2021-11-10 17:00:37 +01:00
|
|
|
|
2021-11-11 16:11:25 +01:00
|
|
|
package eu.dnetlib.dhp.actionmanager.createunresolvedentities;
|
2021-11-08 16:44:19 +01:00
|
|
|
|
2021-12-22 16:44:44 +01:00
|
|
|
import static eu.dnetlib.dhp.actionmanager.Constants.*;
|
2021-11-10 17:00:37 +01:00
|
|
|
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
|
|
|
|
|
|
|
import java.io.Serializable;
|
2021-11-11 16:11:25 +01:00
|
|
|
import java.util.*;
|
|
|
|
import java.util.stream.Collectors;
|
2021-11-10 17:00:37 +01:00
|
|
|
|
2021-11-08 16:44:19 +01:00
|
|
|
import org.apache.commons.io.IOUtils;
|
|
|
|
import org.apache.spark.SparkConf;
|
2021-11-11 16:11:25 +01:00
|
|
|
import org.apache.spark.api.java.function.MapFunction;
|
2021-12-22 21:40:35 +01:00
|
|
|
import org.apache.spark.api.java.function.MapGroupsFunction;
|
2021-11-08 16:44:19 +01:00
|
|
|
import org.apache.spark.sql.Dataset;
|
|
|
|
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;
|
|
|
|
|
2021-11-11 16:11:25 +01:00
|
|
|
import eu.dnetlib.dhp.actionmanager.createunresolvedentities.model.FOSDataModel;
|
2021-11-10 17:00:37 +01:00
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
2022-02-01 08:38:41 +01:00
|
|
|
import eu.dnetlib.dhp.schema.common.ModelConstants;
|
2021-11-11 16:11:25 +01:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.Result;
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
|
2022-08-04 11:39:39 +02:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.Subject;
|
2022-02-01 08:38:41 +01:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
|
2021-11-15 09:58:48 +01:00
|
|
|
import eu.dnetlib.dhp.utils.DHPUtils;
|
2021-11-08 16:44:19 +01:00
|
|
|
|
2021-11-09 11:25:41 +01:00
|
|
|
public class PrepareFOSSparkJob implements Serializable {
|
2021-11-10 17:00:37 +01:00
|
|
|
private static final Logger log = LoggerFactory.getLogger(PrepareFOSSparkJob.class);
|
|
|
|
|
|
|
|
public static void main(String[] args) throws Exception {
|
|
|
|
|
|
|
|
String jsonConfiguration = IOUtils
|
|
|
|
.toString(
|
|
|
|
PrepareFOSSparkJob.class
|
|
|
|
.getResourceAsStream(
|
2021-11-12 10:18:01 +01:00
|
|
|
"/eu/dnetlib/dhp/actionmanager/createunresolvedentities/prepare_parameters.json"));
|
2021-11-10 17:00:37 +01:00
|
|
|
|
|
|
|
final ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
|
|
|
|
|
|
|
|
parser.parseArgument(args);
|
|
|
|
|
|
|
|
Boolean isSparkSessionManaged = isSparkSessionManaged(parser);
|
|
|
|
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
|
|
|
|
|
|
|
|
String sourcePath = parser.get("sourcePath");
|
|
|
|
log.info("sourcePath: {}", sourcePath);
|
|
|
|
|
|
|
|
final String outputPath = parser.get("outputPath");
|
|
|
|
log.info("outputPath: {}", outputPath);
|
|
|
|
|
|
|
|
SparkConf conf = new SparkConf();
|
|
|
|
runWithSparkSession(
|
|
|
|
conf,
|
|
|
|
isSparkSessionManaged,
|
|
|
|
spark -> {
|
|
|
|
distributeFOSdois(
|
|
|
|
spark,
|
|
|
|
sourcePath,
|
|
|
|
|
|
|
|
outputPath);
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
private static void distributeFOSdois(SparkSession spark, String sourcePath, String outputPath) {
|
|
|
|
Dataset<FOSDataModel> fosDataset = readPath(spark, sourcePath, FOSDataModel.class);
|
|
|
|
|
2021-12-22 23:10:16 +01:00
|
|
|
fosDataset
|
