merge upstream

This commit is contained in:
Miriam Baglioni 2020-05-25 10:38:22 +02:00
commit d3d36647d2
45 changed files with 475 additions and 469 deletions

View File

@ -0,0 +1,69 @@
package eu.dnetlib.dhp.schema.common;
import java.util.Comparator;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
public class LicenseComparator implements Comparator<Qualifier> {
@Override
public int compare(Qualifier left, Qualifier right) {
if (left == null && right == null)
return 0;
if (left == null)
return 1;
if (right == null)
return -1;
String lClass = left.getClassid();
String rClass = right.getClassid();
if (lClass.equals(rClass))
return 0;
if (lClass.equals("OPEN SOURCE"))
return -1;
if (rClass.equals("OPEN SOURCE"))
return 1;
if (lClass.equals("OPEN"))
return -1;
if (rClass.equals("OPEN"))
return 1;
if (lClass.equals("6MONTHS"))
return -1;
if (rClass.equals("6MONTHS"))
return 1;
if (lClass.equals("12MONTHS"))
return -1;
if (rClass.equals("12MONTHS"))
return 1;
if (lClass.equals("EMBARGO"))
return -1;
if (rClass.equals("EMBARGO"))
return 1;
if (lClass.equals("RESTRICTED"))
return -1;
if (rClass.equals("RESTRICTED"))
return 1;
if (lClass.equals("CLOSED"))
return -1;
if (rClass.equals("CLOSED"))
return 1;
if (lClass.equals("UNKNOWN"))
return -1;
if (rClass.equals("UNKNOWN"))
return 1;
// Else (but unlikely), lexicographical ordering will do.
return lClass.compareTo(rClass);
}
}

View File

@ -2,6 +2,7 @@
package eu.dnetlib.dhp.schema.oaf; package eu.dnetlib.dhp.schema.oaf;
import java.io.Serializable; import java.io.Serializable;
import java.util.Objects;
public class Field<T> implements Serializable { public class Field<T> implements Serializable {
@ -39,6 +40,6 @@ public class Field<T> implements Serializable {
if (getClass() != obj.getClass()) if (getClass() != obj.getClass())
return false; return false;
Field<T> other = (Field<T>) obj; Field<T> other = (Field<T>) obj;
return getValue().equals(other.getValue()); return Objects.equals(getValue(), other.getValue());
} }
} }

View File

@ -106,6 +106,7 @@ public abstract class OafEntity extends Oaf implements Serializable {
.stream(lists) .stream(lists)
.filter(Objects::nonNull) .filter(Objects::nonNull)
.flatMap(List::stream) .flatMap(List::stream)
.filter(Objects::nonNull)
.distinct() .distinct()
.collect(Collectors.toList()); .collect(Collectors.toList());
} }

