alessia.bardi
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

partial refactoring

This commit is contained in:
Michele Artini 2020-06-16 15:55:42 +02:00
parent 113c9b1de0
commit 9e2c23e391
3 changed files with 83 additions and 54 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

37
dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/GenerateEventsApplication.java
View File

@ -51,7 +51,8 @@ public class GenerateEventsApplication {
public static void main(final String[] args) throws Exception {
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
IOUtils
.toString(GenerateEventsApplication.class
.toString(
GenerateEventsApplication.class
.getResourceAsStream("/eu/dnetlib/dhp/broker/oa/generate_broker_events.json")));
parser.parseArgument(args);
@ -117,11 +118,14 @@ public class GenerateEventsApplication {
return results
.joinWith(mergedRels, results.col("result.id").equalTo(mergedRels.col("source")), "inner")
.groupByKey((MapFunction<Tuple2<OpenaireBrokerResult, Relation>, String>) t -> t._2.getTarget(), Encoders.STRING())
.groupByKey(
(MapFunction<Tuple2<OpenaireBrokerResult, Relation>, String>) t -> t._2.getTarget(), Encoders.STRING())
.agg(aggr)
.map((MapFunction<Tuple2<String, ResultGroup>, ResultGroup>) t -> t._2, Encoders.kryo(ResultGroup.class))
.filter(ResultGroup::isValid)
.map((MapFunction<ResultGroup, EventGroup>) g -> EventFinder.generateEvents(g, dedupConfig), Encoders.kryo(EventGroup.class))
.map(
(MapFunction<ResultGroup, EventGroup>) g -> EventFinder.generateEvents(g, dedupConfig),
Encoders.kryo(EventGroup.class))
.flatMap(group -> group.getData().iterator(), Encoders.kryo(Event.class));
}
@ -130,7 +134,8 @@ public class GenerateEventsApplication {
final String graphPath,
final Class<SRC> sourceClass) {
final Dataset<Project> projects = readPath(spark, graphPath + "/project", Project.class);
final Dataset<eu.dnetlib.dhp.schema.oaf.Dataset> datasets = readPath(spark, graphPath + "/dataset", eu.dnetlib.dhp.schema.oaf.Dataset.class);
final Dataset<eu.dnetlib.dhp.schema.oaf.Dataset> datasets = readPath(
spark, graphPath + "/dataset", eu.dnetlib.dhp.schema.oaf.Dataset.class);
final Dataset<Software> softwares = readPath(spark, graphPath + "/software", Software.class);
final Dataset<Publication> publications = readPath(spark, graphPath + "/publication", Publication.class);
@ -138,14 +143,17 @@ public class GenerateEventsApplication {
.filter(r -> !r.getRelClass().equals(BrokerConstants.IS_MERGED_IN_CLASS))
.cache();
final Dataset<OpenaireBrokerResult> r0 = readPath(spark, graphPath + "/" + sourceClass.getSimpleName().toLowerCase(), Result.class)
final Dataset<OpenaireBrokerResult> r0 = readPath(
spark, graphPath + "/" + sourceClass.getSimpleName().toLowerCase(), Result.class)
.filter(r -> r.getDataInfo().getDeletedbyinference())
.map(ConversionUtils::oafResultToBrokerResult, Encoders.kryo(OpenaireBrokerResult.class));
final Dataset<OpenaireBrokerResult> r1 = join(r0, rels, relatedEntities(projects, rels, RelatedProject.class));
final Dataset<OpenaireBrokerResult> r2 = join(r1, rels, relatedEntities(softwares, rels, RelatedProject.class));
final Dataset<OpenaireBrokerResult> r3 = join(r2, rels, relatedEntities(datasets, rels, RelatedProject.class));
final Dataset<OpenaireBrokerResult> r4 = join(r3, rels, relatedEntities(publications, rels, RelatedProject.class));;
final Dataset<OpenaireBrokerResult> r4 = join(
r3, rels, relatedEntities(publications, rels, RelatedProject.class));
;
return r4;
}
@ -155,7 +163,9 @@ public class GenerateEventsApplication {
final Class<RT> clazz) {
return rels
.joinWith(targets, targets.col("id").equalTo(rels.col("target")), "inner")
.map(t -> RelatedEntityFactory.newRelatedEntity(t._1.getSource(), t._1.getRelType(), t._2, clazz), Encoders.kryo(clazz));
.map(
t -> RelatedEntityFactory.newRelatedEntity(t._1.getSource(), t._1.getRelType(), t._2, clazz),
Encoders.kryo(clazz));
}
private static <T> Dataset<OpenaireBrokerResult> join(final Dataset<OpenaireBrokerResult> sources,
@ -163,11 +173,13 @@ public class GenerateEventsApplication {
final Dataset<T> typedRels) {
final TypedColumn<Tuple2<OpenaireBrokerResult, T>, OpenaireBrokerResult> aggr = new OpenaireBrokerResultAggregator<T>()
.toColumn();;
.toColumn();
;
return sources
.joinWith(typedRels, sources.col("result.id").equalTo(rels.col("source")), "left_outer")
.groupByKey((MapFunction<Tuple2<OpenaireBrokerResult, T>, String>) t -> t._1.getOpenaireId(), Encoders.STRING())
.groupByKey(
(MapFunction<Tuple2<OpenaireBrokerResult, T>, String>) t -> t._1.getOpenaireId(), Encoders.STRING())
.agg(aggr)
.map(t -> t._2, Encoders.kryo(OpenaireBrokerResult.class));
}
@ -186,8 +198,11 @@ public class GenerateEventsApplication {
final ISLookUpService isLookUpService = ISLookupClientFactory.getLookUpService(isLookupUrl);
final String conf = isLookUpService
.getResourceProfileByQuery(String
.format("for $x in /RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value = '%s'] return $x//DEDUPLICATION/text()", profId));
.getResourceProfileByQuery(
String
.format(
"for $x in /RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value = '%s'] return $x//DEDUPLICATION/text()",
profId));
final DedupConfig dedupConfig = new ObjectMapper().readValue(conf, DedupConfig.class);
dedupConfig.getPace().initModel();

