102 lines
3.3 KiB
Java
102 lines
3.3 KiB
Java
|
|
package eu.dnetlib.dhp.oa.graph.dump.complete;
|
|
|
|
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
|
|
|
import java.io.Serializable;
|
|
import java.util.ArrayList;
|
|
import java.util.Arrays;
|
|
import java.util.List;
|
|
import java.util.Optional;
|
|
|
|
import org.apache.commons.io.IOUtils;
|
|
import org.apache.spark.SparkConf;
|
|
import org.apache.spark.sql.*;
|
|
import org.apache.spark.sql.types.*;
|
|
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.oaf.*;
|
|
|
|
/**
|
|
* It selects the valid relations among those present in the graph. One relation is valid if it is not deletedbyinference
|
|
* and if both the source and the target node are present in the graph and are not deleted by inference nor invisible.
|
|
* To check this I made a view of the ids of all the entities in the graph, and select the relations for which a join exists
|
|
* with this view for both the source and the target
|
|
*/
|
|
|
|
public class SparkSelectValidRelationsJob implements Serializable {
|
|
|
|
private static final Logger log = LoggerFactory.getLogger(SparkSelectValidRelationsJob.class);
|
|
|
|
public static void main(String[] args) throws Exception {
|
|
String jsonConfiguration = IOUtils
|
|
.toString(
|
|
SparkSelectValidRelationsJob.class
|
|
.getResourceAsStream(
|
|
"/eu/dnetlib/dhp/oa/graph/dump/input_relationdump_parameters.json"));
|
|
|
|
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);
|
|
selectValidRelation2(spark, inputPath, outputPath);
|
|
|
|
});
|
|
|
|
}
|
|
|
|
private static void selectValidRelation2(SparkSession spark, String inputPath, String outputPath) {
|
|
final StructType structureSchema = new StructType()
|
|
.fromDDL("`id` STRING, `dataInfo` STRUCT<`deletedbyinference`:BOOLEAN,`invisible`:BOOLEAN>");
|
|
|
|
org.apache.spark.sql.Dataset<Row> df = spark.createDataFrame(new ArrayList<Row>(), structureSchema);
|
|
List<String> entities = Arrays
|
|
.asList(
|
|
"publication", "dataset", "otherresearchproduct", "software", "organization", "project", "datasource");
|
|
for (String e : entities)
|
|
df = df
|
|
.union(
|
|
spark
|
|
.read()
|
|
.schema(structureSchema)
|
|
.json(inputPath + "/" + e)
|
|
.filter("dataInfo.deletedbyinference != true and dataInfo.invisible != true"));
|
|
|
|
org.apache.spark.sql.Dataset<Row> relations = spark
|
|
.read()
|
|
.schema(Encoders.bean(Relation.class).schema())
|
|
.json(inputPath + "/relation")
|
|
.filter("dataInfo.deletedbyinference == false");
|
|
|
|
relations
|
|
.join(df, relations.col("source").equalTo(df.col("id")), "leftsemi")
|
|
.join(df, relations.col("target").equalTo(df.col("id")), "leftsemi")
|
|
.write()
|
|
.mode(SaveMode.Overwrite)
|
|
.option("compression", "gzip")
|
|
.json(outputPath);
|
|
}
|
|
|
|
}
|