1
|
package eu.dnetlib.data.mapreduce.hbase.bulktag;
|
2
|
|
3
|
import com.googlecode.protobuf.format.JsonFormat;
|
4
|
import eu.dnetlib.data.bulktag.CommunityConfiguration;
|
5
|
import eu.dnetlib.data.bulktag.CommunityConfigurationFactory;
|
6
|
import eu.dnetlib.data.proto.FieldTypeProtos;
|
7
|
import eu.dnetlib.data.proto.OafProtos;
|
8
|
import eu.dnetlib.data.proto.ResultProtos;
|
9
|
import org.apache.commons.io.IOUtils;
|
10
|
import org.apache.commons.lang3.StringUtils;
|
11
|
import org.apache.hadoop.mapreduce.Counter;
|
12
|
import org.apache.hadoop.mapreduce.Mapper;
|
13
|
import org.dom4j.DocumentException;
|
14
|
import org.junit.Before;
|
15
|
import org.junit.Test;
|
16
|
import org.junit.runner.RunWith;
|
17
|
import org.mockito.Mock;
|
18
|
import org.mockito.junit.MockitoJUnitRunner;
|
19
|
|
20
|
import java.io.IOException;
|
21
|
import java.util.HashSet;
|
22
|
import java.util.List;
|
23
|
import java.util.Set;
|
24
|
import java.util.stream.Collectors;
|
25
|
|
26
|
import static org.junit.Assert.*;
|
27
|
|
28
|
import static org.mockito.Mockito.*;
|
29
|
|
30
|
/**
|
31
|
* Created by miriam on 02/08/2018.
|
32
|
*/
|
33
|
@RunWith(MockitoJUnitRunner.class)
|
34
|
public class ResultTaggerTest {
|
35
|
|
36
|
private String xml;
|
37
|
|
38
|
private CommunityConfiguration cc;
|
39
|
|
40
|
@Mock
|
41
|
private Mapper.Context context;
|
42
|
|
43
|
@Mock
|
44
|
private Counter counter;
|
45
|
|
46
|
private ResultTagger resultTagger = new ResultTagger();
|
47
|
|
48
|
@Before
|
49
|
public void setUp() throws IOException, DocumentException {
|
50
|
xml = IOUtils.toString(getClass().getResourceAsStream("community_configuration.xml"));
|
51
|
|
52
|
when(context.getCounter(anyString(), anyString())).thenReturn(counter);
|
53
|
|
54
|
cc = CommunityConfigurationFactory.newInstance(xml);
|
55
|
assertEquals(cc.size(),4);
|
56
|
cc.getCommunityList().forEach(c -> assertTrue(StringUtils.isNoneBlank(c.getId())));
|
57
|
}
|
58
|
|
59
|
@Test
|
60
|
public void testResultTaggerSubject() throws IOException {
|
61
|
OafProtos.Oaf oaf = getOaf("oaf_subject.json");
|
62
|
|
63
|
assertTrue(oaf.getEntity().getResult().getMetadata().getContextList().isEmpty());
|
64
|
|
65
|
oaf = resultTagger.enrichContext(oaf, cc, context);
|
66
|
List<ResultProtos.Result.Context> contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
67
|
assertNotNull(contextList);
|
68
|
assertEquals(contextList.size(),1);
|
69
|
assertEquals(contextList.get(0).getId(),"ni");
|
70
|
|
71
|
}
|
72
|
|
73
|
@Test
|
74
|
public void testResultTaggerSubject2() throws IOException {
|
75
|
OafProtos.Oaf oaf = getOaf("oaf_subject_2.jason");
|
76
|
|
77
|
assertTrue(oaf.getEntity().getResult().getMetadata().getContextList().isEmpty());
|
78
|
|
79
|
oaf = resultTagger.enrichContext(oaf, cc, context);
|
80
|
List<ResultProtos.Result.Context> contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
81
|
assertNotNull(contextList);
|
82
|
assertEquals(contextList.size(),2);
|
83
|
Set<String> ids = new HashSet<>();
|
84
|
ids.addAll(contextList.stream().map(c->c.getId()).collect(Collectors.toList()));
|
85
|
assertTrue(ids.contains("mes"));
|
86
|
assertTrue(ids.contains("ni"));
|
87
|
System.out.println(oaf.toString());
|
88
|
}
|
89
|
|
90
|
@Test
|
91
|
public void testResultTaggerExistingContext() throws IOException {
|
92
|
OafProtos.Oaf oaf = getOaf("oaf_existingcontext.json");
|
93
|
|
94
|
assertFalse(oaf.getEntity().getResult().getMetadata().getContextList().isEmpty());
|
95
|
|
96
|
oaf = resultTagger.enrichContext(oaf, cc, context);
|
97
|
List<ResultProtos.Result.Context> contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
98
|
assertNotNull(contextList);
|
99
|
assertEquals(contextList.size(),1);
|
100
|
assertEquals(contextList.get(0).getId(),"ni");
|
101
|
assertEquals(contextList.get(0).getDataInfoCount(),1);
|
102
|
assertEquals(contextList.get(0).getDataInfoList().get(0).getInferenceprovenance(),"bulktagging::community");
|
103
|
|
104
|
}
|
105
|
|
106
|
@Test
|
107
|
public void testReadOaf() throws IOException {
|
108
|
|
109
|
