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

refactoring for the generation of relationships from related identifier of type 'OPENAIRE'

This commit is contained in:
Alessia Bardi 2022-09-23 15:17:13 +02:00
parent 982bcc1e35
commit ba33ff71fd
4 changed files with 66 additions and 58 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/MigrateDbEntitiesApplication.java
View File

@ -422,7 +422,8 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final Relation r2 = OafMapperUtils
.getRelation(
orgId, dsId, DATASOURCE_ORGANIZATION, PROVISION, PROVIDES, collectedFrom, info, lastUpdateTimestamp);
orgId, dsId, DATASOURCE_ORGANIZATION, PROVISION, PROVIDES, collectedFrom, info,
lastUpdateTimestamp);
return Arrays.asList(r1, r2);
} catch (final Exception e) {

117
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
View File

@ -25,6 +25,7 @@ import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions;
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
import eu.dnetlib.dhp.schema.oaf.utils.PidType;
public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
@ -391,75 +392,77 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
final String docId = entity.getId();
final List<Oaf> res = new ArrayList<>();
/*
/*
<datacite:relatedIdentifiers>
<datacite:relatedIdentifier relatedIdentifierType="w3id" relationType="HasPart">https://w3id.org/ro-id/13c54585-362e-4925-a785-08afb591fa0d/resources/b4be0f3e-41d7-471f-b34e-f0bd54ff5698</datacite:relatedIdentifier>
<datacite:relatedIdentifier relatedIdentifierType="w3id" relationType="HasPart">https://w3id.org/ro-id/13c54585-362e-4925-a785-08afb591fa0d/resources/5d6e575b-ef84-417a-9d76-61c6702f7cb2</datacite:relatedIdentifier>
<datacite:relatedIdentifier relatedIdentifierType="w3id" relationType="HasPart">https://w3id.org/ro-id/13c54585-362e-4925-a785-08afb591fa0d/resources/35e01545-8c6d-49bd-ab98-5c152df69934</datacite:relatedIdentifier>
</datacite:relatedIdentifiers>
We could extend it to create the relationships targeting w3id, dois, pmcids and other pid types for which we know how to build the target openaire identifier "blindly".
for (final Object o : doc
.selectNodes("//*[local-name()='relatedIdentifier']")) {
.selectNodes("//*[local-name()='relatedIdentifier']")) {
final String originalId = ((Node) o).getText();
final String originalId = ((Node) o).getText().trim();
if (StringUtils.isNotBlank(originalId)) {
final String otherId = createOpenaireId(50, originalId, false);
final String type = ((Node) o).valueOf("@relationType");
switch(type){
case IS_SUPPLEMENT_TO:
break;
case SUPPLEMENT:
break;
case IS_PART_OF:
break;
case HAS_PART:
break;
final String idType = ((Node) o).valueOf("@relatedIdentifierType");
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
}
}
}
}
*/
for (final Object o : doc
.selectNodes("//*[local-name()='relatedIdentifier' and ./@relatedIdentifierType='OPENAIRE']")) {
final String originalId = ((Node) o).getText();
if (StringUtils.isNotBlank(originalId)) {
final String otherId = createOpenaireId(50, originalId, false);
final String type = ((Node) o).valueOf("@relationType");
if (type.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 (type.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 {
// TODO catch more semantics
}
}
}
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);
} else
return null;
}
@Override
protected Qualifier prepareResourceType(final Document doc, final DataInfo info) {
return prepareQualifier(

1
dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java
View File

@ -933,6 +933,7 @@ class MappersTest {
System.out.println("***************");
System.out.println(new ObjectMapper().writeValueAsString(list));
System.out.println("***************");
assertEquals(3, list.size());
final OtherResearchProduct p = (OtherResearchProduct) list.get(0);
assertValidId(p.getId());
assertValidId(p.getCollectedfrom().get(0).getKey());

3
dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/rohub-modified.xml
View File

@ -30,6 +30,9 @@
<datacite:relatedIdentifier relatedIdentifierType="w3id" relationType="HasPart">
https://w3id.org/ro-id/0ab171a7-45c5-4194-82d4-850955504bca/resources/6d3427a8-352e-49f4-9796-f618c44dc16d
</datacite:relatedIdentifier>
<datacite:relatedIdentifier relatedIdentifierType="OPENAIRE" relationType="isSupplementedBy">
fsh_____4119::afc7592914ae190a50570db90f55f9c3
</datacite:relatedIdentifier>
</datacite:relatedIdentifiers>
<datacite:resourceType xs:anyURI="http://purl.org/coar/resource_type/c_1843">RO-crate</datacite:resourceType>
<datacite:rightsList>