Rewrite SparkPropagateRelation exploiting Dataframe API

This commit is contained in:
Giambattista Bloisi 2023-08-28 10:34:54 +02:00
parent 9c8b41475a
commit 0d7b2bf83d
1 changed files with 127 additions and 175 deletions

View File

@ -1,19 +1,6 @@
package eu.dnetlib.dhp.oa.dedup;
import static org.apache.spark.sql.functions.col;
import java.util.Objects;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.function.FilterFunction;
import org.apache.spark.api.java.function.MapFunction;
import org.apache.spark.sql.*;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.common.ModelSupport;
@ -21,16 +8,29 @@ import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.function.FilterFunction;
import org.apache.spark.api.java.function.MapFunction;
import org.apache.spark.api.java.function.ReduceFunction;
import org.apache.spark.sql.*;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import scala.Tuple2;
import scala.Tuple3;
import java.util.Objects;
import static org.apache.spark.sql.functions.col;
public class SparkPropagateRelation extends AbstractSparkAction {
private static final Logger log = LoggerFactory.getLogger(SparkPropagateRelation.class);
enum FieldType {
SOURCE, TARGET
}
private static Encoder<Relation> REL_BEAN_ENC = Encoders.bean(Relation.class);
private static Encoder<Relation> REL_KRYO_ENC = Encoders.kryo(Relation.class);
public SparkPropagateRelation(ArgumentApplicationParser parser, SparkSession spark) {
super(parser, spark);
@ -71,120 +71,42 @@ public class SparkPropagateRelation extends AbstractSparkAction {
Dataset<Relation> mergeRels = spark
.read()
.load(DedupUtility.createMergeRelPath(workingPath, "*", "*"))
.as(Encoders.bean(Relation.class));
.as(REL_BEAN_ENC);
// <mergedObjectID, dedupID>
Dataset<Tuple2<String, String>> mergedIds = mergeRels
Dataset<Row> mergedIds = mergeRels
.where(col("relClass").equalTo(ModelConstants.MERGES))
.select(col("source"), col("target"))
.select(col("source").as("dedupID"), col("target").as("mergedObjectID"))
.distinct()
.map(
(MapFunction<Row, Tuple2<String, String>>) r -> new Tuple2<>(r.getString(1), r.getString(0)),
Encoders.tuple(Encoders.STRING(), Encoders.STRING()))
.cache();
final String relationPath = DedupUtility.createEntityPath(graphBasePath, "relation");
final String inputRelationPath = DedupUtility.createEntityPath(graphBasePath, "relation");
Dataset<Relation> rels = spark.read().textFile(relationPath).map(patchRelFn(), Encoders.bean(Relation.class));
Dataset<Relation> rels = spark.read().schema(REL_BEAN_ENC.schema()).json(inputRelationPath)
.as(REL_BEAN_ENC)
// .map((MapFunction<Relation, Relation>) rel -> {
// if (rel.getDataInfo() == null) {
// rel.setDataInfo(new DataInfo());
// }
// return rel;
// }, REL_BEAN_ENC)
;
Dataset<Relation> newRels = createNewRels(rels, mergedIds, getFixRelFn());
Dataset<Tuple3<Relation, String, String>> dedupedRels = rels
.joinWith(mergedIds, rels.col("source").equalTo(mergedIds.col("mergedObjectID")), "left_outer")
.joinWith(mergedIds, col("_1.target").equalTo(mergedIds.col("mergedObjectID")), "left_outer")
.filter("_1._2 IS NOT NULL OR _2 IS NOT NULL")
.select("_1._1", "_1._2.dedupID", "_2.dedupID")
.as(Encoders.tuple(REL_BEAN_ENC, Encoders.STRING(), Encoders.STRING()))
.cache();
Dataset<Relation> updated = processDataset(
processDataset(rels, mergedIds, FieldType.SOURCE, getDeletedFn()),
mergedIds,
FieldType.TARGET,
getDeletedFn());
mergedIds.unpersist();
save(
distinctRelations(
newRels
.union(updated)
.union(mergeRels)
.map((MapFunction<Relation, Relation>) r -> r, Encoders.kryo(Relation.class)))
.filter((FilterFunction<Relation>) r -> !Objects.equals(r.getSource(), r.getTarget())),
outputRelationPath, SaveMode.Overwrite);
}
private Dataset<Relation> distinctRelations(Dataset<Relation> rels) {
return rels
.filter(getRelationFilterFunction())
.groupByKey(
(MapFunction<Relation, String>) r -> String
.join(r.getSource(), r.getTarget(), r.getRelType(), r.getSubRelType(), r.getRelClass()),
Encoders.STRING())
.agg(new RelationAggregator().toColumn())
.map((MapFunction<Tuple2<String, Relation>, Relation>) Tuple2::_2, Encoders.bean(Relation.class));
}
// redirect the relations to the dedupID
private static Dataset<Relation> createNewRels(
Dataset<Relation> rels, // all the relations to be redirected
Dataset<Tuple2<String, String>> mergedIds, // merge rels: <mergedObjectID, dedupID>
MapFunction<Tuple2<Tuple2<Tuple3<String, Relation, String>, Tuple2<String, String>>, Tuple2<String, String>>, Relation> mapRel) {
