From 9f8a80deb739893d71eb72e396d964f12f549681 Mon Sep 17 00:00:00 2001 From: Sandro La Bruzzo Date: Wed, 1 Sep 2021 14:16:27 +0200 Subject: [PATCH] fixed wrong import of unresolved relation in openaire --- .../dnetlib/dhp/actionmanager/datacite/oozie_app/workflow.xml | 2 +- .../java/eu/dnetlib/dhp/oa/provision/PrepareRelationsJob.java | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/datacite/oozie_app/workflow.xml b/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/datacite/oozie_app/workflow.xml index 021704f54..c6332ff7d 100644 --- a/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/datacite/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/datacite/oozie_app/workflow.xml @@ -70,7 +70,7 @@ --sourcePath${mainPath}/datacite_dump --targetPath${mainPath}/datacite_oaf --isLookupUrl${isLookupUrl} - --exportLinkstrue + --exportLinksfalse --masteryarn-cluster diff --git a/dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/oa/provision/PrepareRelationsJob.java b/dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/oa/provision/PrepareRelationsJob.java index ae899c3d8..fdf397ad7 100644 --- a/dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/oa/provision/PrepareRelationsJob.java +++ b/dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/oa/provision/PrepareRelationsJob.java @@ -131,6 +131,7 @@ public class PrepareRelationsJob { Set relationFilter, int sourceMaxRelations, int targetMaxRelations, int relPartitions) { JavaRDD rels = readPathRelationRDD(spark, inputRelationsPath) + .filter(rel -> !(rel.getSource().startsWith("unresolved") || rel.getTarget().startsWith("unresolved"))) .filter(rel -> !rel.getDataInfo().getDeletedbyinference()) .filter(rel -> !relationFilter.contains(StringUtils.lowerCase(rel.getRelClass())));