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

52 lines
1.5 KiB
Java
Raw Normal View History

package eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets;
import java.util.List;
import java.util.Set;
import java.util.stream.Collectors;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
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.RelatedDataset;
import eu.dnetlib.dhp.broker.oa.util.aggregators.withRels.ResultWithRelations;
import eu.dnetlib.dhp.schema.oaf.Dataset;
public abstract class AbstractEnrichMissingDataset
2020-06-11 11:25:18 +02:00
extends UpdateMatcher<eu.dnetlib.broker.objects.Dataset> {
public AbstractEnrichMissingDataset(final Topic topic) {
2020-06-12 09:47:55 +02:00
super(true,
rel -> topic,
(p, rel) -> p.getDatasets().add(rel),
rel -> rel.getInstances().get(0).getUrl());
}
2020-06-11 11:25:18 +02:00
protected abstract boolean filterByType(String relType);
@Override
2020-06-12 09:47:55 +02:00
protected final List<eu.dnetlib.broker.objects.Dataset> findDifferences(
2020-06-11 11:25:18 +02:00
final ResultWithRelations source,
2020-06-12 09:47:55 +02:00
final ResultWithRelations target) {
final Set<String> existingDatasets = target
2020-06-11 11:25:18 +02:00
.getDatasets()
.stream()
2020-06-11 11:25:18 +02:00
.filter(rel -> filterByType(rel.getRelType()))
.map(RelatedDataset::getRelDataset)
.map(Dataset::getId)
.collect(Collectors.toSet());
return source
2020-06-11 11:25:18 +02:00
.getDatasets()
.stream()
2020-06-11 11:25:18 +02:00
.filter(rel -> filterByType(rel.getRelType()))
.map(RelatedDataset::getRelDataset)
.filter(d -> !existingDatasets.contains(d.getId()))
.map(ConversionUtils::oafDatasetToBrokerDataset)
.collect(Collectors.toList());
}
}