andreas.czerniak
/
BrBETA_dnet-hadoop
forked from D-Net/dnet-hadoop
1
1
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge pull request 'relation-from-odf' (#251) from relation-from-odf into beta 

Reviewed-on: D-Net/dnet-hadoop#251
This commit is contained in:
Claudio Atzori 2022-09-27 15:08:26 +02:00
parent b4b6a4457c 3f90d159e3
commit de7bc9350e
7 changed files with 103 additions and 32 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

@ -30,6 +30,7 @@ import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.*;
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.dhp.schema.oaf.utils.PidType;
public abstract class AbstractMdRecordToOafMapper {
@ -68,6 +69,17 @@ 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;

66
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
View File

@ -22,9 +22,12 @@ 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;
import eu.dnetlib.dhp.schema.oaf.utils.PidType;
public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
@ -393,40 +396,53 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
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(

49
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
@ -912,13 +930,13 @@ class MappersTest {
}
@Test
void testROHub() throws IOException, DocumentException {
void testROHub() throws IOException {
final String xml = IOUtils.toString(Objects.requireNonNull(getClass().getResourceAsStream("rohub.xml")));
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);
System.out.println("***************");
System.out.println(new ObjectMapper().writeValueAsString(list));
System.out.println("***************");
// final Dataset p = (Dataset) list.get(0);
// final OtherResearchProduct p = (OtherResearchProduct) list.get(0);
// assertValidId(p.getId());
// assertValidId(p.getCollectedfrom().get(0).getKey());
// System.out.println(p.getTitle().get(0).getValue());
@ -926,13 +944,36 @@ class MappersTest {
}
@Test
void testROHub2() throws IOException, DocumentException {
void testROHub2() throws IOException {
final String xml = IOUtils
.toString(Objects.requireNonNull(getClass().getResourceAsStream("rohub-modified.xml")));
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);
System.out.println("***************");
System.out.println(new ObjectMapper().writeValueAsString(list));
System.out.println("***************");
assertEquals(7, list.size());
final OtherResearchProduct p = (OtherResearchProduct) list.get(0);
assertValidId(p.getId());
assertValidId(p.getCollectedfrom().get(0).getKey());
assertEquals("50|w3id________::afc7592914ae190a50570db90f55f9c2", p.getId());
assertTrue(StringUtils.isNotBlank(p.getTitle().get(0).getValue()));
assertEquals("w3id", (p.getPid().get(0).getQualifier().getClassid()));
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(6, list.stream().filter(o -> o instanceof Relation).count());
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).getRelType());
}
}
}
@Test

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>

2
pom.xml
View File

@ -807,7 +807,7 @@
<mockito-core.version>3.3.3</mockito-core.version>
<mongodb.driver.version>3.4.2</mongodb.driver.version>
<vtd.version>[2.12,3.0)</vtd.version>
<dhp-schemas.version>[3.14.0]</dhp-schemas.version>
<dhp-schemas.version>[3.15.0]</dhp-schemas.version>
<dnet-actionmanager-api.version>[4.0.3]</dnet-actionmanager-api.version>
<dnet-actionmanager-common.version>[6.0.5]</dnet-actionmanager-common.version>
<dnet-openaire-broker-common.version>[3.1.6]</dnet-openaire-broker-common.version>