View File

@ -30,30 +30,30 @@ import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.application.ArgumentApplicationParser; import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.broker.model.Event; import eu.dnetlib.dhp.broker.model.Event;
import eu.dnetlib.dhp.broker.model.EventFactory; import eu.dnetlib.dhp.broker.model.EventFactory;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingAbstract;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingAuthorOrcid;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingDatasetIsReferencedBy;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingDatasetIsRelatedTo;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingDatasetIsSupplementedBy;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingDatasetIsSupplementedTo;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingDatasetReferences;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingOpenAccess;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingPid;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingProject;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingPublicationDate;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingPublicationIsReferencedBy;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingPublicationIsRelatedTo;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingPublicationIsSupplementedBy;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingPublicationIsSupplementedTo;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingPublicationReferences;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingSoftware;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMissingSubject;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMoreOpenAccess;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMorePid;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMoreProject;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMoreSoftware;
import eu.dnetlib.dhp.broker.oa.matchers.EnrichMoreSubject;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher; import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets.EnrichMissingDatasetIsReferencedBy;
import eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets.EnrichMissingDatasetIsRelatedTo;
import eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets.EnrichMissingDatasetIsSupplementedBy;
import eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets.EnrichMissingDatasetIsSupplementedTo;
import eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets.EnrichMissingDatasetReferences;
import eu.dnetlib.dhp.broker.oa.matchers.relatedProjects.EnrichMissingProject;
import eu.dnetlib.dhp.broker.oa.matchers.relatedProjects.EnrichMoreProject;
import eu.dnetlib.dhp.broker.oa.matchers.relatedPublications.EnrichMissingPublicationIsReferencedBy;
import eu.dnetlib.dhp.broker.oa.matchers.relatedPublications.EnrichMissingPublicationIsRelatedTo;
import eu.dnetlib.dhp.broker.oa.matchers.relatedPublications.EnrichMissingPublicationIsSupplementedBy;
import eu.dnetlib.dhp.broker.oa.matchers.relatedPublications.EnrichMissingPublicationIsSupplementedTo;
import eu.dnetlib.dhp.broker.oa.matchers.relatedPublications.EnrichMissingPublicationReferences;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMissingAbstract;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMissingAuthorOrcid;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMissingOpenAccess;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMissingPid;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMissingPublicationDate;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMissingSoftware;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMissingSubject;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMoreOpenAccess;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMorePid;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMoreSoftware;
import eu.dnetlib.dhp.broker.oa.matchers.simple.EnrichMoreSubject;
import eu.dnetlib.dhp.broker.oa.util.BrokerConstants; import eu.dnetlib.dhp.broker.oa.util.BrokerConstants;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.common.HdfsSupport; import eu.dnetlib.dhp.common.HdfsSupport;

View File

@ -1,38 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingDatasetIsReferencedBy
extends UpdateMatcher<Pair<Result, List<Dataset>>, eu.dnetlib.broker.objects.Dataset> {
public EnrichMissingDatasetIsReferencedBy() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
// TODO Auto-generated method stub
return null;
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
final eu.dnetlib.broker.objects.Dataset highlightValue,
final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_DATASET_IS_REFERENCED_BY,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> p.getDatasets().add(rel),
rel -> rel.getInstances().get(0).getUrl());
}
}

View File

@ -1,38 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingDatasetIsRelatedTo
extends UpdateMatcher<Pair<Result, List<Dataset>>, eu.dnetlib.broker.objects.Dataset> {
public EnrichMissingDatasetIsRelatedTo() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
// TODO Auto-generated method stub
return null;
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
final eu.dnetlib.broker.objects.Dataset highlightValue,
final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_DATASET_IS_RELATED_TO,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> p.getDatasets().add(rel),
rel -> rel.getInstances().get(0).getUrl());
}
}

View File

@ -1,38 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingDatasetIsSupplementedBy
extends UpdateMatcher<Pair<Result, List<Dataset>>, eu.dnetlib.broker.objects.Dataset> {
public EnrichMissingDatasetIsSupplementedBy() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
// TODO Auto-generated method stub
return null;
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
final eu.dnetlib.broker.objects.Dataset highlightValue,
final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_DATASET_IS_SUPPLEMENTED_BY,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> p.getDatasets().add(rel),
rel -> rel.getInstances().get(0).getUrl());
}
}

View File

@ -1,38 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingDatasetIsSupplementedTo
extends UpdateMatcher<Pair<Result, List<Dataset>>, eu.dnetlib.broker.objects.Dataset> {
public EnrichMissingDatasetIsSupplementedTo() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
// TODO Auto-generated method stub
return null;
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
final eu.dnetlib.broker.objects.Dataset highlightValue,
final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_DATASET_IS_SUPPLEMENTED_TO,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> p.getDatasets().add(rel),
rel -> rel.getInstances().get(0).getUrl());
}
}

View File

@ -1,38 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingDatasetReferences
extends UpdateMatcher<Pair<Result, List<Dataset>>, eu.dnetlib.broker.objects.Dataset> {
public EnrichMissingDatasetReferences() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
// TODO Auto-generated method stub
return null;
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
final eu.dnetlib.broker.objects.Dataset highlightValue,
final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_DATASET_REFERENCES,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> p.getDatasets().add(rel),
rel -> rel.getInstances().get(0).getUrl());
}
}