2021-12-23 13:24:28 +01:00
|
|
|
.groupByKey((MapFunction<FOSDataModel, String>) v -> v.getDoi().toLowerCase(), Encoders.STRING())
|
2021-12-22 23:10:16 +01:00
|
|
|
.mapGroups((MapGroupsFunction<String, FOSDataModel, Result>) (k, it) -> {
|
|
|
|
Result r = new Result();
|
|
|
|
FOSDataModel first = it.next();
|
2021-12-23 13:24:28 +01:00
|
|
|
r.setId(DHPUtils.generateUnresolvedIdentifier(k, DOI));
|
2022-02-01 08:38:41 +01:00
|
|
|
|
2021-12-22 23:10:16 +01:00
|
|
|
HashSet<String> level1 = new HashSet<>();
|
|
|
|
HashSet<String> level2 = new HashSet<>();
|
|
|
|
HashSet<String> level3 = new HashSet<>();
|
2023-10-10 09:46:40 +02:00
|
|
|
HashSet<String> level4 = new HashSet<>();
|
|
|
|
addLevels(level1, level2, level3, level4, first);
|
|
|
|
it.forEachRemaining(v -> addLevels(level1, level2, level3, level4, v));
|
2022-08-04 11:39:39 +02:00
|
|
|
List<Subject> sbjs = new ArrayList<>();
|
2023-10-10 09:46:40 +02:00
|
|
|
level1
|
|
|
|
.forEach(l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID)));
|
|
|
|
level2
|
|
|
|
.forEach(l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID)));
|
|
|
|
level3
|
|
|
|
.forEach(
|
|
|
|
l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID, true)));
|
|
|
|
level4
|
|
|
|
.forEach(
|
|
|
|
l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID, true)));
|
2021-12-22 23:10:16 +01:00
|
|
|
r.setSubject(sbjs);
|
2022-02-01 08:38:41 +01:00
|
|
|
r
|
|
|
|
.setDataInfo(
|
|
|
|
OafMapperUtils
|
|
|
|
.dataInfo(
|
|
|
|
false, null, true,
|
|
|
|
false,
|
|
|
|
OafMapperUtils
|
|
|
|
.qualifier(
|
|
|
|
ModelConstants.PROVENANCE_ENRICH,
|
|
|
|
null,
|
|
|
|
ModelConstants.DNET_PROVENANCE_ACTIONS,
|
|
|
|
ModelConstants.DNET_PROVENANCE_ACTIONS),
|
|
|
|
null));
|
2021-12-22 23:10:16 +01:00
|
|
|
return r;
|
|
|
|
}, Encoders.bean(Result.class))
|
2021-11-10 17:00:37 +01:00
|
|
|
.write()
|
|
|
|
.mode(SaveMode.Overwrite)
|
|
|
|
.option("compression", "gzip")
|
2021-11-11 16:11:25 +01:00
|
|
|
.json(outputPath + "/fos");
|
|
|
|
}
|
|
|
|
|
2023-10-10 09:46:40 +02:00
|
|
|
private static void add(List<Subject> sbsjs, Subject sbj) {
|
|
|
|
if (sbj != null)
|
|
|
|
sbsjs.add(sbj);
|
|
|
|
}
|
|
|
|
|
2021-12-22 23:10:16 +01:00
|
|
|
private static void addLevels(HashSet<String> level1, HashSet<String> level2, HashSet<String> level3,
|
2023-10-10 09:46:40 +02:00
|
|
|
HashSet<String> level4,
|
2021-12-22 23:10:16 +01:00
|
|
|
FOSDataModel first) {
|
2021-12-22 21:40:35 +01:00
|
|
|
level1.add(first.getLevel1());
|
|
|
|
level2.add(first.getLevel2());
|
2024-01-12 10:20:28 +01:00
|
|
|
if (Optional.ofNullable(first.getLevel3()).isPresent() &&
|
|
|
|
!first.getLevel3().equalsIgnoreCase(NA) && !first.getLevel3().equalsIgnoreCase(NULL)
|
|
|
|
&& first.getLevel3() != null)
|
|
|
|
level3.add(first.getLevel3() + "@@" + first.getScoreL3());
|
|
|
|
else
|
|
|
|
level3.add(NULL);
|
|
|
|
if (Optional.ofNullable(first.getLevel4()).isPresent() &&
|
|
|
|
!first.getLevel4().equalsIgnoreCase(NA) &&
|
|
|
|
!first.getLevel4().equalsIgnoreCase(NULL) &&
|
|
|
|
first.getLevel4() != null)
|
|
|
|
level4.add(first.getLevel4() + "@@" + first.getScoreL4());
|
|
|
|
else
|
|
|
|
level4.add(NULL);
|
2021-11-11 16:11:25 +01:00
|
|
|
}
|
|
|
|
|
2021-11-08 16:44:19 +01:00
|
|
|
}
|