D-Net
/
dnet-hadoop
13
4
Fork 10
Code Issues 14 Pull Requests 7 Projects 2 Releases 1 Wiki Activity

Merge remote-tracking branch 'origin/beta' into scholix_to_solr

This commit is contained in:
Sandro La Bruzzo 2022-10-11 10:58:53 +02:00
parent 4b8739e45b ece40adc09
commit 818a936468
2 changed files with 4 additions and 6 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)

6
dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/clean/oozie_app/workflow.xml
View File

@ -492,7 +492,7 @@
<arg>--datasourcePath</arg><arg>${workingDir}/working/hostedby</arg>
<arg>--collectedfrom</arg><arg>${collectedfrom}</arg>
</spark>
<ok to="wait_clean_context"/>
<ok to="wait_clean_country"/>
<error to="Kill"/>
</action>
@ -521,7 +521,7 @@
<arg>--datasourcePath</arg><arg>${workingDir}/working/hostedby</arg>
<arg>--collectedfrom</arg><arg>${collectedfrom}</arg>
</spark>
<ok to="wait_clean_context"/>
<ok to="wait_clean_country"/>
<error to="Kill"/>
</action>
@ -550,7 +550,7 @@
<arg>--datasourcePath</arg><arg>${workingDir}/working/hostedby</arg>
<arg>--collectedfrom</arg><arg>${collectedfrom}</arg>
</spark>
<ok to="wait_clean_context"/>
<ok to="wait_clean_country"/>
<error to="Kill"/>
</action>