1
|
package eu.dnetlib.resolver;
|
2
|
|
3
|
import com.googlecode.protobuf.format.JsonFormat;
|
4
|
import eu.dnetlib.data.collector.plugins.datacite.DataciteParser;
|
5
|
import eu.dnetlib.data.proto.dli.ScholixObjectProtos;
|
6
|
import eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions;
|
7
|
import eu.dnetlib.dli.DLIUtils;
|
8
|
import eu.dnetlib.dli.resolver.*;
|
9
|
import eu.dnetlib.dli.resolver.model.CompletionStatus;
|
10
|
import eu.dnetlib.dli.resolver.model.DLIObjectProvenance;
|
11
|
import eu.dnetlib.dli.resolver.model.DLIResolvedObject;
|
12
|
import eu.dnetlib.dli.resolver.model.ObjectProvisionMode;
|
13
|
import eu.dnetlib.dli.resolver.model.serializer.DLIResolverSerializer;
|
14
|
import eu.dnetlib.pid.resolver.AbstractPIDResolver;
|
15
|
import eu.dnetlib.resolver.parser.DMFResolverParser;
|
16
|
import eu.dnetlib.resolver.parser.PMFResolverParser;
|
17
|
import eu.dnetlib.resolver.parser.ScholixResolverParser;
|
18
|
import org.apache.commons.io.IOUtils;
|
19
|
import org.apache.commons.lang3.StringUtils;
|
20
|
import org.apache.commons.logging.Log;
|
21
|
import org.apache.commons.logging.LogFactory;
|
22
|
import org.junit.Assert;
|
23
|
import org.junit.Ignore;
|
24
|
import org.junit.Test;
|
25
|
import org.springframework.beans.factory.annotation.Autowired;
|
26
|
|
27
|
import java.io.IOException;
|
28
|
import java.io.InputStream;
|
29
|
import java.net.URISyntaxException;
|
30
|
import java.net.URL;
|
31
|
import java.time.LocalDateTime;
|
32
|
import java.util.Arrays;
|
33
|
|
34
|
/**
|
35
|
* Created by sandro on 9/12/16.
|
36
|
*/
|
37
|
|
38
|
|
39
|
//@RunWith(SpringJUnit4ClassRunner.class)
|
40
|
//@ContextConfiguration(classes = { ConfigurationTestConfig.class})
|
41
|
@Ignore
|
42
|
public class DLIParserTest extends DLIAbstractTest{
|
43
|
|
44
|
private static final Log log = LogFactory.getLog(DLIParserTest.class);
|
45
|
|
46
|
@Autowired
|
47
|
private DLIResolverSerializer serializer;
|
48
|
|
49
|
@Test
|
50
|
public void testDMFParser() throws IOException {
|
51
|
final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/transform/huge.xml");
|
52
|
String str = IOUtils.toString(resourceAsStream);
|
53
|
DMFResolverParser parser = new DMFResolverParser();
|
54
|
DLIResolvedObject object = parser.parseObject(str);
|
55
|
Assert.assertNotNull(object);
|
56
|
Assert.assertNotNull(object.getTitles());
|
57
|
Assert.assertNotNull(object.getDescription());
|
58
|
Assert.assertNotNull(object.getType());
|
59
|
System.out.println(object);
|
60
|
// object.setTitles(Arrays.asList("TITOLO 1", "TITOLO 2"));
|
61
|
// Assert.assertEquals(CompletionStatus.complete.toString(), object.getCompletionStatus());
|
62
|
// DLIResolvedObject nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordScholix.xml")));
|
63
|
// Assert.assertNull(nullObject);
|
64
|
// nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMFRecord.xml")));
|
65
|
// Assert.assertNull(nullObject);
|
66
|
|
67
|
}
|
68
|
|
69
|
|
70
|
|
71
|
@Test
|
72
|
public void testDOI(){
|
73
|
final String url="https://ssl.fao.org/glisapi/v1/pgrfas?doi=5gg0w";
|
74
|
|
75
|
DLIUtils.isValidDoi(url);
|
76
|
}
|
77
|
|
78
|
|
79
|
@Test
|
80
|
public void testResolvedDate() throws IOException {
|
81
|
final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMF.xml");
|
82
|
