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

Mapping all relationships supported in ModelConstants and ModelSupport

This commit is contained in:
Alessia Bardi 2022-09-26 11:24:13 +02:00
parent c5eb722170
commit fd63e9bfac
5 changed files with 59 additions and 62 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -69,13 +69,15 @@ public abstract class AbstractMdRecordToOafMapper {
nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3);
}
protected static final Set<String> pidTypeWithAuthority = new HashSet<>();
// lowercase pidTypes as keys, normal casing for the values
protected static final Map<String, String> pidTypeWithAuthority = new HashMap<>();
static {
pidTypeWithAuthority.addAll(IdentifierFactory.PID_AUTHORITY.keySet().stream()
.map(PidType::toString)
.map(String::toLowerCase)
.collect(Collectors.toCollection(HashSet::new)));
IdentifierFactory.PID_AUTHORITY
.keySet()
.stream()
.forEach(entry -> pidTypeWithAuthority.put(entry.toString().toLowerCase(), entry.toString()));
}
protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible,

74
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
View File

@ -22,6 +22,8 @@ import com.google.common.collect.Lists;
import eu.dnetlib.dhp.common.PacePerson;
import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.common.RelationInverse;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions;
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
@ -400,71 +402,47 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
if (StringUtils.isNotBlank(originalId)) {
final String idType = ((Node) o).valueOf("@relatedIdentifierType");
final String reltype = ((Node) o).valueOf("@relationType");
final String relType = ((Node) o).valueOf("@relationType");
String otherId = guessRelatedIdentifier(idType, originalId);
if (StringUtils.isNotBlank(otherId)) {
if (reltype.equalsIgnoreCase(IS_SUPPLEMENT_TO)) {
res
.add(
getRelation(
docId, otherId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENT_TO, entity));
res
.add(
getRelation(
otherId, docId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENTED_BY, entity));
} else {
if (reltype.equalsIgnoreCase(IS_SUPPLEMENTED_BY)) {
res
.add(
getRelation(
otherId, docId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENT_TO, entity));
res
.add(
getRelation(
docId, otherId, RESULT_RESULT, SUPPLEMENT, IS_SUPPLEMENTED_BY, entity));
} else {
if (reltype.equalsIgnoreCase(IS_PART_OF)) {
res
.add(
getRelation(
docId, otherId, RESULT_RESULT, PART, IS_PART_OF, entity));
res
.add(
getRelation(
otherId, docId, RESULT_RESULT, PART, HAS_PART, entity));
} else {
if (reltype.equalsIgnoreCase(HAS_PART)) {
res
.add(
getRelation(
otherId, docId, RESULT_RESULT, PART, IS_PART_OF, entity));
res
.add(
getRelation(
docId, otherId, RESULT_RESULT, PART, HAS_PART, entity));
}
// else TODO catch more semantics
}
}
res.addAll(getRelations(relType, docId, otherId, entity));
}
}
}
}
return res;
}
protected String guessRelatedIdentifier(final String idType, final String value) {
if (StringUtils.isBlank(idType) || StringUtils.isBlank(value))
return null;
if (idType.equalsIgnoreCase("OPENAIRE")) return createOpenaireId(50, value, false);
if(pidTypeWithAuthority.contains(idType.toLowerCase())){
return IdentifierFactory.idFromPid("50", idType, value, true);
if (idType.equalsIgnoreCase("OPENAIRE"))
return createOpenaireId(50, value, false);
if (pidTypeWithAuthority.containsKey(idType.toLowerCase())) {
return IdentifierFactory.idFromPid("50", pidTypeWithAuthority.get(idType.toLowerCase()), value, true);
}
return null;
}
protected List<Oaf> getRelations(final String reltype, final String entityId, final String otherId,
final OafEntity entity) {
final List<Oaf> res = new ArrayList<>();
RelationInverse rel = ModelSupport.findRelation(reltype);
if (rel != null) {
res
.add(
getRelation(
entityId, otherId, rel.getRelType(), rel.getSubReltype(), rel.getRelClass(), entity));
res
.add(
getRelation(
otherId, entityId, rel.getRelType(), rel.getSubReltype(), rel.getInverseRelClass(), entity));
}
return res;
}
@Override
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
return prepareQualifier(

