2020-11-18 16:47:31 +01:00
|
|
|
|
|
|
|
package eu.dnetlib.dhp.oa.graph.dump.funderresults;
|
|
|
|
|
|
|
|
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
|
|
|
|
|
|
|
import java.io.Serializable;
|
|
|
|
import java.util.*;
|
|
|
|
|
2020-11-25 13:45:41 +01:00
|
|
|
import eu.dnetlib.dhp.oa.graph.dump.Constants;
|
2020-11-18 16:47:31 +01:00
|
|
|
import org.apache.commons.io.IOUtils;
|
|
|
|
import org.apache.spark.SparkConf;
|
|
|
|
import org.apache.spark.api.java.function.MapFunction;
|
2020-11-24 14:44:53 +01:00
|
|
|
import org.apache.spark.sql.*;
|
2020-11-18 16:47:31 +01:00
|
|
|
import org.slf4j.Logger;
|
|
|
|
import org.slf4j.LoggerFactory;
|
|
|
|
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
2020-11-19 19:15:39 +01:00
|
|
|
import eu.dnetlib.dhp.common.api.zenodo.Community;
|
2020-11-18 16:47:31 +01:00
|
|
|
import eu.dnetlib.dhp.oa.graph.dump.ResultMapper;
|
|
|
|
import eu.dnetlib.dhp.oa.graph.dump.Utils;
|
|
|
|
import eu.dnetlib.dhp.oa.graph.dump.community.CommunityMap;
|
2020-11-19 19:15:39 +01:00
|
|
|
import eu.dnetlib.dhp.schema.dump.oaf.community.CommunityResult;
|
2020-11-24 14:44:53 +01:00
|
|
|
import eu.dnetlib.dhp.schema.dump.oaf.community.Project;
|
2020-11-18 16:47:31 +01:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.Relation;
|
|
|
|
import scala.Tuple2;
|
|
|
|
|
|
|
|
/**
|
2020-11-25 14:08:09 +01:00
|
|
|
* Splits the dumped results by funder and stores them in a folder named as the funder nsp (for all the funders, but the EC
|
|
|
|
* for the EC it specifies also the fundingStream (FP7 or H2020)
|
2020-11-18 16:47:31 +01:00
|
|
|
*/
|
2020-11-18 18:56:48 +01:00
|
|
|
public class SparkDumpFunderResults implements Serializable {
|
|
|
|
private static final Logger log = LoggerFactory.getLogger(SparkDumpFunderResults.class);
|
2020-11-18 16:47:31 +01:00
|
|
|
|
|
|
|
public static void main(String[] args) throws Exception {
|
|
|
|
String jsonConfiguration = IOUtils
|
|
|
|
.toString(
|
2020-11-18 18:56:48 +01:00
|
|
|
SparkDumpFunderResults.class
|
2020-11-18 16:47:31 +01:00
|
|
|
.getResourceAsStream(
|
2020-11-18 16:58:20 +01:00
|
|
|
"/eu/dnetlib/dhp/oa/graph/dump/funder_result_parameters.json"));
|
2020-11-18 16:47:31 +01:00
|
|
|
|
|
|
|
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);
|
|
|
|
|
2020-11-19 19:15:39 +01:00
|
|
|
final String relationPath = parser.get("relationPath");
|
|
|
|
log.info("relationPath: {}", relationPath);
|
2020-11-18 16:47:31 +01:00
|
|
|
|
|
|
|
SparkConf conf = new SparkConf();
|
|
|
|
|
|
|
|
runWithSparkSession(
|
|
|
|
conf,
|
|
|
|
isSparkSessionManaged,
|
|
|
|
spark -> {
|
|
|
|
Utils.removeOutputDir(spark, outputPath);
|
2020-11-20 12:32:23 +01:00
|
|
|
writeResultProjectList(spark, inputPath, outputPath, relationPath);
|
2020-11-18 16:47:31 +01:00
|
|
|
});
|
|
|
|
}
|
|
|
|
|
2020-11-24 14:44:53 +01:00
|
|
|
private static void writeResultProjectList(SparkSession spark, String inputPath, String outputPath,
|
|
|
|
String relationPath) {
|
2020-11-18 16:47:31 +01:00
|
|
|
|
|
|
|
Dataset<Relation> relation = Utils
|
2020-11-20 12:32:23 +01:00
|
|
|
.readPath(spark, relationPath + "/relation", Relation.class)
|
2020-11-25 13:45:41 +01:00
|
|
|
.filter("dataInfo.deletedbyinference = false and lower(relClass) = '" + Constants.RESULT_PROJECT_IS_PRODUCED_BY.toLowerCase()+ "'");
|
2020-11-18 16:47:31 +01:00
|
|
|
|
2020-11-19 19:15:39 +01:00
|
|
|
Dataset<CommunityResult> result = Utils
|
|
|
|
.readPath(spark, inputPath + "/publication", CommunityResult.class)
|
|
|
|
.union(Utils.readPath(spark, inputPath + "/dataset", CommunityResult.class))
|
2020-11-24 16:47:07 +01:00
|
|
|
.union(Utils.readPath(spark, inputPath + "/orp", CommunityResult.class))
|
2020-11-19 19:15:39 +01:00
|
|
|
.union(Utils.readPath(spark, inputPath + "/software", CommunityResult.class));
|
2020-11-18 16:47:31 +01:00
|
|
|
|
2020-11-24 14:44:53 +01:00
|
|
|
List<String> funderList = relation
|
|
|
|
.select("target")
|
|
|
|
.map((MapFunction<Row, String>) value -> value.getString(0).substring(0, 15), Encoders.STRING())
|
|
|
|
.distinct()
|
|
|
|
.collectAsList();
|
|
|
|
|
|
|
|
|
2020-11-25 13:45:41 +01:00
|
|
|
funderList.forEach(funder -> {
|
|
|
|
String fundernsp = funder.substring(3);
|
|
|
|
String funderdump;
|
|
|
|
if (fundernsp.startsWith("corda")){
|
|
|
|
funderdump = "EC_";
|
|
|
|
if(fundernsp.endsWith("h2020")){
|
|
|
|
funderdump += "H2020";
|
|
|
|
}else{
|
|
|
|
funderdump += "FP7";
|
|
|
|
}
|
|
|
|
}else{
|
|
|
|
funderdump = fundernsp.substring(0, fundernsp.indexOf("_")).toUpperCase();
|
|
|
|
}
|
|
|
|
writeFunderResult(funder, result, outputPath + "/" + funderdump);
|
|
|
|
});
|
2020-11-24 14:44:53 +01:00
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
private static void writeFunderResult(String funder, Dataset<CommunityResult> results, String outputPath) {
|
|
|
|
|
|
|
|
results.map((MapFunction<CommunityResult, CommunityResult>) r -> {
|
|
|
|
if (!Optional.ofNullable(r.getProjects()).isPresent()) {
|
|
|
|
return null;
|
|
|
|
}
|
|
|
|
for (Project p : r.getProjects()) {
|
|
|
|
if (p.getId().startsWith(funder)) {
|
|
|
|
return r;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return null;
|
|
|
|
}, Encoders.bean(CommunityResult.class))
|
|
|
|
.filter(Objects::nonNull)
|
2020-11-19 19:15:39 +01:00
|
|
|
.write()
|
|
|
|
.mode(SaveMode.Overwrite)
|
2020-11-24 14:44:53 +01:00
|
|
|
.option("compression", "gzip")
|
2020-11-25 13:45:41 +01:00
|
|
|
.json(outputPath);
|
2020-11-18 16:47:31 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
}
|