Search in sources :

Example 66 with Dataset

use of com.google.cloud.datalabeling.v1beta1.Dataset in project geonetwork-microservices by geonetwork.

the class DcatConverter method convert.

/**
 * Convert an index document into a DCAT object.
 */
public CatalogRecord convert(JsonNode doc) {
    CatalogRecord catalogRecord = null;
    Dataset dcatDataset = null;
    try {
        IndexRecord record = new ObjectMapper().enable(DeserializationFeature.ACCEPT_SINGLE_VALUE_AS_ARRAY).readValue(doc.get(IndexRecordFieldNames.source).toString(), IndexRecord.class);
        String recordIdentifier = record.getMetadataIdentifier();
        String recordUri = formatterConfiguration.buildLandingPageLink(record.getMetadataIdentifier());
        Optional<ResourceIdentifier> resourceIdentifier = record.getResourceIdentifier().stream().filter(Objects::nonNull).findFirst();
        // TODO: Define strategy to build IRI
        final String resourceIdentifierUri = resourceIdentifier.isPresent() ? "local:" + resourceIdentifier.get().getCode() : null;
        String language = record.getMainLanguage() == null ? defaultLanguage : record.getMainLanguage();
        String languageUpperCase = language.toUpperCase();
        // TODO: Need language mapper
        String iso2letterLanguage = language.substring(0, 2);
        List<String> resourceLanguage = record.getResourceLanguage();
        List<String> resourceType = record.getResourceType();
        boolean isInspireResource = resourceType.contains("dataset") || resourceType.contains("series") || resourceType.contains("service");
        // TODO: Add multilingual support
        // TODO .resource("https://creativecommons.org/publicdomain/zero/1.0/deed")
        DatasetBuilder datasetBuilder = Dataset.builder().identifier(record.getResourceIdentifier().stream().map(c -> c.getCode()).collect(Collectors.toList())).title(listOfNullable(record.getResourceTitle().get(defaultText))).description(listOfNullable(record.getResourceAbstract().get(defaultText))).landingPage(listOfNullable(DcatDocument.builder().foafDocument(FoafDocument.builder().about(formatterConfiguration.buildLandingPageLink(record.getMetadataIdentifier())).title(record.getResourceTitle().get(defaultText)).build()).build())).provenance(record.getResourceLineage().stream().map(l -> Provenance.builder().provenanceStatement(ProvenanceStatement.builder().label(l.get(defaultText)).build()).build()).collect(Collectors.toList())).type(record.getResourceType().stream().map(t -> new RdfResource(null, "dcat:" + RESSOURCE_TYPE_MAPPING.get(t), null)).collect(Collectors.toList())).modified(toDate(record.getChangeDate())).theme(record.getCodelists().get(topic).stream().map(t -> Subject.builder().skosConcept(SkosConcept.builder().prefLabel(t.getProperties().get(defaultText)).build()).build()).collect(Collectors.toList())).theme(record.getTag().stream().map(t -> Subject.builder().skosConcept(SkosConcept.builder().prefLabel(t.get(defaultText)).build()).build()).collect(Collectors.toList()));
        record.getResourceDate().stream().filter(d -> "creation".equals(d.getType())).forEach(d -> datasetBuilder.created(toDate(d.getDate())));
        record.getResourceDate().stream().filter(d -> "publication".equals(d.getType())).forEach(d -> datasetBuilder.issued(toDate(d.getDate())));
        record.getResourceDate().stream().filter(d -> "revision".equals(d.getType())).forEach(d -> datasetBuilder.modified(toDate(d.getDate())));
        // TODO: Convert to meter ?
        datasetBuilder.spatialResolutionInMeters(record.getResolutionScaleDenominator().stream().map(BigDecimal::new).collect(Collectors.toList()));
        // INSPIRE
        if (record.getSpecificationConformance().size() > 0) {
            datasetBuilder.wasUsedBy(record.getSpecificationConformance().stream().map(c -> DcatActivity.builder().activity(// https://github.com/SEMICeu/iso-19139-to-dcat-ap/blob/master/iso-19139-to-dcat-ap.xsl#L837-L840