26
dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java
View File

@ -579,8 +579,10 @@ class MappersTest {
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);
assertEquals(1, list.size());
assertEquals(3, list.size());
assertTrue(list.get(0) instanceof Software);
assertTrue(list.get(1) instanceof Relation);
assertTrue(list.get(2) instanceof Relation);
final Software s = (Software) list.get(0);
@ -590,6 +592,22 @@ class MappersTest {
assertTrue(s.getAuthor().size() > 0);
assertTrue(s.getSubject().size() > 0);
assertTrue(s.getInstance().size() > 0);
final Relation r1 = (Relation) list.get(1);
final Relation r2 = (Relation) list.get(2);
assertEquals(s.getId(), r1.getSource());
assertEquals("50|doi_________::b453e7b4b2130ace57ff0c3db470a982", r1.getTarget());
assertEquals(ModelConstants.RESULT_RESULT, r1.getRelType());
assertEquals(ModelConstants.RELATIONSHIP, r1.getSubRelType());
assertEquals(ModelConstants.IS_REFERENCED_BY, r1.getRelClass());
assertEquals(s.getId(), r2.getTarget());
assertEquals("50|doi_________::b453e7b4b2130ace57ff0c3db470a982", r2.getSource());
assertEquals(ModelConstants.RESULT_RESULT, r2.getRelType());
assertEquals(ModelConstants.RELATIONSHIP, r2.getSubRelType());
assertEquals(ModelConstants.REFERENCES, r2.getRelClass());
}
@Test
@ -945,14 +963,14 @@ class MappersTest {
assertEquals(1, list.stream().filter(o -> o instanceof OtherResearchProduct).count());
assertEquals(6, list.stream().filter(o -> o instanceof Relation).count());
for(Oaf oaf : list){
if(oaf instanceof Relation){
for (Oaf oaf : list) {
if (oaf instanceof Relation) {
String source = ((Relation) oaf).getSource();
String target = ((Relation) oaf).getTarget();
assertNotEquals(source, target);
assertTrue(source.equals(p.getId()) || target.equals(p.getId()));
assertNotNull(((Relation) oaf).getSubRelType());
assertNotNull( ((Relation) oaf).getRelClass());
assertNotNull(((Relation) oaf).getRelClass());
assertNotNull(((Relation) oaf).getRelType());
}
}

1
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_dataset.xml
View File

@ -69,7 +69,6 @@
</dates>
<resourceType resourceTypeGeneral="Dataset"/>
<relatedIdentifiers>
<relatedIdentifier relatedIdentifierType="DOI" relationType="IsVersionOf">10.5281/zenodo.3234525</relatedIdentifier>
<relatedIdentifier relatedIdentifierType="URL" relationType="IsPartOf">https://zenodo.org/communities/epfl</relatedIdentifier>
</relatedIdentifiers>
<version>1.0.0</version>

2
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_software.xml
View File

@ -20,7 +20,7 @@
<datacite:publisher>bio.tools</datacite:publisher>
<datacite:relatedIdentifiers>
<datacite:relatedIdentifier relatedIdentifierType="URL" relationType="IsDocumentedBy">http://maplab.imppc.org/chainy/</datacite:relatedIdentifier>
<datacite:relatedIdentifier relatedIdentifierType="DOI" relationType="isReferencedBy">10.1093/bioinformatics/btw839</datacite:relatedIdentifier>
<datacite:relatedIdentifier relatedIdentifierType="DOI" relationType="isreferencedBy">10.1093/bioinformatics/btw839</datacite:relatedIdentifier>
</datacite:relatedIdentifiers>
<datacite:alternateIdentifiers>
<datacite:alternateIdentifier alternateIdentifierType="LandingPage">https://bio.tools/</datacite:alternateIdentifier>