tests for d4science catalog

This commit is contained in:
Alessia Bardi 2023-09-19 13:38:25 +02:00
parent 77a2199837
commit 19abba8fa7
6 changed files with 237 additions and 4 deletions

View File

@ -1067,6 +1067,28 @@ class MappersTest {
System.out.println("***************");
}
@Test
public void testD4ScienceTraining() throws IOException {
final String xml = IOUtils
.toString(Objects.requireNonNull(getClass().getResourceAsStream("d4science-1-training.xml")));
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);
final OtherResearchProduct trainingMaterial = (OtherResearchProduct) list.get(0);
System.out.println("***************");
System.out.println(new ObjectMapper().writeValueAsString(trainingMaterial));
System.out.println("***************");
}
@Test
public void testD4ScienceDataset() throws IOException {
final String xml = IOUtils
.toString(Objects.requireNonNull(getClass().getResourceAsStream("d4science-2-dataset.xml")));
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);
final Dataset trainingMaterial = (Dataset) list.get(0);
System.out.println("***************");
System.out.println(new ObjectMapper().writeValueAsString(trainingMaterial));
System.out.println("***************");
}
@Test
void testNotWellFormed() throws IOException {
final String xml = IOUtils

View File

@ -0,0 +1,93 @@
<?xml version="1.0" encoding="UTF-8"?>
<oai:record xmlns:dr="http://www.driver-repository.eu/namespace/dr"
xmlns:dri="http://www.driver-repository.eu/namespace/dri"
xmlns:oaf="http://namespace.openaire.eu/oaf" xmlns:oai="http://www.openarchives.org/OAI/2.0/">
<oai:header>
<dri:objIdentifier>alessia_____::104c2d4ba8878c16fa824dce5b1bea57</dri:objIdentifier>
<dri:recordIdentifier>12d8f77e-d66f-46f5-8d88-af7db23bc4c9</dri:recordIdentifier>
<dri:dateOfCollection>2023-09-08T10:12:35.864+02:00</dri:dateOfCollection>
<oaf:datasourceprefix>alessia_____</oaf:datasourceprefix>
<dr:dateOfTransformation>2023-09-08T11:31:45.692+02:00</dr:dateOfTransformation>
</oai:header>
<oai:metadata>
<datacite:resource
xmlns:datacite="http://datacite.org/schema/kernel-4"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4/metadata.xsd">
<datacite:identifier identifierType="URL">http://data.d4science.org/ctlg/ResourceCatalogue/visual_analytics_for_data_scientists</datacite:identifier>
<datacite:alternateIdentifiers/>
<datacite:creators>
<datacite:creator>
<datacite:creatorName>BRAGHIERI MARCO</datacite:creatorName>
</datacite:creator>
</datacite:creators>
<datacite:titles>
<datacite:title>Visual Analytics for Data Scientists</datacite:title>
</datacite:titles>
<datacite:publisher>SoBigData++</datacite:publisher>
<datacite:publicationYear/>
<datacite:dates>
<datacite:date dateType="Issued"/>
</datacite:dates>
<datacite:resourceType resourceTypeGeneral="TrainingMaterial">TrainingMaterial</datacite:resourceType>
<datacite:descriptions>
<datacite:description descriptionType="Abstract">Participants to this module shall
- Learn the principles and rules underlying the design of visual data
representations and human-computer interactions
- Understand, adapt and apply representative visual analytics methods and systems for diverse types
of data and problems
- Analyse and evaluate the structure and properties
of data to select or devise appropriate methods for data exploration
- Combine visualization, interactive techniques, and computational
processing to develop practical data analysis for problem solving
(This teaching material on Visual Analytics for Data Scientists is part of a MSc module at City University London).
The author did not intend to violate any copyright on figures or content. In case you are the legal owner of any copyrighted content, please contact info@sobigdata.eu and we will immediately remove it</datacite:description>
</datacite:descriptions>
<datacite:subjects>
<datacite:subject>Visual analytics</datacite:subject>
</datacite:subjects>
<datacite:formats>
<datacite:format>Slides</datacite:format>
<datacite:format>Other</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>PDF</datacite:format>
<datacite:format>ZIP</datacite:format>
</datacite:formats>
</datacite:resource>
<oaf:accessrights>OPEN</oaf:accessrights>
<dr:CobjCategory type="other">0010</dr:CobjCategory>
<oaf:dateAccepted/>
<oaf:hostedBy id="alessia_____::alessia" name="Alessia"/>
<oaf:collectedFrom id="alessia_____::alessia" name="Alessia"/>
<oaf:license>other-open</oaf:license>
<oaf:projectid>corda__h2020::871042</oaf:projectid>
</oai:metadata>
<about xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:prov="http://www.openarchives.org/OAI/2.0/provenance" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<provenance xmlns="http://www.openarchives.org/OAI/2.0/provenance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/provenance http://www.openarchives.org/OAI/2.0/provenance.xsd">
<originDescription altered="true" harvestDate="2023-09-08T10:12:35.864+02:00">
<baseURL>https%3A%2F%2Fapi.d4science.org%2Fcatalogue%2Fitems</baseURL>
<identifier/>
<datestamp/>
<metadataNamespace/>
</originDescription>
</provenance>
<oaf:datainfo>
<oaf:inferred>false</oaf:inferred>
<oaf:deletedbyinference>false</oaf:deletedbyinference>
<oaf:trust>0.9</oaf:trust>
<oaf:inferenceprovenance/>
<oaf:provenanceaction classid="sysimport:crosswalk"
classname="Harvested" schemeid="dnet:provenanceActions" schemename="dnet:provenanceActions"/>
</oaf:datainfo>
</about>
</oai:record>

