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

extended mapping from ODF relations (PR#251)

This commit is contained in:
Claudio Atzori 2022-11-14 18:26:13 +01:00
parent a929dc5fee
commit eb45ba7af0
6 changed files with 61 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -72,6 +72,17 @@ public abstract class AbstractMdRecordToOafMapper {
nsContext.put("datacite", DATACITE_SCHEMA_KERNEL_3);
}
// lowercase pidTypes as keys, normal casing for the values
protected static final Map<String, String> pidTypeWithAuthority = new HashMap<>();
static {
IdentifierFactory.PID_AUTHORITY
.keySet()
.stream()
.forEach(entry -> pidTypeWithAuthority.put(entry.toString().toLowerCase(), entry.toString()));
}
protected AbstractMdRecordToOafMapper(final VocabularyGroup vocs, final boolean invisible,
final boolean shouldHashId, final boolean forceOriginalId) {
this.vocs = vocs;

69
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
View File

@ -10,6 +10,8 @@ import java.net.URLDecoder;
import java.util.*;
import java.util.stream.Collectors;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.common.RelationInverse;
import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document;
import org.dom4j.Element;
@ -377,48 +379,61 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
@Override
protected List<Oaf> addOtherResultRels(
final Document doc,
final OafEntity entity) {
final Document doc,
final OafEntity entity) {
final String docId = entity.getId();
final List<Oaf> res = new ArrayList<>();
for (final Object o : doc
.selectNodes("//*[local-name()='relatedIdentifier' and ./@relatedIdentifierType='OPENAIRE']")) {
.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");
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
final String idType = ((Node) o).valueOf("@relatedIdentifierType");
final String relType = ((Node) o).valueOf("@relationType");
String otherId = guessRelatedIdentifier(idType, originalId);
if (StringUtils.isNotBlank(otherId)) {
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.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(

8
dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java
View File

@ -579,7 +579,7 @@ 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);
final Software s = (Software) list.get(0);
@ -918,7 +918,7 @@ class MappersTest {
System.out.println("***************");
System.out.println(new ObjectMapper().writeValueAsString(list));
System.out.println("***************");
assertEquals(1, list.size());
assertEquals(5, list.size());
final OtherResearchProduct p = (OtherResearchProduct) list.get(0);
assertValidId(p.getId());
assertTrue(p.getId().startsWith("50|w3id"));
@ -945,7 +945,7 @@ class MappersTest {
System.out.println("***************");
System.out.println(new ObjectMapper().writeValueAsString(list));
System.out.println("***************");
assertEquals(1, list.size());
assertEquals(7, list.size());
final OtherResearchProduct p = (OtherResearchProduct) list.get(0);
assertValidId(p.getId());
assertValidId(p.getCollectedfrom().get(0).getKey());
@ -955,7 +955,7 @@ class MappersTest {
assertEquals("https://w3id.org/ro-id/0ab171a7-45c5-4194-82d4-850955504bca", (p.getPid().get(0).getValue()));
assertEquals(1, list.stream().filter(o -> o instanceof OtherResearchProduct).count());
assertEquals(0, list.stream().filter(o -> o instanceof Relation).count());
assertEquals(6, list.stream().filter(o -> o instanceof Relation).count());
for (Oaf oaf : list) {
if (oaf instanceof Relation) {

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>

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>