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

dataset based provision WIP, fixed spark2EventLogDir

This commit is contained in:
Claudio Atzori 2020-04-02 16:15:50 +02:00
parent 9c7092416a
commit daa26acc9d
2 changed files with 30 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

48
dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/oa/provision/GraphJoiner_v2.java
View File

@ -86,24 +86,36 @@ public class GraphJoiner_v2 implements Serializable {
Dataset<TypedRow> publication = readPathEntity(jsc, getInputPath(), "publication");
// create the union between all the entities
Dataset<Tuple2<String, TypedRow>> entities =
datasource
.union(organization)
.union(project)
.union(dataset)
.union(otherresearchproduct)
.union(software)
.union(publication)
.map((MapFunction<TypedRow, Tuple2<String, TypedRow>>) value -> new Tuple2<>(
value.getId(),
value),
Encoders.tuple(Encoders.STRING(), Encoders.kryo(TypedRow.class)))
.cache();
datasource
.union(organization)
.union(project)
.union(dataset)
.union(otherresearchproduct)
.union(software)
.union(publication)
.repartition(20000)
.write()
.parquet(getOutPath() + "/entities");
Dataset<Tuple2<String, TypedRow>> entities = getSpark()
.read()
.load(getOutPath() + "/entities")
.map((MapFunction<Row, Tuple2<String, TypedRow>>) r -> {
TypedRow t = new TypedRow();
t.setId(r.getAs("id"));
t.setDeleted(r.getAs("deleted"));
t.setType(r.getAs("type"));
t.setOaf(r.getAs("oaf"));
return new Tuple2<>(t.getId(), t);
}, Encoders.tuple(Encoders.STRING(), Encoders.kryo(TypedRow.class)));
System.out.println("Entities, number of partitions: " + entities.rdd().getNumPartitions());
System.out.println("Entities schema:");
entities.printSchema();
// reads the relationships
/*
// reads the relationships
Dataset<Relation> rels = readPathRelation(jsc, getInputPath())
.groupByKey((MapFunction<Relation, SortableRelationKey>) t -> SortableRelationKey.from(t), Encoders.kryo(SortableRelationKey.class))
.flatMapGroups((FlatMapGroupsFunction<SortableRelationKey, Relation, Relation>) (key, values) -> Iterators.limit(values, MAX_RELS), Encoders.bean(Relation.class))
@ -126,7 +138,7 @@ public class GraphJoiner_v2 implements Serializable {
e.setRelation(t._1()._2());
e.setTarget(asRelatedEntity(t._2()._2()));
return e;
}, Encoders.bean(EntityRelEntity.class))
}, Encoders.kryo(EntityRelEntity.class))
.map((MapFunction<EntityRelEntity, Tuple2<String, EntityRelEntity>>) e -> new Tuple2<>(e.getRelation().getSource(), e),
Encoders.tuple(Encoders.STRING(), Encoders.kryo(EntityRelEntity.class)));
@ -160,14 +172,12 @@ public class GraphJoiner_v2 implements Serializable {
final XmlRecordFactory recordFactory = new XmlRecordFactory(accumulators, contextMapper, false, schemaLocation, otherDsTypeId);
grouped
.map((MapFunction<JoinedEntity, String>) value -> recordFactory.build(value), Encoders.STRING())
.write()
.text(getOutPath() + "/xml");
/*
.javaRDD()
.mapToPair((PairFunction<Tuple2<String, String>, String, String>) t -> new Tuple2<>(t._1(), t._2()))
.saveAsHadoopFile(getOutPath() + "/xml", Text.class, Text.class, SequenceFileOutputFormat.class, GzipCodec.class);
*/
*/
return this;
}

2
dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/config-default.xml
View File

@ -25,6 +25,6 @@
</property>
<property>
<name>spark2EventLogDir</name>
<value>/user/spark/applicationHistory</value>
<value>/user/spark/spark2ApplicationHistory</value>
</property>
</configuration>