View File

@ -0,0 +1,72 @@
<?xml version="1.0" encoding="UTF-8"?>
<oai:record xmlns:dr="http://www.driver-repository.eu/namespace/dr"
xmlns:dri="http://www.driver-repository.eu/namespace/dri"
xmlns:oaf="http://namespace.openaire.eu/oaf" xmlns:oai="http://www.openarchives.org/OAI/2.0/">
<oai:header>
<dri:objIdentifier>alessia_____::028879484548f4e1c630e1c503e35231</dri:objIdentifier>
<dri:recordIdentifier>4fed018e-c2ff-4afa-b7b5-1ca1beebf850</dri:recordIdentifier>
<dri:dateOfCollection>2023-09-08T12:14:27.615+02:00</dri:dateOfCollection>
<oaf:datasourceprefix>alessia_____</oaf:datasourceprefix>
<dr:dateOfTransformation>2023-09-08T12:14:51.7+02:00</dr:dateOfTransformation>
</oai:header>
<oai:metadata>
<datacite:resource
xmlns:datacite="http://datacite.org/schema/kernel-4"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4/metadata.xsd">
<datacite:identifier identifierType="URL">http://data.d4science.org/ctlg/ResourceCatalogue/city-to-city_migration</datacite:identifier>
<datacite:alternateIdentifiers>
<datacite:alternateIdentifier type="URL"/>
</datacite:alternateIdentifiers>
<datacite:creators>
<datacite:creator>
<datacite:creatorName>Pappalardo, Luca</datacite:creatorName>
<datacite:affiliation/>
<datacite:nameIdentifier nameIdentifierScheme="ORCID" schemeURI="http://orcid.org">0000-0002-1547-6007</datacite:nameIdentifier>
</datacite:creator>
</datacite:creators>
<datacite:titles>
<datacite:title>City-to-city migration</datacite:title>
</datacite:titles>
<datacite:publisher>SoBigData++</datacite:publisher>
<datacite:publicationYear/>
<datacite:dates>
<datacite:date dateType="Issued">2018-02-15</datacite:date>
</datacite:dates>
<datacite:resourceType resourceTypeGeneral="Dataset">Dataset</datacite:resourceType>
<datacite:descriptions>
<datacite:description descriptionType="Abstract">Census data recording the migration of people between metropolitan areas in
the US</datacite:description>
</datacite:descriptions>
<datacite:subjects>
<datacite:subject>Human Mobility data</datacite:subject>
</datacite:subjects>
<datacite:formats/>
</datacite:resource>
<oaf:accessrights>OPEN</oaf:accessrights>
<dr:CobjCategory type="dataset">0021</dr:CobjCategory>
<oaf:dateAccepted>2018-02-15</oaf:dateAccepted>
<oaf:hostedBy id="alessia_____::alessia" name="Alessia"/>
<oaf:collectedFrom id="alessia_____::alessia" name="Alessia"/>
<oaf:license>AFL-3.0</oaf:license>
<oaf:projectid>corda__h2020::871042</oaf:projectid>
</oai:metadata>
<about xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:prov="http://www.openarchives.org/OAI/2.0/provenance" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<provenance xmlns="http://www.openarchives.org/OAI/2.0/provenance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/provenance http://www.openarchives.org/OAI/2.0/provenance.xsd">
<originDescription altered="true" harvestDate="2023-09-08T12:14:27.615+02:00">
<baseURL>https%3A%2F%2Fapi.d4science.org%2Fcatalogue%2Fitems</baseURL>
<identifier/>
<datestamp/>
<metadataNamespace/>
</originDescription>
</provenance>
<oaf:datainfo>
<oaf:inferred>false</oaf:inferred>
<oaf:deletedbyinference>false</oaf:deletedbyinference>
<oaf:trust>0.9</oaf:trust>
<oaf:inferenceprovenance/>
<oaf:provenanceaction classid="sysimport:crosswalk"
classname="Harvested" schemeid="dnet:provenanceActions" schemename="dnet:provenanceActions"/>
</oaf:datainfo>
</about>
</oai:record>