View File

@ -1,42 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.Arrays;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingPublicationIsReferencedBy
extends UpdateMatcher<Pair<Result, List<Publication>>, eu.dnetlib.broker.objects.Publication> {
public EnrichMissingPublicationIsReferencedBy() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Publication>> findUpdates(
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
// TODO Auto-generated method stub
return Arrays.asList();
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Publication> generateUpdateInfo(
final eu.dnetlib.broker.objects.Publication highlightValue,
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_PUBLICATION_IS_REFERENCED_BY,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> {
}, // p.getPublications().add(rel), //TODO available in the future release of dnet-openaire-broker-common
rel -> rel.getOriginalId());
}
}

View File

@ -1,42 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.Arrays;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingPublicationIsRelatedTo
extends UpdateMatcher<Pair<Result, List<Publication>>, eu.dnetlib.broker.objects.Publication> {
public EnrichMissingPublicationIsRelatedTo() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Publication>> findUpdates(
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
// TODO Auto-generated method stub
return Arrays.asList();
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Publication> generateUpdateInfo(
final eu.dnetlib.broker.objects.Publication highlightValue,
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_PUBLICATION_IS_RELATED_TO,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> {
}, // p.getPublications().add(rel), //TODO available in the future release of dnet-openaire-broker-common
rel -> rel.getOriginalId());
}
}

View File

@ -1,42 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.Arrays;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingPublicationIsSupplementedBy
extends UpdateMatcher<Pair<Result, List<Publication>>, eu.dnetlib.broker.objects.Publication> {
public EnrichMissingPublicationIsSupplementedBy() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Publication>> findUpdates(
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
// TODO Auto-generated method stub
return Arrays.asList();
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Publication> generateUpdateInfo(
final eu.dnetlib.broker.objects.Publication highlightValue,
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_PUBLICATION_IS_SUPPLEMENTED_BY,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> {
}, // p.getPublications().add(rel), //TODO available in the future release of dnet-openaire-broker-common
rel -> rel.getOriginalId());
}
}

View File

@ -1,42 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.Arrays;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingPublicationIsSupplementedTo
extends UpdateMatcher<Pair<Result, List<Publication>>, eu.dnetlib.broker.objects.Publication> {
public EnrichMissingPublicationIsSupplementedTo() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Publication>> findUpdates(
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
// TODO Auto-generated method stub
return Arrays.asList();
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Publication> generateUpdateInfo(
final eu.dnetlib.broker.objects.Publication highlightValue,
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_PUBLICATION_IS_SUPPLEMENTED_TO,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> {
}, // p.getPublications().add(rel), //TODO available in the future release of dnet-openaire-broker-common
rel -> rel.getOriginalId());
}
}

View File

@ -1,42 +0,0 @@
package eu.dnetlib.dhp.broker.oa.matchers;
import java.util.Arrays;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Result;
public class EnrichMissingPublicationReferences
extends UpdateMatcher<Pair<Result, List<Publication>>, eu.dnetlib.broker.objects.Publication> {
public EnrichMissingPublicationReferences() {
super(true);
}
@Override
protected List<UpdateInfo<eu.dnetlib.broker.objects.Publication>> findUpdates(
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
// TODO Auto-generated method stub
return Arrays.asList();
}
@Override
protected UpdateInfo<eu.dnetlib.broker.objects.Publication> generateUpdateInfo(
final eu.dnetlib.broker.objects.Publication highlightValue,
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_PUBLICATION_REFERENCES,
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> {
}, // p.getPublications().add(rel), //TODO available in the future release of dnet-openaire-broker-common
rel -> rel.getOriginalId());
}
}

