1 |
46233
|
sandro.lab
|
package eu.dnetlib.dli.proto;
|
2 |
|
|
|
3 |
46249
|
claudio.at
|
import java.util.*;
|
4 |
|
|
import java.util.stream.Collectors;
|
5 |
|
|
|
6 |
46238
|
sandro.lab
|
import com.googlecode.protobuf.format.JsonFormat;
|
7 |
46233
|
sandro.lab
|
import eu.dnetlib.data.proto.*;
|
8 |
|
|
import eu.dnetlib.data.proto.dli.DLIObjectProtos;
|
9 |
|
|
|
10 |
|
|
import static eu.dnetlib.data.proto.dli.DLIObjectProtos.DLIObjectSummary;
|
11 |
|
|
import static eu.dnetlib.data.proto.dli.DLIObjectProtos.TypedIdentifier;
|
12 |
|
|
|
13 |
|
|
/**
|
14 |
|
|
* Created by sandro on 3/13/17.
|
15 |
|
|
*/
|
16 |
|
|
|
17 |
|
|
public class DNGFDLISummaryConverter {
|
18 |
|
|
|
19 |
|
|
private DNGFProtos.DNGF mainDNGFEntity;
|
20 |
|
|
|
21 |
|
|
private int relatedDatasets = 0;
|
22 |
|
|
|
23 |
|
|
private int relatedPublications = 0;
|
24 |
|
|
|
25 |
|
|
private int relatedUnknown = 0;
|
26 |
|
|
|
27 |
|
|
|
28 |
|
|
public void setMainEntity(final DNGFProtos.DNGF mainEntity) {
|
29 |
|
|
this.mainDNGFEntity = mainEntity;
|
30 |
|
|
}
|
31 |
|
|
|
32 |
|
|
public void addRelation(final DNGFProtos.DNGF relation) {
|
33 |
|
|
final TypeProtos.Type targetType = relation.getRel().getTargetType();
|
34 |
|
|
|
35 |
|
|
switch (targetType) {
|
36 |
|
|
case dataset:
|
37 |
|
|
this.relatedDatasets++;
|
38 |
|
|
break;
|
39 |
|
|
case publication:
|
40 |
|
|
this.relatedPublications++;
|
41 |
|
|
break;
|
42 |
|
|
case unknown:
|
43 |
|
|
this.relatedUnknown++;
|
44 |
|
|
break;
|
45 |
|
|
}
|
46 |
|
|
|
47 |
|
|
}
|
48 |
|
|
|
49 |
46238
|
sandro.lab
|
public String convertAsJson() {
|
50 |
49184
|
sandro.lab
|
final DLIObjectSummary convert = convert();
|
51 |
|
|
if (convert != null)
|
52 |
|
|
return JsonFormat.printToString(convert);
|
53 |
|
|
return null;
|
54 |
46238
|
sandro.lab
|
}
|
55 |
|
|
|
56 |
46233
|
sandro.lab
|
public DLIObjectSummary convert() {
|
57 |
46249
|
claudio.at
|
if (!isValid()) {
|
58 |
46233
|
sandro.lab
|
return null;
|
59 |
|
|
}
|
60 |
|
|
|
61 |
|
|
final DNGFProtos.DNGFEntity entity = mainDNGFEntity.getEntity();
|
62 |
|
|
final DLIObjectSummary.Builder summary = DLIObjectSummary.newBuilder();
|
63 |
46249
|
claudio.at
|
|
64 |
46233
|
sandro.lab
|
summary.setId(entity.getId());
|
65 |
|
|
final List<FieldTypeProtos.StructuredProperty> identifiers = entity.getExtension(DliProtos.typedIdentifier);
|
66 |
|
|
if (identifiers == null || identifiers.isEmpty()) {
|
67 |
49184
|
sandro.lab
|
this.mainDNGFEntity = null;
|
68 |
|
|
return null;
|
69 |
|
|
// throw new IllegalStateException(String.format("missing typedIdentifiers on main Entity \n%s", entity));
|
70 |
46233
|
sandro.lab
|
}
|
71 |
|
|
|
72 |
|
|
//ADD ALL IDENTIFIERS
|
73 |
|
|
identifiers.stream()
|
74 |
|
|
.collect(
|
75 |
|
|
Collectors.groupingBy(s -> String.format("%s::%s", s.getValue().toLowerCase().trim(), s.getQualifier().getClassid().toLowerCase().trim())))