            ProvActivity.builder().used(new RdfResource(null, resourceIdentifierUri, null)).qualifiedAssociation(ProvQualifiedAssociation.builder().hadPlan(ProvHadPlan.builder().wasDerivedFrom(new RdfResource("Resource", null, null, c.getTitle(), null)).build()).build()).generated(ProvGenerated.builder().type(new RdfResource("http://inspire.ec.europa.eu/metadata-codelist/DegreeOfConformity/" + INSPIRE_DEGREE_OF_CONFORMITY.get(c.getPass()), null)).build()).build()).build()).collect(Collectors.toList()));
        }
        if (record.getResourceLanguage() != null) {
            // TODO: Where to put resource language ?
            datasetBuilder.language(record.getResourceLanguage().stream().map(l -> new RdfResource(null, "http://publications.europa.eu/resource/authority/language/" + l.toUpperCase(), null)).collect(Collectors.toList()));
        }
        ArrayList<Codelist> updateFrequencyList = record.getCodelists().get(Codelists.maintenanceAndUpdateFrequency);
        if (updateFrequencyList != null && updateFrequencyList.size() > 0) {
            datasetBuilder.accrualPeriodicity(new RdfResource(null, ACCRUAL_PERIODICITY_URI_PREFIX + ACCRUAL_PERIODICITY_TO_ISO.get(updateFrequencyList.get(0).getProperties().get(CommonField.key)), null));
        }
        // <dct:spatial rdf:parseType="Resource">
        datasetBuilder.spatial(record.getGeometries().stream().map(g -> DctSpatial.builder().location(DctLocation.builder().geometry(g).build()).build()).collect(Collectors.toList()));
        datasetBuilder.temporal(record.getResourceTemporalExtentDateRange().stream().map(range -> {
            DctPeriodOfTimeBuilder periodOfTime = DctPeriodOfTime.builder();
            if (StringUtils.isNotEmpty(range.getGte())) {
                periodOfTime.startDate(toDate(range.getGte()));
            }
            if (StringUtils.isNotEmpty(range.getLte())) {
                periodOfTime.endDate(toDate(range.getLte()));
            }
            return DctTemporal.builder().periodOfTime(periodOfTime.build()).build();
        }).collect(Collectors.toList()));
        record.getLinks().stream().forEach(link -> {
            DcatDistributionBuilder dcatDistributionBuilder = DcatDistribution.builder().title(listOfNullable(link.getName())).description(listOfNullable(link.getDescription())).representationTechnique(Subject.builder().skosConcept(SkosConcept.builder().prefLabel(link.getProtocol()).build()).build());
            // TODO: depending on function/protocol build page/accessUrl/downloadUrl
            dcatDistributionBuilder.accessUrl(link.getUrl());
            datasetBuilder.distribution(listOfNullable(DcatDistributionContainer.builder().distribution(dcatDistributionBuilder.build()).build()));
        });
        datasetBuilder.contactPoint(record.getContactForResource().stream().map(contact -> DcatContactPoint.builder().contact(VcardContact.builder().title(contact.getOrganisation()).role(contact.getRole()).hasEmail(contact.getEmail()).build()).build()).collect(Collectors.toList()));
        dcatDataset = datasetBuilder.build();
        catalogRecord = CatalogRecord.builder().identifier(listOfNullable(record.getMetadataIdentifier())).created(toDate(record.getCreateDate())).modified(toDate(record.getChangeDate())).language(listOfNullable(new RdfResource(null, "http://publications.europa.eu/resource/authority/language/" + record.getMainLanguage().toUpperCase()))).primaryTopic(listOfNullable(new ResourceContainer(dcatDataset, null))).build();
    } catch (JsonMappingException e) {
        e.printStackTrace();
    } catch (JsonProcessingException e) {
        e.printStackTrace();
    }
    return catalogRecord;
}