// <sourceID, relation, targetID>
Dataset<Tuple3<String, Relation, String>> mapped = rels
.map(
(MapFunction<Relation, Tuple3<String, Relation, String>>) r -> new Tuple3<>(getId(r, FieldType.SOURCE),
r, getId(r, FieldType.TARGET)),
Encoders.tuple(Encoders.STRING(), Encoders.kryo(Relation.class), Encoders.STRING()));
// < <sourceID, relation, target>, <sourceID, dedupID> >
Dataset<Tuple2<Tuple3<String, Relation, String>, Tuple2<String, String>>> relSource = mapped
.joinWith(mergedIds, mapped.col("_1").equalTo(mergedIds.col("_1")), "left_outer");
// < <<sourceID, relation, targetID>, <sourceID, dedupID>>, <targetID, dedupID> >
Dataset<Tuple2<Tuple2<Tuple3<String, Relation, String>, Tuple2<String, String>>, Tuple2<String, String>>> relSourceTarget = relSource
.joinWith(mergedIds, relSource.col("_1._3").equalTo(mergedIds.col("_1")), "left_outer");
return relSourceTarget
.filter(
(FilterFunction<Tuple2<Tuple2<Tuple3<String, Relation, String>, Tuple2<String, String>>, Tuple2<String, String>>>) r -> r
._1()
._1() != null || r._2() != null)
.map(mapRel, Encoders.bean(Relation.class))
.distinct();
}
private static Dataset<Relation> processDataset(
Dataset<Relation> rels,
Dataset<Tuple2<String, String>> mergedIds,
FieldType type,
MapFunction<Tuple2<Tuple2<String, Relation>, Tuple2<String, String>>, Relation> mapFn) {
final Dataset<Tuple2<String, Relation>> mapped = rels
.map(
(MapFunction<Relation, Tuple2<String, Relation>>) r -> new Tuple2<>(getId(r, type), r),
Encoders.tuple(Encoders.STRING(), Encoders.kryo(Relation.class)));
return mapped
.joinWith(mergedIds, mapped.col("_1").equalTo(mergedIds.col("_1")), "left_outer")
.map(mapFn, Encoders.bean(Relation.class));
}
private FilterFunction<Relation> getRelationFilterFunction() {
return r -> StringUtils.isNotBlank(r.getSource()) ||
StringUtils.isNotBlank(r.getTarget()) ||
StringUtils.isNotBlank(r.getRelType()) ||
StringUtils.isNotBlank(r.getSubRelType()) ||
StringUtils.isNotBlank(r.getRelClass());
}
private static String getId(Relation r, FieldType type) {
switch (type) {
case SOURCE:
return r.getSource();
case TARGET:
return r.getTarget();
default:
throw new IllegalArgumentException("");
}
}
private static MapFunction<Tuple2<Tuple2<Tuple3<String, Relation, String>, Tuple2<String, String>>, Tuple2<String, String>>, Relation> getFixRelFn() {
return value -> {
Relation r = value._1()._1()._2();
String newSource = value._1()._2() != null ? value._1()._2()._2() : null;
String newTarget = value._2() != null ? value._2()._2() : null;
Dataset<Relation> newRels = dedupedRels
.map((MapFunction<Tuple3<Relation, String, String>, Relation>) t -> {
Relation r = t._1();
String newSource = t._2();
String newTarget = t._3();
if (r.getDataInfo() == null) {
r.setDataInfo(new DataInfo());
@ -198,21 +120,51 @@ public class SparkPropagateRelation extends AbstractSparkAction {
r.setTarget(newTarget);
return r;
};
}
}, REL_BEAN_ENC)
.distinct();
private static MapFunction<Tuple2<Tuple2<String, Relation>, Tuple2<String, String>>, Relation> getDeletedFn() {
return value -> {
if (value._2() != null) {
Relation r = value._1()._2();
Dataset<Relation> updated = dedupedRels
.map((MapFunction<Tuple3<Relation, String, String>, Relation>) t -> {
Relation r = t._1();
if (r.getDataInfo() == null) {
r.setDataInfo(new DataInfo());
}
r.getDataInfo().setDeletedbyinference(true);
return r;
}
return value._1()._2();
};
}, REL_BEAN_ENC);
save(
distinctRelations(
newRels
.union(updated)
.union(mergeRels)
.map((MapFunction<Relation, Relation>) r -> r, REL_KRYO_ENC)
)
.filter((FilterFunction<Relation>) r -> !Objects.equals(r.getSource(), r.getTarget())),
outputRelationPath, SaveMode.Overwrite);
}
private Dataset<Relation> distinctRelations(Dataset<Relation> rels) {
return rels
.filter(getRelationFilterFunction())
.groupByKey(
(MapFunction<Relation, String>) r -> String
.join(" ", r.getSource(), r.getTarget(), r.getRelType(), r.getSubRelType(), r.getRelClass()),
Encoders.STRING())
.reduceGroups((ReduceFunction<Relation>) (b, a) -> {
b.mergeFrom(a);
return b;
}
)
.map((MapFunction<Tuple2<String, Relation>, Relation>) Tuple2::_2, REL_BEAN_ENC);
}
private FilterFunction<Relation> getRelationFilterFunction() {
return r -> StringUtils.isNotBlank(r.getSource()) ||
StringUtils.isNotBlank(r.getTarget()) ||
StringUtils.isNotBlank(r.getRelType()) ||
StringUtils.isNotBlank(r.getSubRelType()) ||
StringUtils.isNotBlank(r.getRelClass());
}
}