D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki Activity

orcid events

This commit is contained in:
Michele Artini 2020-06-11 14:24:02 +02:00
parent a41e0cb648
commit 48959e9a17
1 changed files with 38 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

49
dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/matchers/simple/EnrichMissingAuthorOrcid.java
View File

@ -1,41 +1,68 @@
package eu.dnetlib.dhp.broker.oa.matchers.simple;
import java.util.Arrays;
import java.util.ArrayList;
import java.util.List;
import org.apache.commons.lang3.tuple.Pair;
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.UpdateInfo;
import eu.dnetlib.dhp.broker.oa.util.aggregators.withRels.ResultWithRelations;
import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
import eu.dnetlib.pace.config.DedupConfig;
public class EnrichMissingAuthorOrcid extends UpdateMatcher<Pair<String, String>> {
public class EnrichMissingAuthorOrcid extends UpdateMatcher<String> {
public EnrichMissingAuthorOrcid() {
super(true);
}
@Override
protected List<UpdateInfo<Pair<String, String>>> findUpdates(final ResultWithRelations source,
protected List<UpdateInfo<String>> findUpdates(final ResultWithRelations source,
final ResultWithRelations target,
final DedupConfig dedupConfig) {
// TODO
// return Arrays.asList(new EnrichMissingAbstract("xxxxxxx", 0.9f));
return Arrays.asList();
final Set<String> existingOrcids = target
.getResult()
.getAuthor()
.stream()
.map(Author::getPid)
.flatMap(List::stream)
.filter(pid -> pid.getQualifier().getClassid().equalsIgnoreCase("orcid"))
.map(pid -> pid.getValue())
.collect(Collectors.toSet());
final List<UpdateInfo<String>> list = new ArrayList<>();
for (final Author author : source.getResult().getAuthor()) {
final String name = author.getFullname();
for (final StructuredProperty pid : author.getPid()) {
if (pid.getQualifier().getClassid().equalsIgnoreCase("orcid")
&& !existingOrcids.contains(pid.getValue())) {
list
.add(
generateUpdateInfo(name + " [ORCID: " + pid.getValue() + "]", source, target, dedupConfig));
;
}
}
}
public UpdateInfo<Pair<String, String>> generateUpdateInfo(final Pair<String, String> highlightValue,
return list;
}
public UpdateInfo<String> generateUpdateInfo(final String highlightValue,
final ResultWithRelations source,
final ResultWithRelations target,
final DedupConfig dedupConfig) {
return new UpdateInfo<>(
Topic.ENRICH_MISSING_AUTHOR_ORCID,
highlightValue, source, target,
(p, pair) -> p.getCreators().add(pair.getLeft() + " - ORCID: " + pair.getRight()),
pair -> pair.getLeft() + "::" + pair.getRight(),
(p, aut) -> p.getCreators().add(aut),
aut -> aut,
dedupConfig);
}
}