D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki Activity

[cleaning wf] fixed OAF record navigation, a mapping defined on a container object would have prevented the natvigation to continue on its properties

This commit is contained in:
Claudio Atzori 2021-12-01 15:49:15 +01:00
parent 4fe7888817
commit d85af6fc25
4 changed files with 20 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/OafCleaner.java
View File

@ -30,6 +30,11 @@ public class OafCleaner implements Serializable {
}
} else if (hasMapping(o, mapping)) {
mapping.get(o.getClass()).accept(o);
for (final Field f : getAllFields(o.getClass())) {
f.setAccessible(true);
final Object val = f.get(o);
navigate(val, mapping);
}
} else {
for (final Field f : getAllFields(o.getClass())) {
f.setAccessible(true);

3
dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/clean/GraphCleaningFunctionsTest.java
View File

@ -68,6 +68,9 @@ public class GraphCleaningFunctionsTest {
Relation r_out = OafCleaner.apply(r_in, mapping);
assertTrue(vocabularies.getTerms(ModelConstants.DNET_RELATION_RELCLASS).contains(r_out.getRelClass()));
assertTrue(vocabularies.getTerms(ModelConstants.DNET_RELATION_SUBRELTYPE).contains(r_out.getSubRelType()));
assertEquals("iis", r_out.getDataInfo().getProvenanceaction().getClassid());
assertEquals("Inferred by OpenAIRE", r_out.getDataInfo().getProvenanceaction().getClassname());
}
}

20
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/clean/relation.json
View File

@ -1,10 +1,10 @@
{"relType":"resultResult","subRelType":"citation","relClass":"cites","source":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","target":"50|openaire____::007a4870b31056f89b768cf508e1538e"}
{"relType":"resultResult","subRelType":"citation","relClass":"isCitedBy","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"supplement","relClass":"isSupplementTo","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"supplement","relClass":"isSupplementedBy","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"part","relClass":"isPartOf","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"part","relClass":"hasPart","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"review","relClass":"isReviewedBy","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"review","relClass":"reviews","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"relationship","relClass":"isRelatedTo","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"publicationDataset","relClass":"isRelatedTo","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556"}
{"relType":"resultResult","subRelType":"citation","relClass":"cites","source":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","target":"50|openaire____::007a4870b31056f89b768cf508e1538e","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"citation","relClass":"isCitedBy","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"supplement","relClass":"isSupplementTo","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"supplement","relClass":"isSupplementedBy","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"part","relClass":"isPartOf","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"part","relClass":"hasPart","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"review","relClass":"isReviewedBy","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"review","relClass":"reviews","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"relationship","relClass":"isRelatedTo","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}
{"relType":"resultResult","subRelType":"publicationDataset","relClass":"isRelatedTo","source":"50|openaire____::007a4870b31056f89b768cf508e1538e","target":"50|4ScienceCRIS::f66f1bd369679b5b077dcdf006089556","dataInfo": {"provenanceaction": {"classid": "iis", "classname": "erroneous label to be cleaned","schemeid": "dnet:provenanceActions", "schemename": "dnet:provenanceActions"}}}

1
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/clean/synonyms.txt
View File

@ -1242,3 +1242,4 @@ dnet:relation_relClass @=@ IsSupplementTo @=@ isSupplementTo
dnet:relation_relClass @=@ IsSupplementedBy @=@ isSupplementedBy
dnet:relation_relClass @=@ IsRelatedTo @=@ isRelatedTo
dnet:relation_subRelType @=@ relationship @=@ publicationDataset
dnet:provenanceActions @=@ iis @=@ erroneous label to be cleaned