assertNotNull(getOaf("oaf_existingcontext_withprovenance.json"));
|
110
|
}
|
111
|
|
112
|
@Test
|
113
|
public void testResultTaggerExistingContextWithProvenance() throws IOException{
|
114
|
OafProtos.Oaf oaf = getOaf("oaf_existingcontext_withprovenance.json");
|
115
|
|
116
|
List<ResultProtos.Result.Context> contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
117
|
assertFalse(contextList.isEmpty());
|
118
|
assertEquals(contextList.size(),1);
|
119
|
assertEquals(contextList.get(0).getId(),"ni");
|
120
|
assertEquals(contextList.get(0).getDataInfoCount(),1);
|
121
|
assertEquals(contextList.get(0).getDataInfoList().get(0).getInferenceprovenance(),"bulktagging::community");
|
122
|
oaf = resultTagger.enrichContext(oaf, cc, context);
|
123
|
contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
124
|
assertEquals(contextList.size(),1);
|
125
|
assertEquals(contextList.get(0).getId(),"ni");
|
126
|
assertEquals(contextList.get(0).getDataInfoCount(),1);
|
127
|
assertEquals(contextList.get(0).getDataInfoList().get(0).getInferenceprovenance(),"bulktagging::community");
|
128
|
}
|
129
|
|
130
|
@Test
|
131
|
public void testResultTaggerDatasource() throws IOException{
|
132
|
OafProtos.Oaf oaf = getOaf("oaf_datasource.json");
|
133
|
List<ResultProtos.Result.Context> contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
134
|
assertTrue(contextList.isEmpty());
|
135
|
oaf = resultTagger.enrichContext(oaf,cc,context);
|
136
|
contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
137
|
assertFalse(contextList.isEmpty());
|
138
|
assertEquals(contextList.size(),2);
|
139
|
Set<String> ids = new HashSet<>();
|
140
|
ids.addAll(contextList.stream().map(c->{
|
141
|
assertTrue(c.getDataInfoCount()== 1);
|
142
|
return c.getId();
|
143
|
}).collect(Collectors.toList()));
|
144
|
assertTrue(ids.contains("aginfra"));
|
145
|
assertTrue(ids.contains("ni"));
|
146
|
}
|
147
|
|
148
|
@Test
|
149
|
public void testResultTaggerDatasource2() throws IOException{
|
150
|
OafProtos.Oaf oaf = getOaf("oaf_datasource2.json");
|
151
|
List<ResultProtos.Result.Context> contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
152
|
assertTrue(contextList.isEmpty());
|
153
|
oaf = resultTagger.enrichContext(oaf,cc,context);
|
154
|
contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
155
|
assertFalse(contextList.isEmpty());
|
156
|
assertEquals(contextList.size(),3);
|
157
|
Set<String> ids = new HashSet<>();
|
158
|
ids.addAll(contextList.stream().map(c->{
|
159
|
assertTrue(c.getDataInfoCount()== 1);
|
160
|
return c.getId();
|
161
|
}).collect(Collectors.toList()));
|
162
|
assertTrue(ids.contains("aginfra"));
|
163
|
assertTrue(ids.contains("ni"));
|
164
|
assertTrue(ids.contains("mes"));
|
165
|
}
|
166
|
|
167
|
@Test
|
168
|
public void testResultTaggerDatasource3() throws IOException{
|
169
|
OafProtos.Oaf oaf = getOaf("oaf_existingcontext3.json");
|
170
|
List<ResultProtos.Result.Context> contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
171
|
assertFalse(contextList.isEmpty());
|
172
|
oaf = resultTagger.enrichContext(oaf,cc,context);
|
173
|
contextList = oaf.getEntity().getResult().getMetadata().getContextList();
|
174
|
assertFalse(contextList.isEmpty());
|
175
|
assertEquals(contextList.size(),1);
|
176
|
assertTrue(contextList.get(0).getId().equals("ni"));
|
177
|
assertTrue(contextList.get(0).getDataInfoCount() == 2);
|
178
|
Set<String> provenance = new HashSet<>();
|
179
|
|
180
|
final List<FieldTypeProtos.DataInfo> dataInfoList = contextList.get(0).getDataInfoList();
|
181
|
provenance.addAll(dataInfoList.stream().map(d->d.getInferenceprovenance()).collect(Collectors.toList()));
|
182
|
assertTrue(provenance.size()==2);
|
183
|
assertTrue(provenance.contains("bulktagging::community"));
|
184
|
|
185
|
}
|
186
|
|
187
|
private OafProtos.Oaf getOaf(String oafjson) throws IOException {
|
188
|
final String json = IOUtils.toString(getClass().getResourceAsStream(oafjson));
|
189
|
|
190
|
final OafProtos.Oaf.Builder oaf = OafProtos.Oaf.newBuilder();
|
191
|
JsonFormat.merge(json, oaf);
|
192
|
|
193
|
return oaf.build();
|
194
|
}
|
195
|
}
|