|
76 |
|
|
.values()
|
77 |
|
|
.forEach(
|
78 |
|
|
v -> summary.addLocalIdentifier(
|
79 |
|
|
TypedIdentifier.newBuilder()
|
80 |
|
|
.setId(v.get(0).getValue())
|
81 |
|
|
.setType(v.get(0).getQualifier().getClassid())
|
82 |
|
|
));
|
83 |
|
|
|
84 |
|
|
|
85 |
|
|
// ADD ALL DATASOURCES
|
86 |
|
|
final Map<String, DLIObjectProtos.CollectedFromType> datasources = new HashMap<>();
|
87 |
|
|
|
88 |
|
|
|
89 |
|
|
final List<FieldTypeProtos.KeyValue> collectedFromList = entity.getCollectedfromList();
|
90 |
|
|
if (collectedFromList != null) {
|
91 |
46549
|
sandro.lab
|
collectedFromList.forEach(collectedFromValue -> {
|
92 |
|
|
try {
|
93 |
|
|
datasources.put(collectedFromValue.getKey(),
|
94 |
|
|
DLIObjectProtos.CollectedFromType.newBuilder()
|
95 |
|
|
.setDatasourceId(collectedFromValue.getKey())
|
96 |
|
|
.setDatasourceName(collectedFromValue.getValue())
|
97 |
|
|
.setCompletionStatus(DLIObjectProtos.CompletionStatus.valueOf(collectedFromValue.getExtension(DliFieldTypeProtos.completionStatus)))
|
98 |
|
|
.build());
|
99 |
|
|
} catch (IllegalArgumentException e) {
|
100 |
|
|
System.out.println("Error on generate datasource from record " + entity.toString());
|
101 |
|
|
throw new RuntimeException(e);
|
102 |
|
|
}
|
103 |
|
|
}
|
104 |
46233
|
sandro.lab
|
);
|
105 |
|
|
}
|
106 |
|
|
final List<FieldTypeProtos.KeyValue> resolvedFromList = entity.getExtension(DliProtos.resolvedfrom);
|
107 |
|
|
if (resolvedFromList != null) {
|
108 |
|
|
Map<String, FieldTypeProtos.KeyValue> setRFrom = new HashMap<>();
|
109 |
|
|
resolvedFromList.forEach(resolvedFromValue -> setRFrom.put(resolvedFromValue.getKey(), resolvedFromValue));
|
110 |
|
|
setRFrom.values().forEach(r ->
|
111 |
|
|
datasources.put(r.getKey(),
|
112 |
|
|
DLIObjectProtos.CollectedFromType.newBuilder()
|
113 |
|
|
.setCompletionStatus(DLIObjectProtos.CompletionStatus.complete)
|
114 |
|
|
.setDatasourceId(r.getKey())
|
115 |
|
|
.setDatasourceName(r.getValue())
|
116 |
|
|
.build())
|
117 |
|
|
);
|
118 |
|
|
}
|
119 |
|
|
summary.addAllDatasources(datasources.values());
|
120 |
|
|
|
121 |
46249
|
claudio.at
|
final Set<String> authorList = DngfDliUtils.parseAuthor(entity);
|
122 |
|
|
|
123 |
46233
|
sandro.lab
|
switch (entity.getType()) {
|
124 |
|
|
case dataset:
|
125 |
|
|
summary.setTypology(DLIObjectProtos.Typology.dataset);
|
126 |
|
|
setMetadataFromDataset(entity.getDataset(), summary, authorList);
|
127 |
|
|
break;
|
128 |
|
|
case publication:
|
129 |
|
|
summary.setTypology(DLIObjectProtos.Typology.publication);
|
130 |
|
|
setMetadataFromPublication(entity.getPublication(), summary, authorList);
|
131 |
|
|
break;
|
132 |
|
|
case unknown:
|
133 |
|
|
summary.setTypology(DLIObjectProtos.Typology.unknown);
|
134 |
|
|
break;
|
135 |
|
|
default:
|
136 |
|
|