View File

@ -24,10 +24,7 @@ import eu.dnetlib.dhp.oa.provision.model.RelatedEntity;
import eu.dnetlib.dhp.oa.provision.model.RelatedEntityWrapper;
import eu.dnetlib.dhp.oa.provision.utils.ContextMapper;
import eu.dnetlib.dhp.oa.provision.utils.XmlRecordFactory;
import eu.dnetlib.dhp.schema.oaf.Datasource;
import eu.dnetlib.dhp.schema.oaf.Project;
import eu.dnetlib.dhp.schema.oaf.Publication;
import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.schema.oaf.*;
public class XmlRecordFactoryTest {
@ -196,4 +193,51 @@ public class XmlRecordFactoryTest {
assertEquals("dnet:pid_types", ((Element) pids.get(0)).attribute("schemeid").getValue());
assertEquals("dnet:pid_types", ((Element) pids.get(0)).attribute("schemename").getValue());
}
@Test
public void testD4ScienceTraining() throws DocumentException, IOException {
final ContextMapper contextMapper = new ContextMapper();
final XmlRecordFactory xmlRecordFactory = new XmlRecordFactory(contextMapper, false,
XmlConverterJob.schemaLocation);
final OtherResearchProduct p = OBJECT_MAPPER
.readValue(
IOUtils.toString(getClass().getResourceAsStream("d4science-1-training.json")),
OtherResearchProduct.class);
final String xml = xmlRecordFactory.build(new JoinedEntity<>(p));
assertNotNull(xml);
final Document doc = new SAXReader().read(new StringReader(xml));
assertNotNull(doc);
System.out.println(doc.asXML());
}
@Test
public void testD4ScienceDataset() throws DocumentException, IOException {
final ContextMapper contextMapper = new ContextMapper();
final XmlRecordFactory xmlRecordFactory = new XmlRecordFactory(contextMapper, false,
XmlConverterJob.schemaLocation);
final OtherResearchProduct p = OBJECT_MAPPER
.readValue(
IOUtils.toString(getClass().getResourceAsStream("d4science-2-dataset.json")),
OtherResearchProduct.class);
final String xml = xmlRecordFactory.build(new JoinedEntity<>(p));
assertNotNull(xml);
final Document doc = new SAXReader().read(new StringReader(xml));
assertNotNull(doc);
System.out.println(doc.asXML());
}
}

File diff suppressed because one or more lines are too long

File diff suppressed because one or more lines are too long