String str = IOUtils.toString(resourceAsStream);
|
83
|
|
84
|
PMFResolverParser parser = new PMFResolverParser();
|
85
|
DLIResolvedObject object = parser.parseObject(str);
|
86
|
Assert.assertNotNull(object.getResolvedDate());
|
87
|
}
|
88
|
|
89
|
|
90
|
@Test
|
91
|
public void testDataciteParser() throws Exception {
|
92
|
|
93
|
final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordOpenaire.xml");
|
94
|
String str = IOUtils.toString(resourceAsStream);
|
95
|
DMFResolverParser parser = new DMFResolverParser();
|
96
|
DLIResolvedObject object = parser.parseObject(str);
|
97
|
System.out.println(object.getType());
|
98
|
object.getRelations().forEach(it -> System.out.println(it.getTargetPID()));
|
99
|
Assert.assertTrue(object.getCompletionStatus().equals("complete"));
|
100
|
System.out.println("object = " + object);
|
101
|
|
102
|
final String replacingXML = serializer.serializeReplacingXML(str, object);
|
103
|
|
104
|
log.debug(replacingXML);
|
105
|
|
106
|
|
107
|
|
108
|
}
|
109
|
|
110
|
@Test
|
111
|
public void testScholixParser() throws IOException, URISyntaxException {
|
112
|
final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/ieda.xml");
|
113
|
|
114
|
String str = IOUtils.toString(resourceAsStream);
|
115
|
ScholixResolverParser parser = new ScholixResolverParser();
|
116
|
DLIResolvedObject object = parser.parseObject(str);
|
117
|
|
118
|
final ScholixObjectProtos.Scholix.Builder scholix = ScholixObjectProtos.Scholix.newBuilder();
|
119
|
|
120
|
scholix.addLinkproviderBuilder()
|
121
|
.setName(object.getDatasourceProvenance().get(0).getDatasource())
|
122
|
.addIdentifiersBuilder()
|
123
|
.setIdentifier(object.getDatasourceProvenance().get(0).getDatasourceId())
|
124
|
.setSchema("dnetIdentifier");
|
125
|
|
126
|
|
127
|
|
128
|
scholix.setRelationship(ScholixObjectProtos.ScholixRelationship.newBuilder()
|
129
|
.setName(object.getRelations().get(0).getRelationSemantics())
|
130
|
.setInverse(object.getRelations().get(0).getInverseRelation())
|
131
|
.setSchema("datacite")
|
132
|
.build());
|
133
|
final CrossRefParserJSON parserJSON = new CrossRefParserJSON();
|
134
|
DLIResolvedObject dli_source = parserJSON.parseRecord(IOUtils.toString(new URL("https://api.crossref.org/works/10.1002/2016JG003630").openStream()));
|
135
|
DLIResolvedObject dli_target = DataciteResolver.parseResponse(AbstractPIDResolver.requestURL("http://search.datacite.org/api?wt=json&q=doi:10.6073/pasta/6038bd9745347610df8a51c92ca0e452"));
|
136
|
final ScholixObjectProtos.ScholixResource source = generateResource(dli_source);
|
137
|
final ScholixObjectProtos.ScholixResource target = generateResource(dli_target);
|
138
|
scholix.setSource(source);
|
139
|
scholix.setTarget(target);
|
140
|
scholix.setPublicationDate(LocalDateTime.now().toString());
|
141
|
System.out.println(JsonFormat.printToString(scholix.build()));
|
142
|
|
143
|
System.out.println(generateIdentifier(dli_source, dli_target));
|
144
|
|
145
|
|
146
|
}
|
147
|
|
148
|
private String generateIdentifier(final DLIResolvedObject source, DLIResolvedObject target) {
|
149
|
|
150
|
return AbstractDNetXsltFunctions.md5(String.format("%s::%s", source.getPid().toLowerCase().trim(), target.getPid().toLowerCase().trim()));
|
151
|
|
152
|
}
|
153
|
|
154
|
private ScholixObjectProtos.ScholixResource generateResource(DLIResolvedObject result) {
