2020-08-13 12:12:40 +02:00
|
|
|
|
2020-11-05 17:38:18 +01:00
|
|
|
package eu.dnetlib.dhp.oa.graph.dump.complete;
|
2020-07-24 17:21:48 +02:00
|
|
|
|
2020-07-29 17:03:19 +02:00
|
|
|
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
|
|
|
|
|
|
|
import java.io.Serializable;
|
|
|
|
import java.util.Optional;
|
|
|
|
|
|
|
|
import org.apache.commons.io.IOUtils;
|
|
|
|
import org.apache.spark.SparkConf;
|
2020-08-24 17:07:34 +02:00
|
|
|
import org.apache.spark.api.java.function.MapFunction;
|
2020-10-30 14:12:28 +01:00
|
|
|
import org.apache.spark.sql.Dataset;
|
2020-07-29 17:03:19 +02:00
|
|
|
import org.apache.spark.sql.Encoders;
|
|
|
|
import org.apache.spark.sql.SaveMode;
|
|
|
|
import org.apache.spark.sql.SparkSession;
|
|
|
|
import org.slf4j.Logger;
|
|
|
|
import org.slf4j.LoggerFactory;
|
|
|
|
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
|
|
|
import eu.dnetlib.dhp.oa.graph.dump.Utils;
|
|
|
|
import eu.dnetlib.dhp.schema.common.ModelSupport;
|
|
|
|
import eu.dnetlib.dhp.schema.dump.oaf.Provenance;
|
|
|
|
import eu.dnetlib.dhp.schema.dump.oaf.graph.Node;
|
|
|
|
import eu.dnetlib.dhp.schema.dump.oaf.graph.RelType;
|
2020-10-30 14:12:28 +01:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.DataInfo;
|
2020-07-29 17:03:19 +02:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.Relation;
|
|
|
|
|
2020-08-13 12:12:40 +02:00
|
|
|
/**
|
|
|
|
* Dumps eu.dnetlib.dhp.schema.oaf.Relation in eu.dnetlib.dhp.schema.dump.oaf.graph.Relation
|
|
|
|
*/
|
2020-07-29 17:03:19 +02:00
|
|
|
public class SparkDumpRelationJob implements Serializable {
|
|
|
|
|
|
|
|
private static final Logger log = LoggerFactory.getLogger(SparkDumpRelationJob.class);
|
|
|
|
|
|
|
|
public static void main(String[] args) throws Exception {
|
|
|
|
String jsonConfiguration = IOUtils
|
|
|
|
.toString(
|
2020-07-29 17:42:50 +02:00
|
|
|
SparkDumpRelationJob.class
|
2020-07-29 17:03:19 +02:00
|
|
|
.getResourceAsStream(
|
2020-11-05 17:38:18 +01:00
|
|
|
"/eu/dnetlib/dhp/oa/graph/dump/complete/input_relationdump_parameters.json"));
|
2020-07-29 17:03:19 +02:00
|
|
|
|
|
|
|
final ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
|
|
|
|
parser.parseArgument(args);
|
|
|
|
|
|
|
|
Boolean isSparkSessionManaged = Optional
|
|
|
|
.ofNullable(parser.get("isSparkSessionManaged"))
|
|
|
|
.map(Boolean::valueOf)
|
|
|
|
.orElse(Boolean.TRUE);
|
|
|
|
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
|
|
|
|
|
|
|
|
final String inputPath = parser.get("sourcePath");
|
|
|
|
log.info("inputPath: {}", inputPath);
|
|
|
|
|
|
|
|
final String outputPath = parser.get("outputPath");
|
|
|
|
log.info("outputPath: {}", outputPath);
|
|
|
|
|
|
|
|
SparkConf conf = new SparkConf();
|
|
|
|
|
|
|
|
runWithSparkSession(
|
|
|
|
conf,
|
|
|
|
isSparkSessionManaged,
|
|
|
|
spark -> {
|
|
|
|
Utils.removeOutputDir(spark, outputPath);
|
|
|
|
dumpRelation(spark, inputPath, outputPath);
|
|
|
|
|
|
|
|
});
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
private static void dumpRelation(SparkSession spark, String inputPath, String outputPath) {
|
2020-10-30 14:12:28 +01:00
|
|
|
Dataset<Relation> relations = Utils.readPath(spark, inputPath, Relation.class);
|
|
|
|
relations
|
2020-08-24 17:07:34 +02:00
|
|
|
.map((MapFunction<Relation, eu.dnetlib.dhp.schema.dump.oaf.graph.Relation>) relation -> {
|
2021-10-04 18:40:07 +02:00
|
|
|
eu.dnetlib.dhp.schema.dump.oaf.graph.Relation relNew = new eu.dnetlib.dhp.schema.dump.oaf.graph.Relation();
|
|
|
|
relNew
|
2020-07-29 17:03:19 +02:00
|
|
|
.setSource(
|
|
|
|
Node
|
|
|
|
.newInstance(
|
|
|
|
relation.getSource(),
|
|
|
|
ModelSupport.idPrefixEntity.get(relation.getSource().substring(0, 2))));
|
|
|
|
|
2021-10-04 18:40:07 +02:00
|
|
|
relNew
|
2020-07-29 17:03:19 +02:00
|
|
|
.setTarget(
|
|
|
|
Node
|
|
|
|
.newInstance(
|
|
|
|
relation.getTarget(),
|
|
|
|
ModelSupport.idPrefixEntity.get(relation.getTarget().substring(0, 2))));
|
|
|
|
|
2021-10-04 18:40:07 +02:00
|
|
|
relNew
|
2020-07-29 17:03:19 +02:00
|
|
|
.setReltype(
|
|
|
|
RelType
|
|
|
|
.newInstance(
|
|
|
|
relation.getRelClass(),
|
|
|
|
relation.getSubRelType()));
|
|
|
|
|
2020-10-30 14:12:28 +01:00
|
|
|
Optional<DataInfo> odInfo = Optional.ofNullable(relation.getDataInfo());
|
|
|
|
if (odInfo.isPresent()) {
|
|
|
|
DataInfo dInfo = odInfo.get();
|
2021-10-04 18:40:07 +02:00
|
|
|
if (Optional.ofNullable(dInfo.getProvenanceaction()).isPresent() &&
|
|
|
|
Optional.ofNullable(dInfo.getProvenanceaction().getClassname()).isPresent()) {
|
|
|
|
relNew
|
|
|
|
.setProvenance(
|
|
|
|
Provenance
|
|
|
|
.newInstance(
|
|
|
|
dInfo.getProvenanceaction().getClassname(),
|
|
|
|
dInfo.getTrust()));
|
2020-10-30 14:12:28 +01:00
|
|
|
}
|
|
|
|
}
|
2021-10-04 18:40:07 +02:00
|
|
|
if (Boolean.TRUE.equals(relation.getValidated())) {
|
|
|
|
relNew.setValidated(relation.getValidated());
|
|
|
|
relNew.setValidationDate(relation.getValidationDate());
|
2021-08-06 18:56:18 +02:00
|
|
|
}
|
2020-10-30 14:12:28 +01:00
|
|
|
|
2021-10-04 18:40:07 +02:00
|
|
|
return relNew;
|
2020-07-29 17:03:19 +02:00
|
|
|
|
|
|
|
}, Encoders.bean(eu.dnetlib.dhp.schema.dump.oaf.graph.Relation.class))
|
|
|
|
.write()
|
|
|
|
.option("compression", "gzip")
|
|
|
|
.mode(SaveMode.Overwrite)
|
|
|
|
.json(outputPath);
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2020-07-24 17:21:48 +02:00
|
|
|
}
|