sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

[cleaning] fixing NPE in the country cleaning phase

This commit is contained in:
Claudio Atzori 2022-10-11 10:10:20 +02:00
parent d51275a965
commit ece40adc09
1 changed files with 1 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/country/GetDatasourceFromCountry.java
View File

@ -65,7 +65,6 @@ public class GetDatasourceFromCountry implements Serializable {
conf,
isSparkSessionManaged,
spark -> {
getDatasourceFromCountry(spark, country, inputPath, workingPath);
});
}
@ -83,7 +82,6 @@ public class GetDatasourceFromCountry implements Serializable {
(FilterFunction<Organization>) o -> !o.getDataInfo().getDeletedbyinference() &&
o.getCountry().getClassid().length() > 0 &&
o.getCountry().getClassid().equals(country));
;
// filtering of the relations taking the non deleted by inference and those with IsProvidedBy as relclass
Dataset<Relation> relation = spark
@ -97,7 +95,7 @@ public class GetDatasourceFromCountry implements Serializable {
!rel.getDataInfo().getDeletedbyinference());
organization
.joinWith(relation, organization.col("id").equalTo(relation.col("target")), "left")
.joinWith(relation, organization.col("id").equalTo(relation.col("target")))
.map((MapFunction<Tuple2<Organization, Relation>, String>) t2 -> t2._2().getSource(), Encoders.STRING())
.write()
.mode(SaveMode.Overwrite)