28
dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ConversionUtils.java
View File

@ -57,7 +57,8 @@ public class ConversionUtils {
.setOriginalId(d.getOriginalId().get(0))
.setTitle(structPropValue(d.getTitle()))
.setPids(d.getPid().stream().map(ConversionUtils::oafPidToBrokerPid).collect(Collectors.toList()))
.setInstances(d
.setInstances(
d
.getInstance()
.stream()
.map(ConversionUtils::oafInstanceToBrokerInstances)
@ -72,7 +73,8 @@ public class ConversionUtils {
.setOriginalId(p.getOriginalId().get(0))
.setTitle(structPropValue(p.getTitle()))
.setPids(p.getPid().stream().map(ConversionUtils::oafPidToBrokerPid).collect(Collectors.toList()))
.setInstances(p
.setInstances(
p
.getInstance()
.stream()
.map(ConversionUtils::oafInstanceToBrokerInstances)
@ -92,22 +94,26 @@ public class ConversionUtils {
.setAbstracts(fieldList(result.getDescription()))
.setLanguage(result.getLanguage().getClassid())
.setSubjects(structPropTypedList(result.getSubject()))
.setCreators(result.getAuthor().stream().map(ConversionUtils::oafAuthorToBrokerAuthor).collect(Collectors.toList()))
.setCreators(
result.getAuthor().stream().map(ConversionUtils::oafAuthorToBrokerAuthor).collect(Collectors.toList()))
.setPublicationdate(result.getDateofacceptance().getValue())
.setPublisher(fieldValue(result.getPublisher()))
.setEmbargoenddate(fieldValue(result.getEmbargoenddate()))
.setContributor(fieldList(result.getContributor()))
.setJournal(result instanceof Publication ? oafJournalToBrokerJournal(((Publication) result).getJournal()) : null)
.setJournal(
result instanceof Publication ? oafJournalToBrokerJournal(((Publication) result).getJournal()) : null)
.setCollectedFromId(result.getCollectedfrom().stream().map(KeyValue::getKey).findFirst().orElse(null))
.setCollectedFromName(result.getCollectedfrom().stream().map(KeyValue::getValue).findFirst().orElse(null))
.setPids(result.getPid().stream().map(ConversionUtils::oafPidToBrokerPid).collect(Collectors.toList()))
.setInstances(result
.setInstances(
result
.getInstance()
.stream()
.map(ConversionUtils::oafInstanceToBrokerInstances)
.flatMap(List::stream)
.collect(Collectors.toList()))
.setExternalReferences(result
.setExternalReferences(
result
.getExternalReference()
.stream()
.map(ConversionUtils::oafExtRefToBrokerExtRef)
@ -118,7 +124,8 @@ public class ConversionUtils {
private static List<TypedValue> structPropTypedList(final List<StructuredProperty> list) {
return list
.stream()
.map(p -> new TypedValue()
.map(
p -> new TypedValue()
.setValue(p.getValue())
.setType(p.getQualifier().getClassid()))
.collect(Collectors.toList());
@ -127,7 +134,8 @@ public class ConversionUtils {
private static eu.dnetlib.broker.objects.Author oafAuthorToBrokerAuthor(final Author author) {
return author != null ? new eu.dnetlib.broker.objects.Author()
.setFullname(author.getFullname())
.setOrcid(author
.setOrcid(
author
.getPid()
.stream()
.filter(pid -> pid.getQualifier().getClassid().equalsIgnoreCase("orcid"))
@ -155,7 +163,9 @@ public class ConversionUtils {
}
public static final eu.dnetlib.broker.objects.Project oafProjectToBrokerProject(final Project p) {
if (p == null) { return null; }
if (p == null) {
return null;
}
final eu.dnetlib.broker.objects.Project res = new eu.dnetlib.broker.objects.Project()
.setTitle(fieldValue(p.getTitle()))

12
dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/aggregators/withRels/RelatedEntityFactory.java
View File

@ -16,13 +16,17 @@ public class RelatedEntityFactory {
final Class<RT> clazz) {
if (clazz == RelatedProject.class) {
return (RT) new RelatedProject(sourceId, relType, ConversionUtils.oafProjectToBrokerProject((Project) target));
return (RT) new RelatedProject(sourceId, relType,
ConversionUtils.oafProjectToBrokerProject((Project) target));
} else if (clazz == RelatedSoftware.class) {
return (RT) new RelatedSoftware(sourceId, relType, ConversionUtils.oafSoftwareToBrokerSoftware((Software) target));
return (RT) new RelatedSoftware(sourceId, relType,
ConversionUtils.oafSoftwareToBrokerSoftware((Software) target));
} else if (clazz == RelatedDataset.class) {
return (RT) new RelatedDataset(sourceId, relType, ConversionUtils.oafDatasetToBrokerDataset((Dataset) target));
return (RT) new RelatedDataset(sourceId, relType,
ConversionUtils.oafDatasetToBrokerDataset((Dataset) target));
} else if (clazz == RelatedPublication.class) {
return (RT) new RelatedPublication(sourceId, relType, ConversionUtils.oafPublicationToBrokerPublication((Publication) target));
return (RT) new RelatedPublication(sourceId, relType,
ConversionUtils.oafPublicationToBrokerPublication((Publication) target));
} else {
return null;
}