throw new IllegalStateException(String.format("wrong Type for mainEntity: \n%s", entity));
|
137 |
|
|
}
|
138 |
|
|
|
139 |
47293
|
sandro.lab
|
summary.addAllPublisher(DngfDliUtils.parsePublishers(mainDNGFEntity.getEntity()));
|
140 |
46233
|
sandro.lab
|
summary.setRelatedPublications(relatedPublications);
|
141 |
|
|
summary.setRelatedDatasets(relatedDatasets);
|
142 |
|
|
summary.setRelatedUnknown(relatedUnknown);
|
143 |
|
|
|
144 |
|
|
return summary.build();
|
145 |
|
|
}
|
146 |
|
|
|
147 |
|
|
|
148 |
|
|
|
149 |
|
|
private void setMetadataFromDataset(final DatasetProtos.Dataset dataset, final DLIObjectSummary.Builder metadata, final Set<String> authorSet) {
|
150 |
|
|
final List<FieldTypeProtos.StructuredProperty> titleList = dataset.getMetadata().getTitleList();
|
151 |
|
|
//ADD Titles
|
152 |
|
|
if (titleList != null) {
|
153 |
|
|
titleList.forEach(title -> metadata.addTitle(title.getValue()));
|
154 |
|
|
}
|
155 |
|
|
|
156 |
|
|
//ADD AUTHORS
|
157 |
|
|
final List<PersonProtos.Person> authorList = dataset.getAuthorList();
|
158 |
|
|
if (authorList != null) {
|
159 |
|
|
authorList.forEach(person -> authorSet.add(person.getMetadata().getFullname().getValue()));
|
160 |
|
|
}
|
161 |
|
|
|
162 |
|
|
if (dataset.getMetadata().getContributorList() != null)
|
163 |
|
|
dataset.getMetadata().getContributorList().forEach(c -> authorSet.add(c.getValue()));
|
164 |
|
|
|
165 |
|
|
metadata.addAllAuthor(authorSet);
|
166 |
|
|
|
167 |
|
|
//ADD Abstract
|
168 |
|
|
manageDatasetDates(dataset, metadata);
|
169 |
|
|
if (dataset.getMetadata().getDescriptionList() != null) {
|
170 |
|
|
dataset.getMetadata().getDescriptionList().forEach(desc -> metadata.setAbstract(desc.getValue()));
|
171 |
|
|
}
|
172 |
|
|
|
173 |
|
|
//ADD Subjects
|
174 |
|
|
if (dataset.getMetadata().getSubjectList() != null) {
|
175 |
|
|
dataset.getMetadata().getSubjectList().forEach(
|
176 |
|
|
subject -> metadata.addSubject(
|
177 |
|
|
DLIObjectProtos.SchemeValue.newBuilder()
|
178 |
|
|
.setValue(subject.getValue())
|
179 |
|
|
.setScheme(subject.getQualifier().getClassid())
|
180 |
|
|
.build()
|
181 |
|
|
)
|
182 |
|
|
);
|
183 |
|
|
}
|
184 |
|
|
|
185 |
|
|
|
186 |
|
|
}
|
187 |
|
|
|
188 |
|
|
private void setMetadataFromPublication(final PublicationProtos.Publication publication, final DLIObjectSummary.Builder metadata, final Set<String> authorSet) {
|
189 |
|
|
final List<FieldTypeProtos.StructuredProperty> titleList = publication.getMetadata().getTitleList();
|
190 |
|
|
//ADD Titles
|
191 |
|
|
if (titleList != null) {
|
192 |
|
|
titleList.forEach(title -> metadata.addTitle(title.getValue()));
|
193 |
|
|
}
|
194 |
|
|
//ADD AUTHORS
|
195 |
|
|
final List<PersonProtos.Person> authorList = publication.getAuthorList();
|
196 |
|
|
if (authorList != null) {
|
197 |
|
|
authorList.forEach(person -> authorSet.add(person.getMetadata().getFullname().getValue()));