View File

@ -0,0 +1,63 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets;
import java.util.List;
import java.util.Set;
import java.util.stream.Collectors;
import org.apache.commons.lang3.tuple.Pair;
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;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Result;
public abstract class AbstractEnrichMissingDataset
extends UpdateMatcher<Pair<Result, List<Dataset>>, eu.dnetlib.broker.objects.Dataset> {
private final Topic topic;
public AbstractEnrichMissingDataset(final Topic topic) {
super(true);
this.topic = topic;
}
@Override
protected final List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(
final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
final Set<String> existingDatasets = target
.getRight()
.stream()
.map(Dataset::getId)
.collect(Collectors.toSet());
return source
.getRight()
.stream()
.filter(d -> !existingDatasets.contains(d.getId()))
.map(ConversionUtils::oafDatasetToBrokerDataset)
.map(i -> generateUpdateInfo(i, source, target))
.collect(Collectors.toList());
}
@Override
protected final UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
final eu.dnetlib.broker.objects.Dataset highlightValue,
final Pair<Result, List<Dataset>> source,
final Pair<Result, List<Dataset>> target) {
return new UpdateInfo<>(
getTopic(),
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> p.getDatasets().add(rel),
rel -> rel.getInstances().get(0).getUrl());
}
public Topic getTopic() {
return topic;
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingDatasetIsReferencedBy extends AbstractEnrichMissingDataset {
public EnrichMissingDatasetIsReferencedBy() {
super(Topic.ENRICH_MISSING_DATASET_IS_REFERENCED_BY);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingDatasetIsRelatedTo extends AbstractEnrichMissingDataset {
public EnrichMissingDatasetIsRelatedTo() {
super(Topic.ENRICH_MISSING_DATASET_IS_RELATED_TO);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingDatasetIsSupplementedBy extends AbstractEnrichMissingDataset {
public EnrichMissingDatasetIsSupplementedBy() {
super(Topic.ENRICH_MISSING_DATASET_IS_SUPPLEMENTED_BY);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingDatasetIsSupplementedTo extends AbstractEnrichMissingDataset {
public EnrichMissingDatasetIsSupplementedTo() {
super(Topic.ENRICH_MISSING_DATASET_IS_SUPPLEMENTED_TO);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedDatasets;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingDatasetReferences extends AbstractEnrichMissingDataset {
public EnrichMissingDatasetReferences() {
super(Topic.ENRICH_MISSING_DATASET_REFERENCES);
}
}

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.relatedProjects;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
@ -7,6 +7,7 @@ import java.util.List;
import org.apache.commons.lang3.tuple.Pair; import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic; import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Project; import eu.dnetlib.dhp.schema.oaf.Project;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.relatedProjects;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
@ -7,6 +7,7 @@ import java.util.List;
import org.apache.commons.lang3.tuple.Pair; import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic; import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Project; import eu.dnetlib.dhp.schema.oaf.Project;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -0,0 +1,64 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedPublications;
import java.util.List;
import java.util.Set;
import java.util.stream.Collectors;
import org.apache.commons.lang3.tuple.Pair;
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;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Result;
public abstract class AbstractEnrichMissingPublication
extends UpdateMatcher<Pair<Result, List<Publication>>, eu.dnetlib.broker.objects.Publication> {
private final Topic topic;
public AbstractEnrichMissingPublication(final Topic topic) {
super(true);
this.topic = topic;
}
@Override
protected final List<UpdateInfo<eu.dnetlib.broker.objects.Publication>> findUpdates(
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
final Set<String> existingPublications = target
.getRight()
.stream()
.map(Publication::getId)
.collect(Collectors.toSet());
return source
.getRight()
.stream()
.filter(d -> !existingPublications.contains(d.getId()))
.map(ConversionUtils::oafPublicationToBrokerPublication)
.map(i -> generateUpdateInfo(i, source, target))
.collect(Collectors.toList());
}
@Override
protected final UpdateInfo<eu.dnetlib.broker.objects.Publication> generateUpdateInfo(
final eu.dnetlib.broker.objects.Publication highlightValue,
final Pair<Result, List<Publication>> source,
final Pair<Result, List<Publication>> target) {
return new UpdateInfo<>(
getTopic(),
highlightValue, source.getLeft(), target.getLeft(),
(p, rel) -> {
}, // p.getPublications().add(rel), //TODO available in the future release of dnet-openaire-broker-common
rel -> rel.getInstances().get(0).getUrl());
}
public Topic getTopic() {
return topic;
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedPublications;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingPublicationIsReferencedBy extends AbstractEnrichMissingPublication {
public EnrichMissingPublicationIsReferencedBy() {
super(Topic.ENRICH_MISSING_PUBLICATION_IS_REFERENCED_BY);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedPublications;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingPublicationIsRelatedTo extends AbstractEnrichMissingPublication {
public EnrichMissingPublicationIsRelatedTo() {
super(Topic.ENRICH_MISSING_PUBLICATION_IS_RELATED_TO);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedPublications;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingPublicationIsSupplementedBy extends AbstractEnrichMissingPublication {
public EnrichMissingPublicationIsSupplementedBy() {
super(Topic.ENRICH_MISSING_PUBLICATION_IS_SUPPLEMENTED_BY);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedPublications;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingPublicationIsSupplementedTo extends AbstractEnrichMissingPublication {
public EnrichMissingPublicationIsSupplementedTo() {
super(Topic.ENRICH_MISSING_PUBLICATION_IS_SUPPLEMENTED_TO);
}
}

