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

73 lines
2.1 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;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
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;
2020-06-09 16:01:31 +02:00
import eu.dnetlib.pace.config.DedupConfig;
public abstract class AbstractEnrichMissingDataset
2020-06-11 11:25:18 +02:00
extends UpdateMatcher<eu.dnetlib.broker.objects.Dataset> {
private final Topic topic;
public AbstractEnrichMissingDataset(final Topic topic) {
super(true);
this.topic = topic;
}
2020-06-11 11:25:18 +02:00
protected abstract boolean filterByType(String relType);
@Override
protected final List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(
2020-06-11 11:25:18 +02:00
final ResultWithRelations source,
final ResultWithRelations target,
2020-06-09 16:01:31 +02:00
final DedupConfig dedupConfig) {
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)
2020-06-09 16:01:31 +02:00
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
.collect(Collectors.toList());
}
protected final UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
final eu.dnetlib.broker.objects.Dataset highlightValue,
2020-06-11 11:25:18 +02:00
final ResultWithRelations source,
final ResultWithRelations target,
2020-06-09 16:01:31 +02:00
final DedupConfig dedupConfig) {
return new UpdateInfo<>(
getTopic(),
2020-06-11 11:25:18 +02:00
highlightValue, source, target,
(p, rel) -> p.getDatasets().add(rel),
2020-06-09 16:01:31 +02:00
rel -> rel.getInstances().get(0).getUrl(),
dedupConfig);
}
public Topic getTopic() {
return topic;
}
2020-06-11 11:25:18 +02:00
}