|
198 |
|
|
}
|
199 |
|
|
|
200 |
|
|
if (publication.getMetadata().getContributorList() != null)
|
201 |
|
|
publication.getMetadata().getContributorList().forEach(c -> authorSet.add(c.getValue()));
|
202 |
|
|
|
203 |
|
|
metadata.addAllAuthor(authorSet);
|
204 |
|
|
|
205 |
|
|
//ADD Abstract
|
206 |
|
|
managePublicationDates(publication, metadata);
|
207 |
|
|
if (publication.getMetadata().getDescriptionList() != null) {
|
208 |
|
|
publication.getMetadata().getDescriptionList().forEach(desc -> metadata.setAbstract(desc.getValue()));
|
209 |
|
|
}
|
210 |
|
|
|
211 |
|
|
//ADD Subjects
|
212 |
|
|
if (publication.getMetadata().getSubjectList() != null) {
|
213 |
|
|
publication.getMetadata().getSubjectList().forEach(
|
214 |
|
|
subject -> metadata.addSubject(
|
215 |
|
|
DLIObjectProtos.SchemeValue.newBuilder()
|
216 |
|
|
.setValue(subject.getValue())
|
217 |
|
|
.setScheme(subject.getQualifier().getClassid())
|
218 |
|
|
.build()
|
219 |
|
|
)
|
220 |
|
|
);
|
221 |
|
|
}
|
222 |
|
|
|
223 |
47293
|
sandro.lab
|
|
224 |
46233
|
sandro.lab
|
}
|
225 |
|
|
|
226 |
|
|
private void managePublicationDates(final PublicationProtos.Publication publication, final DLIObjectSummary.Builder metadata) {
|
227 |
|
|
if (publication.getMetadata().getDateofacceptance() != null) {
|
228 |
|
|
metadata.addDate(publication.getMetadata().getDateofacceptance().getValue());
|
229 |
|
|
}
|
230 |
|
|
|
231 |
|
|
if (publication.getMetadata().getRelevantdateList() != null && !publication.getMetadata().getRelevantdateList().isEmpty()) {
|
232 |
|
|
publication.getMetadata().getRelevantdateList().forEach(
|
233 |
|
|
date -> metadata.addDate(date.getValue())
|
234 |
|
|
);
|
235 |
|
|
}
|
236 |
|
|
}
|
237 |
|
|
|
238 |
|
|
private void manageDatasetDates(final DatasetProtos.Dataset dataset, final DLIObjectSummary.Builder metadata) {
|
239 |
|
|
if (dataset.getMetadata().getDateofacceptance() != null) {
|
240 |
|
|
metadata.addDate(dataset.getMetadata().getDateofacceptance().getValue());
|
241 |
|
|
}
|
242 |
|
|
|
243 |
|
|
if (dataset.getMetadata().getRelevantdateList() != null && !dataset.getMetadata().getRelevantdateList().isEmpty()) {
|
244 |
|
|
dataset.getMetadata().getRelevantdateList().forEach(
|
245 |
|
|
date -> metadata.addDate(date.getValue())
|
246 |
|
|
);
|
247 |
|
|
}
|
248 |
|
|
|
249 |
|
|
}
|
250 |
|
|
|
251 |
|
|
|
252 |
|
|
public boolean isValid() {
|
253 |
|
|
return mainDNGFEntity != null && !(this.relatedUnknown == 0 && this.relatedDatasets == 0 && this.relatedPublications == 0);
|
254 |
|
|
}
|
255 |
|
|
|
256 |
|
|
public DNGFProtos.DNGF getMainDNGFEntity() {
|
257 |
|
|
return mainDNGFEntity;
|
258 |
|
|
}
|
259 |
|
|
|
260 |
|
|
public void setMainDNGFEntity(DNGFProtos.DNGF mainDNGFEntity) {
|
261 |
|
|
this.mainDNGFEntity = mainDNGFEntity;
|
262 |
|
|
}
|
263 |
|
|
}
|