View File

@ -0,0 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers.relatedPublications;
import eu.dnetlib.dhp.broker.model.Topic;
public class EnrichMissingPublicationReferences extends AbstractEnrichMissingPublication {
public EnrichMissingPublicationReferences() {
super(Topic.ENRICH_MISSING_PUBLICATION_REFERENCES);
}
}

View File

@ -1,11 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.ArrayList; import java.util.ArrayList;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
import eu.dnetlib.dhp.broker.model.Topic; import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
@ -7,6 +7,7 @@ import java.util.List;
import org.apache.commons.lang3.tuple.Pair; import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic; import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
@ -7,6 +7,7 @@ import java.util.stream.Collectors;
import eu.dnetlib.broker.objects.Instance; import eu.dnetlib.broker.objects.Instance;
import eu.dnetlib.dhp.broker.model.Topic; 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.BrokerConstants;
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils; import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
@ -7,6 +7,7 @@ import java.util.stream.Collectors;
import eu.dnetlib.broker.objects.Pid; import eu.dnetlib.broker.objects.Pid;
import eu.dnetlib.dhp.broker.model.Topic; 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.ConversionUtils;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -1,11 +1,12 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.ArrayList; import java.util.ArrayList;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
import eu.dnetlib.dhp.broker.model.Topic; import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
@ -7,6 +7,7 @@ import java.util.List;
import org.apache.commons.lang3.tuple.Pair; import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic; import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.Software; import eu.dnetlib.dhp.schema.oaf.Software;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.List; import java.util.List;
import java.util.Set; import java.util.Set;
@ -8,6 +8,7 @@ import java.util.stream.Collectors;
import org.apache.commons.lang3.tuple.Pair; import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic; 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.ConversionUtils;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Qualifier; import eu.dnetlib.dhp.schema.oaf.Qualifier;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.List; import java.util.List;
import java.util.Set; import java.util.Set;
@ -7,6 +7,7 @@ import java.util.stream.Collectors;
import eu.dnetlib.broker.objects.Instance; import eu.dnetlib.broker.objects.Instance;
import eu.dnetlib.dhp.broker.model.Topic; 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.BrokerConstants;
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils; import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.List; import java.util.List;
import java.util.Set; import java.util.Set;
@ -7,6 +7,7 @@ import java.util.stream.Collectors;
import eu.dnetlib.broker.objects.Pid; import eu.dnetlib.broker.objects.Pid;
import eu.dnetlib.dhp.broker.model.Topic; 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.ConversionUtils;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.Arrays; import java.util.Arrays;
import java.util.List; import java.util.List;
@ -7,6 +7,7 @@ import java.util.List;
import org.apache.commons.lang3.tuple.Pair; import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic; import eu.dnetlib.dhp.broker.model.Topic;
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.Software; import eu.dnetlib.dhp.schema.oaf.Software;

View File

@ -1,5 +1,5 @@
package eu.dnetlib.dhp.broker.oa.matchers; package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.List; import java.util.List;
import java.util.Set; import java.util.Set;
@ -8,6 +8,7 @@ import java.util.stream.Collectors;
import org.apache.commons.lang3.tuple.Pair; import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.dhp.broker.model.Topic; 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.ConversionUtils;
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo; import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;

