dnet-hadoop/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/matchers/simple/EnrichMoreSubject.java

44 lines
1.3 KiB
Java
Raw Normal View History

2020-05-08 16:49:47 +02:00
package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.List;
2020-05-15 12:25:37 +02:00
import java.util.Set;
import java.util.stream.Collectors;
2020-05-13 12:00:27 +02:00
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
2020-05-15 12:25:37 +02:00
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
2020-06-11 11:25:18 +02:00
import eu.dnetlib.dhp.broker.oa.util.aggregators.withRels.ResultWithRelations;
2020-06-11 11:25:18 +02:00
public class EnrichMoreSubject extends UpdateMatcher<Pair<String, String>> {
2020-05-13 12:00:27 +02:00
public EnrichMoreSubject() {
2020-06-12 09:47:55 +02:00
super(true,
pair -> Topic.fromPath("ENRICH/MORE/SUBJECT/" + pair.getLeft()),
(p, pair) -> p.getSubjects().add(pair.getRight()),
pair -> pair.getLeft() + "::" + pair.getRight());
2020-05-13 12:00:27 +02:00
}
2020-05-13 12:00:27 +02:00
@Override
2020-06-12 09:47:55 +02:00
protected List<Pair<String, String>> findDifferences(final ResultWithRelations source,
final ResultWithRelations target) {
2020-05-15 12:25:37 +02:00
final Set<String> existingSubjects = target
2020-06-11 11:25:18 +02:00
.getResult()
2020-05-15 12:25:37 +02:00
.getSubject()
.stream()
.map(pid -> pid.getQualifier().getClassid() + "::" + pid.getValue())
.collect(Collectors.toSet());
return source
2020-06-11 11:25:18 +02:00
.getResult()
2020-05-15 12:25:37 +02:00
.getPid()
.stream()
.filter(pid -> !existingSubjects.contains(pid.getQualifier().getClassid() + "::" + pid.getValue()))
.map(ConversionUtils::oafSubjectToPair)
.collect(Collectors.toList());
}
}