|
155
|
final ScholixObjectProtos.ScholixResource.Builder builder = ScholixObjectProtos.ScholixResource.newBuilder();
|
156
|
if (result.getDatasourceProvenance() != null)
|
157
|
result.getDatasourceProvenance().forEach(
|
158
|
objectProvenance -> {
|
159
|
builder.addCollectedFrom(ScholixObjectProtos.ScholixCollectedFrom.newBuilder()
|
160
|
.setProvisionMode(((DLIObjectProvenance) objectProvenance).getProvisionMode())
|
161
|
.setCompletionStatus(((DLIObjectProvenance) objectProvenance).getCompletionStatus())
|
162
|
.setProvider(ScholixObjectProtos.ScholixEntityId.newBuilder()
|
163
|
.setName(objectProvenance.getDatasource())
|
164
|
.addIdentifiers(ScholixObjectProtos.ScholixIdentifier.newBuilder().setIdentifier(objectProvenance.getDatasourceId())
|
165
|
.setSchema("dnetIdentifier").build())
|
166
|
.build()));
|
167
|
if (StringUtils.isNotEmpty(((DLIObjectProvenance) objectProvenance).getPublisher())) {
|
168
|
builder.addPublisher(ScholixObjectProtos.ScholixEntityId.newBuilder()
|
169
|
.setName(((DLIObjectProvenance) objectProvenance).getPublisher())
|
170
|
.build());
|
171
|
}
|
172
|
|
173
|
});
|
174
|
builder.addIdentifier(ScholixObjectProtos.ScholixIdentifier.newBuilder().
|
175
|
setIdentifier(result.getPid())
|
176
|
.setSchema(result.getPidType())
|
177
|
.build());
|
178
|
builder.setObjectType(result.getType().toString());
|
179
|
if (result.getTitles() != null && result.getTitles().size() > 0)
|
180
|
builder.setTitle(result.getTitles().get(0));
|
181
|
if (result.getAuthors() != null)
|
182
|
result.getAuthors().forEach(author -> builder.addCreator(
|
183
|
ScholixObjectProtos.ScholixEntityId.newBuilder()
|
184
|
.setName(author)
|
185
|
.build()));
|
186
|
if (StringUtils.isNotBlank(result.getDate())) {
|
187
|
builder.setPublicationDate(result.getDate());
|
188
|
}
|
189
|
|
190
|
String tp = null;
|
191
|
|
192
|
switch (result.getType()) {
|
193
|
case dataset:
|
194
|
tp = "60";
|
195
|
break;
|
196
|
case unknown:
|
197
|
tp = "70";
|
198
|
break;
|
199
|
case publication:
|
200
|
tp = "50";
|
201
|
break;
|
202
|
}
|
203
|
builder.setDnetIdentifier(tp + "|dnet________::" + result.getIdentifier());
|
204
|
return builder.build();
|
205
|
}
|
206
|
|
207
|
@Test
|
208
|
public void testScholixCompleteParser() throws IOException {
|
209
|
final InputStream resourceAsStream = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordScholix.xml");
|
210
|
|
211
|
String str = IOUtils.toString(resourceAsStream);
|
212
|
ScholixResolverParser parser = new ScholixResolverParser();
|
213
|
DLIResolvedObject object = parser.parseObject(str);
|
214
|
|
215
|
log.debug("object = " + object);
|
216
|
|
217
|
Assert.assertNotNull(object);
|
218
|
log.debug("object.getCompletionStatus() = " + object.getCompletionStatus());
|
219
|
|
220
|
// Assert.assertTrue(object.getCompletionStatus().equals(CompletionStatus.complete.toString()));
|
221
|
//
|
222
|
// Assert.assertNotNull(object.getDate());
|
223
|
// Assert.assertNotNull(object.getTitles());
|
224
|
// Assert.assertTrue(object.getTitles().size() == 1);
|
225
|
//
|
226
|
// Assert.assertNotNull(object.getDatasourceProvenance());
|
227
|
// Assert.assertTrue(object.getDatasourceProvenance().size() == 1);
|
228
|
// Assert.assertNotNull(object.getDatasourceProvenance().get(0).getPublisher());
