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

67 lines
1.7 KiB
Java
Raw Normal View History

2020-05-15 12:25:37 +02:00
package eu.dnetlib.dhp.broker.oa.matchers.simple;
2020-05-15 12:25:37 +02:00
2020-07-01 08:42:39 +02:00
import java.util.ArrayList;
2020-05-15 12:25:37 +02:00
import java.util.List;
import java.util.Set;
import java.util.stream.Collectors;
2020-06-22 08:51:31 +02:00
import eu.dnetlib.broker.objects.OaBrokerMainEntity;
import eu.dnetlib.broker.objects.OaBrokerTypedValue;
2020-05-15 12:25:37 +02:00
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
2020-07-01 08:42:39 +02:00
import eu.dnetlib.dhp.broker.oa.util.BrokerConstants;
2020-05-15 12:25:37 +02:00
2020-06-22 08:51:31 +02:00
public class EnrichMissingSubject extends UpdateMatcher<OaBrokerTypedValue> {
2020-05-15 12:25:37 +02:00
public EnrichMissingSubject() {
2020-06-26 11:20:45 +02:00
super(20,
2020-12-15 14:54:11 +01:00
s -> {
switch (s.getType().toLowerCase()) {
case "acm":
return Topic.ENRICH_MISSING_SUBJECT_ACM;
case "arxiv":
return Topic.ENRICH_MISSING_SUBJECT_ARXIV;
case "ddc":
return Topic.ENRICH_MISSING_SUBJECT_DDC;
case "jel":
return Topic.ENRICH_MISSING_SUBJECT_JEL;
case "mesh":
return Topic.ENRICH_MISSING_SUBJECT_MESHEUROPMC;
case "rvk":
return Topic.ENRICH_MISSING_SUBJECT_RVK;
default:
return null;
}
},
2020-06-16 12:34:13 +02:00
(p, s) -> p.getSubjects().add(s),
s -> subjectAsString(s));
2020-05-15 12:25:37 +02:00
}
@Override
2020-06-22 08:51:31 +02:00
protected List<OaBrokerTypedValue> findDifferences(final OaBrokerMainEntity source,
final OaBrokerMainEntity target) {
2020-07-01 08:42:39 +02:00
if (target.getSubjects().size() >= BrokerConstants.MAX_LIST_SIZE) {
return new ArrayList<>();
}
2020-06-16 12:34:13 +02:00
final Set<String> existingSubject = target
.getSubjects()
2020-05-15 12:25:37 +02:00
.stream()
2020-06-16 12:34:13 +02:00
.map(s -> subjectAsString(s))
2020-05-15 12:25:37 +02:00
.collect(Collectors.toSet());
return source
2020-06-16 12:34:13 +02:00
.getSubjects()
2020-05-15 12:25:37 +02:00
.stream()
2020-06-16 12:34:13 +02:00
.filter(s -> !existingSubject.contains(subjectAsString(s)))
2020-05-15 12:25:37 +02:00
.collect(Collectors.toList());
}
2020-06-22 08:51:31 +02:00
private static String subjectAsString(final OaBrokerTypedValue s) {
2020-06-16 12:34:13 +02:00
return s.getType() + "::" + s.getValue();
}
2020-05-15 12:25:37 +02:00
}