
|
64
|
.build());
|
65
|
oafList.add(getOaf("0.6").setEntity(
|
66
|
getEntity("id_6", null, "pid_6").setResult(
|
67
|
Result.newBuilder().setMetadata(
|
68
|
Result.Metadata.newBuilder().setResulttype(OafTest.getQualifier("publication", "dnet:result_typologies"))
|
69
|
.addDescription(OafTest.sf("new description"))))).build());
|
70
|
}
|
71
|
|
72
|
@Test
|
73
|
public void test_merge() {
|
74
|
|
75
|
Oaf.Builder builder = Oaf.newBuilder();
|
76
|
|
77
|
for (Oaf oaf : oafList) {
|
78
|
builder.mergeFrom(oaf);
|
79
|
}
|
80
|
|
81
|
Result.Metadata.Builder metadata = builder.getEntityBuilder().getResultBuilder().getMetadataBuilder();
|
82
|
Iterable<StructuredProperty> filter = Iterables.filter(metadata.getTitleList(), new Predicate<StructuredProperty>() {
|
83
|
|
84
|
@Override
|
85
|
public boolean apply(StructuredProperty sp) {
|
86
|
return sp.getQualifier() != null && sp.getQualifier().getClassname().equals("main title");
|
87
|
}
|
88
|
});
|
89
|
|
90
|
StructuredProperty last = Iterables.getLast(filter);
|
91
|
|
92
|
metadata.clearTitle().addAllTitle(Lists.newArrayList(last));
|
93
|
|
94
|
System.out.println(builder.build().toString());
|
95
|
}
|
96
|
|
97
|
@Test
|
98
|
public void test_merger() {
|
99
|
|
100
|
Oaf merge = merger.mergeEntities("id", oafList).build();
|
101
|
|
102
|
System.out.println(merge.toString());
|
103
|
}
|
104
|
|
105
|
// @Test
|
106
|
// public void test_sort() {
|
107
|
// Queue<Result> q = new PriorityQueue<Result>(3, DedupReducer.cmp);
|
108
|
// for (Oaf oaf : oafList) {
|
109
|
// q.add(oaf.getEntity().getResult());
|
110
|
// }
|
111
|
//
|
112
|
// while (!q.isEmpty()) {
|
113
|
// Result r = q.remove();
|
114
|
// List<StructuredProperty> titles = r.getMetadata().getTitleList();
|
115
|
// if (!titles.isEmpty()) {
|
116
|
// System.out.println(titles.get(0).getValue());
|
117
|
// }
|
118
|
// }
|
119
|
// }
|
120
|
|
121
|
private Oaf.Builder getOaf(String trust) {
|
122
|
return Oaf.newBuilder().setKind(Kind.entity).setDataInfo(OafTest.getDataInfo(trust)).setTimestamp(System.currentTimeMillis());
|
123
|
}
|
124
|
|
125
|
private OafEntity.Builder getEntity(String id, String originalId, String pid) {
|
126
|
Builder entity = OafEntity.newBuilder().setType(Type.result).setId(id).addOriginalId(originalId != null ? originalId : UUID.randomUUID().toString());
|
127
|
|
128
|
if (pid != null) {
|
129
|
entity.addPid(OafTest.getStructuredproperty(pid, "class", "scheme"));
|
130
|
}
|
131
|
|
132
|
return entity;
|
133
|
}
|
134
|
|
135
|
}
|