Also used : StringUtils(org.apache.commons.lang.StringUtils) DatasetBuilder(org.fao.geonet.index.model.dcat2.Dataset.DatasetBuilder) DcatDistributionBuilder(org.fao.geonet.index.model.dcat2.DcatDistribution.DcatDistributionBuilder) Date(java.util.Date) Autowired(org.springframework.beans.factory.annotation.Autowired) DeserializationFeature(com.fasterxml.jackson.databind.DeserializationFeature) ProvHadPlan(org.fao.geonet.index.model.dcat2.ProvHadPlan) IndexRecord(org.fao.geonet.index.model.gn.IndexRecord) BigDecimal(java.math.BigDecimal) Codelists(org.fao.geonet.index.model.gn.IndexRecordFieldNames.Codelists) DcatDistribution(org.fao.geonet.index.model.dcat2.DcatDistribution) Map(java.util.Map) JsonNode(com.fasterxml.jackson.databind.JsonNode) ACCRUAL_PERIODICITY_TO_ISO(org.fao.geonet.index.model.dcat2.Dataset.ACCRUAL_PERIODICITY_TO_ISO) Dataset(org.fao.geonet.index.model.dcat2.Dataset) RdfResource(org.fao.geonet.index.model.dcat2.RdfResource) DcatActivity(org.fao.geonet.index.model.dcat2.DcatActivity) Instant(java.time.Instant) DcatDocument(org.fao.geonet.index.model.dcat2.DcatDocument) Collectors(java.util.stream.Collectors) CatalogRecord(org.fao.geonet.index.model.dcat2.CatalogRecord) ACCRUAL_PERIODICITY_URI_PREFIX(org.fao.geonet.index.model.dcat2.Dataset.ACCRUAL_PERIODICITY_URI_PREFIX) Objects(java.util.Objects) List(java.util.List) CommonField.defaultText(org.fao.geonet.index.model.gn.IndexRecordFieldNames.CommonField.defaultText) DctTemporal(org.fao.geonet.index.model.dcat2.DctTemporal) Optional(java.util.Optional) JsonMappingException(com.fasterxml.jackson.databind.JsonMappingException) DctPeriodOfTime(org.fao.geonet.index.model.dcat2.DctPeriodOfTime) FoafDocument(org.fao.geonet.index.model.dcat2.FoafDocument) ProvGenerated(org.fao.geonet.index.model.dcat2.ProvGenerated) DctPeriodOfTimeBuilder(org.fao.geonet.index.model.dcat2.DctPeriodOfTime.DctPeriodOfTimeBuilder) CommonField(org.fao.geonet.index.model.gn.IndexRecordFieldNames.CommonField) ArrayList(java.util.ArrayList) DctLocation(org.fao.geonet.index.model.dcat2.DctLocation) Value(org.springframework.beans.factory.annotation.Value) ProvenanceStatement(org.fao.geonet.index.model.dcat2.ProvenanceStatement) DctSpatial(org.fao.geonet.index.model.dcat2.DctSpatial) Codelists.topic(org.fao.geonet.index.model.gn.IndexRecordFieldNames.Codelists.topic) ResourceContainer(org.fao.geonet.index.model.dcat2.ResourceContainer) Codelist(org.fao.geonet.index.model.gn.Codelist) IndexRecordFieldNames(org.fao.geonet.index.model.gn.IndexRecordFieldNames) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) ProvActivity(org.fao.geonet.index.model.dcat2.ProvActivity) JsonProcessingException(com.fasterxml.jackson.core.JsonProcessingException) ProvQualifiedAssociation(org.fao.geonet.index.model.dcat2.ProvQualifiedAssociation) SkosConcept(org.fao.geonet.index.model.dcat2.SkosConcept) DcatContactPoint(org.fao.geonet.index.model.dcat2.DcatContactPoint) Provenance(org.fao.geonet.index.model.dcat2.Provenance) ResourceIdentifier(org.fao.geonet.index.model.gn.ResourceIdentifier) AbstractMap(java.util.AbstractMap) Component(org.springframework.stereotype.Component) DateTimeFormatter(java.time.format.DateTimeFormatter) Subject(org.fao.geonet.index.model.dcat2.Subject) Collections(java.util.Collections) DcatDistributionContainer(org.fao.geonet.index.model.dcat2.DcatDistributionContainer) VcardContact(org.fao.geonet.index.model.dcat2.VcardContact) Codelist(org.fao.geonet.index.model.gn.Codelist) DctPeriodOfTimeBuilder(org.fao.geonet.index.model.dcat2.DctPeriodOfTime.DctPeriodOfTimeBuilder) Dataset(org.fao.geonet.index.model.dcat2.Dataset) IndexRecord(org.fao.geonet.index.model.gn.IndexRecord) BigDecimal(java.math.BigDecimal) ResourceIdentifier(org.fao.geonet.index.model.gn.ResourceIdentifier) CatalogRecord(org.fao.geonet.index.model.dcat2.CatalogRecord) JsonMappingException(com.fasterxml.jackson.databind.JsonMappingException) DcatDistributionBuilder(org.fao.geonet.index.model.dcat2.DcatDistribution.DcatDistributionBuilder) RdfResource(org.fao.geonet.index.model.dcat2.RdfResource) DatasetBuilder(org.fao.geonet.index.model.dcat2.Dataset.DatasetBuilder) JsonProcessingException(com.fasterxml.jackson.core.JsonProcessingException) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) ResourceContainer(org.fao.geonet.index.model.dcat2.ResourceContainer)

