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

60 lines
1.8 KiB
Java

package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.List;
import java.util.Set;
import java.util.stream.Collectors;
import eu.dnetlib.broker.objects.Instance;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.BrokerConstants;
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.broker.oa.util.aggregators.withRels.ResultWithRelations;
import eu.dnetlib.pace.config.DedupConfig;
public class EnrichMoreOpenAccess extends UpdateMatcher<Instance> {
public EnrichMoreOpenAccess() {
super(true);
}
@Override
protected List<UpdateInfo<Instance>> findUpdates(final ResultWithRelations source,
final ResultWithRelations target,
final DedupConfig dedupConfig) {
final Set<String> urls = target
.getResult()
.getInstance()
.stream()
.filter(i -> i.getAccessright().getClassid().equals(BrokerConstants.OPEN_ACCESS))
.map(i -> i.getUrl())
.flatMap(List::stream)
.collect(Collectors.toSet());
return source
.getResult()
.getInstance()
.stream()
.filter(i -> i.getAccessright().getClassid().equals(BrokerConstants.OPEN_ACCESS))
.map(ConversionUtils::oafInstanceToBrokerInstances)
.flatMap(List::stream)
.filter(i -> !urls.contains(i.getUrl()))
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
.collect(Collectors.toList());
}
public UpdateInfo<Instance> generateUpdateInfo(final Instance highlightValue,
final ResultWithRelations source,
final ResultWithRelations target,
final DedupConfig dedupConfig) {
return new UpdateInfo<>(
Topic.ENRICH_MORE_OA_VERSION,
highlightValue, source, target,
(p, i) -> p.getInstances().add(i),
Instance::getUrl, dedupConfig);
}
}