added control to check if the entity exists

This commit is contained in:
Sandro La Bruzzo 2021-07-22 16:08:54 +02:00
parent 62ae36a3d2
commit 058b636d4d
1 changed files with 39 additions and 32 deletions

View File

@ -23,6 +23,7 @@ import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.common.HdfsSupport;
import eu.dnetlib.dhp.schema.common.EntityType;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.*;
@ -77,17 +78,22 @@ public class SparkUpdateEntity extends AbstractSparkAction {
(type, clazz) -> {
final String outputPath = dedupGraphPath + "/" + type;
removeOutputDir(spark, outputPath);
final String ip = DedupUtility.createEntityPath(graphBasePath, type.toString());
if (HdfsSupport.exists(ip, sc.hadoopConfiguration())) {
JavaRDD<String> sourceEntity = sc
.textFile(DedupUtility.createEntityPath(graphBasePath, type.toString()));
if (mergeRelExists(workingPath, type.toString())) {
final String mergeRelPath = DedupUtility.createMergeRelPath(workingPath, "*", type.toString());
final String mergeRelPath = DedupUtility
.createMergeRelPath(workingPath, "*", type.toString());
final String dedupRecordPath = DedupUtility
.createDedupRecordPath(workingPath, "*", type.toString());
final Dataset<Relation> rel = spark.read().load(mergeRelPath).as(Encoders.bean(Relation.class));
final Dataset<Relation> rel = spark
.read()
.load(mergeRelPath)
.as(Encoders.bean(Relation.class));
final JavaPairRDD<String, String> mergedIds = rel
.where("relClass == 'merges'")
@ -119,6 +125,7 @@ public class SparkUpdateEntity extends AbstractSparkAction {
}
sourceEntity.saveAsTextFile(outputPath, GzipCodec.class);
}
});
}