|
229
|
// Assert.assertNotNull(object.getDatasourceProvenance().get(0).getPublisherId());
|
230
|
//
|
231
|
//
|
232
|
// Assert.assertNotNull(object.getRelations());
|
233
|
//
|
234
|
// Assert.assertTrue(object.getRelations().size() == 1);
|
235
|
//
|
236
|
// Assert.assertNotNull(object.getRelations().get(0).getTargetPID().getId());
|
237
|
// Assert.assertNotNull(object.getRelations().get(0).getTargetPID().getType());
|
238
|
//
|
239
|
//
|
240
|
// DLIResolvedObject nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordDMF.xml")));
|
241
|
// Assert.assertNull(nullObject);
|
242
|
// nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMFRecord.xml")));
|
243
|
// Assert.assertNull(nullObject);
|
244
|
|
245
|
}
|
246
|
|
247
|
|
248
|
@Test
|
249
|
public void testPMFParser() throws Exception {
|
250
|
final InputStream resource = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputPMF.xml");
|
251
|
String xml = IOUtils.toString(resource);
|
252
|
|
253
|
PMFResolverParser parser = new PMFResolverParser();
|
254
|
final DLIResolvedObject resolvedObject = parser.parseObject(xml);
|
255
|
Assert.assertNotNull(resolvedObject);
|
256
|
|
257
|
Assert.assertNotNull(resolvedObject.getPidType());
|
258
|
log.debug("resolvedObject = " + resolvedObject);
|
259
|
|
260
|
Assert.assertNotNull(resolvedObject.getRelations());
|
261
|
resolvedObject.getRelations().forEach(it -> Assert.assertNotNull(it));
|
262
|
|
263
|
DLIResolvedObject nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordScholix.xml")));
|
264
|
Assert.assertNull(nullObject);
|
265
|
nullObject = parser.parseObject(IOUtils.toString(getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordDMF.xml")));
|
266
|
Assert.assertNull(nullObject);
|
267
|
}
|
268
|
|
269
|
@Test
|
270
|
public void testOpenAIREParser() throws Exception {
|
271
|
|
272
|
final InputStream resource = getClass().getResourceAsStream("/eu/dnetlib/dli/parser/InputRecordOpenaire.xml");
|
273
|
|
274
|
final String record = IOUtils.toString(resource);
|
275
|
|
276
|
final OpenAireParser parser = new OpenAireParser();
|
277
|
final DLIResolvedObject object = parser.parseObject(record);
|
278
|
object.setPid("od_______908::ff0270cae25acd9367ef26a53ee5e4a3");
|
279
|
object.setPidType("openaire");
|
280
|
|
281
|
final DLIObjectProvenance provenance = new DLIObjectProvenance();
|
282
|
provenance.setCompletionStatus(CompletionStatus.complete.toString());
|
283
|
provenance.setDatasourceId("dli::r3d100010778");
|
284
|
provenance.setDatasource("NCBI Nucleotide");
|
285
|
provenance.setProvisionMode(ObjectProvisionMode.resolved.toString());
|
286
|
object.setDatasourceProvenance(Arrays.asList(provenance));
|
287
|
|
288
|
log.debug(object);
|
289
|
|
290
|
}
|
291
|
|
292
|
@Test
|
293
|
public void testProteinParser() throws IOException {
|
294
|
final NCBINResolver resolver = new NCBINResolver();
|
295
|
resolver.setNCBINParser(new NCBINParser());
|
296
|
final DLIResolvedObject p02768 = resolver.resolve("U36380", "protein");
|
297
|
|
298
|
log.debug(p02768.getDescription());
|
299
|
Assert.assertNotNull(p02768);
|
300
|
Assert.assertNotNull(p02768.getDescription());
|
301
|
|
302
|
log.debug("p02768 = " + p02768);
|
303
|
|
304
|
}
|
305
|
|
306
|
|
307
|
|
308
|
|
309
|
|
310
|
|
311
|
|
312
|
|
313
|
|
314
|
}
|