View File

@ -7,6 +7,8 @@ import org.apache.commons.lang3.tuple.Pair;
import eu.dnetlib.broker.objects.Instance; import eu.dnetlib.broker.objects.Instance;
import eu.dnetlib.broker.objects.Pid; import eu.dnetlib.broker.objects.Pid;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty; import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
public class ConversionUtils { public class ConversionUtils {
@ -33,4 +35,15 @@ public class ConversionUtils {
return Pair.of(sp.getQualifier().getClassid(), sp.getValue()); return Pair.of(sp.getQualifier().getClassid(), sp.getValue());
} }
public static final eu.dnetlib.broker.objects.Dataset oafDatasetToBrokerDataset(final Dataset d) {
final eu.dnetlib.broker.objects.Dataset res = new eu.dnetlib.broker.objects.Dataset();
// TODO
return res;
}
public static final eu.dnetlib.broker.objects.Publication oafPublicationToBrokerPublication(final Publication d) {
final eu.dnetlib.broker.objects.Publication res = new eu.dnetlib.broker.objects.Publication();
// TODO
return res;
}
} }

View File

@ -20,12 +20,7 @@ public class AuthorMerger {
public static List<Author> merge(List<List<Author>> authors) { public static List<Author> merge(List<List<Author>> authors) {
authors.sort(new Comparator<List<Author>>() { authors.sort((o1, o2) -> -Integer.compare(countAuthorsPids(o1), countAuthorsPids(o2)));
@Override
public int compare(List<Author> o1, List<Author> o2) {
return -Integer.compare(countAuthorsPids(o1), countAuthorsPids(o2));
}
});
List<Author> author = new ArrayList<>(); List<Author> author = new ArrayList<>();
@ -86,12 +81,21 @@ public class AuthorMerger {
.stream() .stream()
.map(ba -> new Tuple2<>(sim(ba, a._2()), ba)) .map(ba -> new Tuple2<>(sim(ba, a._2()), ba))
.max(Comparator.comparing(Tuple2::_1)); .max(Comparator.comparing(Tuple2::_1));
if (simAuthor.isPresent() && simAuthor.get()._1() > THRESHOLD) {
Author r = simAuthor.get()._2(); if (simAuthor.isPresent()) {
if (r.getPid() == null) { double th = THRESHOLD;
r.setPid(new ArrayList<>()); // increase the threshold if the surname is too short
if (simAuthor.get()._2().getSurname() != null
&& simAuthor.get()._2().getSurname().length() <= 3)
th = 0.99;
if (simAuthor.get()._1() > th) {
Author r = simAuthor.get()._2();
if (r.getPid() == null) {
r.setPid(new ArrayList<>());
}
r.getPid().add(a._1());
} }
r.getPid().add(a._1());
} }
}); });
} }
@ -99,7 +103,8 @@ public class AuthorMerger {
public static String pidToComparableString(StructuredProperty pid) { public static String pidToComparableString(StructuredProperty pid) {
return (pid.getQualifier() != null return (pid.getQualifier() != null
? pid.getQualifier().getClassid() != null ? pid.getQualifier().getClassid().toLowerCase() : "" ? pid.getQualifier().getClassid() != null ? pid.getQualifier().getClassid().toLowerCase() : ""
: "") + (pid.getValue() != null ? pid.getValue().toLowerCase() : ""); : "")
+ (pid.getValue() != null ? pid.getValue().toLowerCase() : "");
} }
public static int countAuthorsPids(List<Author> authors) { public static int countAuthorsPids(List<Author> authors) {
@ -120,9 +125,10 @@ public class AuthorMerger {
final Person pa = parse(a); final Person pa = parse(a);
final Person pb = parse(b); final Person pb = parse(b);
// if both are accurate (e.g. they have name and surname)
if (pa.isAccurate() & pb.isAccurate()) { if (pa.isAccurate() & pb.isAccurate()) {
return new JaroWinkler() return new JaroWinkler().score(normalize(pa.getSurnameString()), normalize(pb.getSurnameString())) * 0.5
.score(normalize(pa.getSurnameString()), normalize(pb.getSurnameString())); + new JaroWinkler().score(normalize(pa.getNameString()), normalize(pb.getNameString())) * 0.5;
} else { } else {
return new JaroWinkler() return new JaroWinkler()
.score(normalize(pa.getNormalisedFullname()), normalize(pb.getNormalisedFullname())); .score(normalize(pa.getNormalisedFullname()), normalize(pb.getNormalisedFullname()));

