2020-05-08 16:49:47 +02:00
|
|
|
|
2020-05-22 17:17:41 +02:00
|
|
|
package eu.dnetlib.dhp.broker.oa.matchers.simple;
|
2020-05-07 12:31:26 +02:00
|
|
|
|
2020-06-11 14:30:24 +02:00
|
|
|
import java.util.ArrayList;
|
2020-05-07 12:31:26 +02:00
|
|
|
import java.util.List;
|
2020-06-11 14:30:24 +02:00
|
|
|
import java.util.Set;
|
|
|
|
import java.util.stream.Collectors;
|
2020-05-07 12:31:26 +02:00
|
|
|
|
2020-05-13 12:00:27 +02:00
|
|
|
import eu.dnetlib.dhp.broker.model.Topic;
|
2020-05-22 17:17:41 +02:00
|
|
|
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
2020-06-11 11:25:18 +02:00
|
|
|
import eu.dnetlib.dhp.broker.oa.util.aggregators.withRels.ResultWithRelations;
|
2020-06-11 14:30:24 +02:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.Author;
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
|
2020-05-07 12:31:26 +02:00
|
|
|
|
2020-06-11 14:30:24 +02:00
|
|
|
public class EnrichMissingAuthorOrcid extends UpdateMatcher<String> {
|
2020-05-07 12:31:26 +02:00
|
|
|
|
2020-05-13 12:00:27 +02:00
|
|
|
public EnrichMissingAuthorOrcid() {
|
2020-06-12 09:47:55 +02:00
|
|
|
super(true,
|
|
|
|
aut -> Topic.ENRICH_MISSING_AUTHOR_ORCID,
|
|
|
|
(p, aut) -> p.getCreators().add(aut),
|
|
|
|
aut -> aut);
|
2020-05-07 12:31:26 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2020-06-12 09:47:55 +02:00
|
|
|
protected List<String> findDifferences(final ResultWithRelations source,
|
|
|
|
final ResultWithRelations target) {
|
2020-06-11 14:30:24 +02:00
|
|
|
|
|
|
|
final Set<String> existingOrcids = target
|
|
|
|
.getResult()
|
|
|
|
.getAuthor()
|
|
|
|
.stream()
|
|
|
|
.map(Author::getPid)
|
|
|
|
.flatMap(List::stream)
|
|
|
|
.filter(pid -> pid.getQualifier().getClassid().equalsIgnoreCase("orcid"))
|
|
|
|
.map(pid -> pid.getValue())
|
|
|
|
.collect(Collectors.toSet());
|
|
|
|
|
2020-06-12 09:47:55 +02:00
|
|
|
final List<String> list = new ArrayList<>();
|
2020-06-11 14:30:24 +02:00
|
|
|
|
|
|
|
for (final Author author : source.getResult().getAuthor()) {
|
|
|
|
final String name = author.getFullname();
|
|
|
|
|
|
|
|
for (final StructuredProperty pid : author.getPid()) {
|
|
|
|
if (pid.getQualifier().getClassid().equalsIgnoreCase("orcid")
|
|
|
|
&& !existingOrcids.contains(pid.getValue())) {
|
2020-06-12 09:47:55 +02:00
|
|
|
list.add(name + " [ORCID: " + pid.getValue() + "]");
|
2020-06-11 14:30:24 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return list;
|
2020-05-07 12:31:26 +02:00
|
|
|
}
|
|
|
|
}
|