View File

@ -21,6 +21,7 @@ import scala.Tuple2;
public class EntityMergerTest implements Serializable { public class EntityMergerTest implements Serializable {
List<Tuple2<String, Publication>> publications; List<Tuple2<String, Publication>> publications;
List<Tuple2<String, Publication>> publications2;
String testEntityBasePath; String testEntityBasePath;
DataInfo dataInfo; DataInfo dataInfo;
@ -36,6 +37,7 @@ public class EntityMergerTest implements Serializable {
.getAbsolutePath(); .getAbsolutePath();
publications = readSample(testEntityBasePath + "/publication_merge.json", Publication.class); publications = readSample(testEntityBasePath + "/publication_merge.json", Publication.class);
publications2 = readSample(testEntityBasePath + "/publication_merge2.json", Publication.class);
pub_top = getTopPub(publications); pub_top = getTopPub(publications);
@ -90,6 +92,17 @@ public class EntityMergerTest implements Serializable {
assertEquals(AuthorMerger.countAuthorsPids(pub_merged.getAuthor()), 4); assertEquals(AuthorMerger.countAuthorsPids(pub_merged.getAuthor()), 4);
} }
@Test
public void publicationMergerTest2() throws InstantiationException, IllegalAccessException, IOException {
Publication pub_merged = DedupRecordFactory
.entityMerger(dedupId, publications2.iterator(), 0, dataInfo, Publication.class);
assertEquals(pub_merged.getAuthor().size(), 27);
// insert assertions here
}
public DataInfo setDI() { public DataInfo setDI() {
DataInfo dataInfo = new DataInfo(); DataInfo dataInfo = new DataInfo();
dataInfo.setTrust("0.9"); dataInfo.setTrust("0.9");

File diff suppressed because one or more lines are too long

View File

@ -10,23 +10,10 @@ import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.listFields;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.oaiIProvenance; import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.oaiIProvenance;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier; import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty; import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
import static eu.dnetlib.dhp.schema.common.ModelConstants.DATASET_DEFAULT_RESULTTYPE; import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import static eu.dnetlib.dhp.schema.common.ModelConstants.DNET_PID_TYPES; import static eu.dnetlib.dhp.schema.common.ModelConstants.DNET_ACCESS_MODES;
import static eu.dnetlib.dhp.schema.common.ModelConstants.IS_PRODUCED_BY;
import static eu.dnetlib.dhp.schema.common.ModelConstants.ORP_DEFAULT_RESULTTYPE;
import static eu.dnetlib.dhp.schema.common.ModelConstants.OUTCOME;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PRODUCES;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PUBLICATION_DEFAULT_RESULTTYPE;
import static eu.dnetlib.dhp.schema.common.ModelConstants.REPOSITORY_PROVENANCE_ACTIONS;
import static eu.dnetlib.dhp.schema.common.ModelConstants.RESULT_PROJECT;
import static eu.dnetlib.dhp.schema.common.ModelConstants.SOFTWARE_DEFAULT_RESULTTYPE;
import java.util.ArrayList; import java.util.*;
import java.util.Arrays;
import java.util.Date;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import org.apache.commons.lang3.StringUtils; import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document; import org.dom4j.Document;
@ -34,6 +21,7 @@ import org.dom4j.DocumentFactory;
import org.dom4j.DocumentHelper; import org.dom4j.DocumentHelper;
import org.dom4j.Node; import org.dom4j.Node;
import eu.dnetlib.dhp.schema.common.LicenseComparator;
import eu.dnetlib.dhp.schema.oaf.Author; import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.Context; import eu.dnetlib.dhp.schema.oaf.Context;
import eu.dnetlib.dhp.schema.oaf.DataInfo; import eu.dnetlib.dhp.schema.oaf.DataInfo;
@ -285,7 +273,9 @@ public abstract class AbstractMdRecordToOafMapper {
r.setCoverage(prepareCoverages(doc, info)); r.setCoverage(prepareCoverages(doc, info));
r.setContext(prepareContexts(doc, info)); r.setContext(prepareContexts(doc, info));
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy)); final List<Instance> instances = prepareInstances(doc, info, collectedFrom, hostedBy);
r.setInstance(instances);
r.setBestaccessright(getBestAccessRights(instances));
} }
private List<Context> prepareContexts(final Document doc, final DataInfo info) { private List<Context> prepareContexts(final Document doc, final DataInfo info) {
@ -368,6 +358,34 @@ public abstract class AbstractMdRecordToOafMapper {
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info); protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
protected static Qualifier getBestAccessRights(List<Instance> instanceList) {
if (instanceList != null) {
final Optional<Qualifier> min = instanceList
.stream()
.map(i -> i.getAccessright())
.min(new LicenseComparator());
final Qualifier rights = min.isPresent() ? min.get() : new Qualifier();
if (StringUtils.isBlank(rights.getClassid())) {
rights.setClassid(UNKNOWN);
}
if (StringUtils.isBlank(rights.getClassname())
|| UNKNOWN.equalsIgnoreCase(rights.getClassname())) {
rights.setClassname(NOT_AVAILABLE);
}
if (StringUtils.isBlank(rights.getSchemeid())) {
rights.setSchemeid(DNET_ACCESS_MODES);
}
if (StringUtils.isBlank(rights.getSchemename())) {
rights.setSchemename(DNET_ACCESS_MODES);
}
return rights;
}
return null;
}
private Journal prepareJournal(final Document doc, final DataInfo info) { private Journal prepareJournal(final Document doc, final DataInfo info) {
final Node n = doc.selectSingleNode("//oaf:journal"); final Node n = doc.selectSingleNode("//oaf:journal");
if (n != null) { if (n != null) {

View File

@ -85,8 +85,19 @@ public class MappersTest {
assertTrue(p.getSubject().size() > 0); assertTrue(p.getSubject().size() > 0);
assertTrue(StringUtils.isNotBlank(p.getJournal().getIssnOnline())); assertTrue(StringUtils.isNotBlank(p.getJournal().getIssnOnline()));
assertTrue(StringUtils.isNotBlank(p.getJournal().getName())); assertTrue(StringUtils.isNotBlank(p.getJournal().getName()));
assertTrue(p.getInstance().size() > 0);
assertNotNull(p.getInstance());
assertTrue(p.getInstance().size() > 0);
p
.getInstance()
.stream()
.forEach(i -> {
assertNotNull(i.getAccessright());
assertEquals("OPEN", i.getAccessright().getClassid());
});
assertNotNull(p.getBestaccessright());
assertEquals("OPEN", p.getBestaccessright().getClassid());
assertValidId(r1.getSource()); assertValidId(r1.getSource());
assertValidId(r1.getTarget()); assertValidId(r1.getTarget());
assertValidId(r2.getSource()); assertValidId(r2.getSource());
@ -164,6 +175,16 @@ public class MappersTest {
assertTrue(d.getContext().size() > 0); assertTrue(d.getContext().size() > 0);
assertTrue(d.getContext().get(0).getId().length() > 0); assertTrue(d.getContext().get(0).getId().length() > 0);
assertNotNull(d.getInstance());
assertTrue(d.getInstance().size() > 0);
d
.getInstance()
.stream()
.forEach(i -> {
assertNotNull(i.getAccessright());
assertEquals("OPEN", i.getAccessright().getClassid());
});
assertValidId(r1.getSource()); assertValidId(r1.getSource());
assertValidId(r1.getTarget()); assertValidId(r1.getTarget());
assertValidId(r2.getSource()); assertValidId(r2.getSource());