Example 67 with Dataset

use of com.google.cloud.datalabeling.v1beta1.Dataset in project geonetwork-microservices by geonetwork.

the class DcatTest method testJsonToDcat.

@Test
public void testJsonToDcat() throws IOException {
    String identifier = "1567-765175-6561756";
    Dataset dcatDataset = Dataset.builder().title(List.of("Rivers on earth")).description(List.of("Water drop ...")).identifier(List.of(identifier)).subject(List.of(Subject.builder().skosConcept(SkosConcept.builder().about("https://registry.org/hydrology").prefLabel("Hydrology").build()).build(), Subject.builder().skosConcept(SkosConcept.builder().prefLabel("Earth observation").build()).build())).landingPage(List.of(DcatDocument.builder().foafDocument(FoafDocument.builder().about("https://data/file.zip").title("Download dataset").description("CC0").format("Shapefile").build()).build())).qualifiedRelation(List.of(DcatRelationship.builder().relation(DcatRelation.builder().relation("http://...TODO").build()).build())).spatialResolutionInMeters(List.of(new BigDecimal(25000))).temporalResolution(List.of(Duration.ofDays(15))).accessRights(List.of(DcatAccessRights.builder().rightsStatement(RightsStatement.builder().label("public access limited according to Article 13(1)(b) of the INSPIRE Directive").about("https://registry.inspire/...").build()).build())).accrualPeriodicity(new RdfResource(null, ACCRUAL_PERIODICITY_URI_PREFIX + ACCRUAL_PERIODICITY_TO_ISO.get("daily"), null)).conformsTo(new RdfResource(null, "http://iso19115-3.schema.org", null)).created(new Date()).isReferencedBy(List.of(new RdfResource(null, "https://isReferencedBy", null))).relation(List.of(new RdfResource(null, "https://relation", null))).language(List.of(new RdfResource(null, "http://publications.europa.eu/resource/authority/language/FRE", null))).type(List.of(Subject.builder().skosConcept(SkosConcept.builder().prefLabel("dataset").build()).build())).page(List.of(DcatDocument.builder().foafDocument(FoafDocument.builder().about("https://apps.titellus.net/ogcapi/collections/main/items/" + identifier).build()).build())).versionInfo("1.0").qualifiedAttribution(List.of(DcatQualifiedAttribution.builder().attribution(ProvAttribution.builder().agent(new RdfResource(null, "http://agent", null)).hadRole(new RdfResource(null, "http://role/creator", null)).build()).build())).comment(List.of("Comments ...")).distribution(List.of(DcatDistributionContainer.builder().distribution(DcatDistribution.builder().accessUrl("https://sdi.eea.europa.eu/webdav/continental/europe/natural_areas/birds_directive/eea_v_3035_10_mio_art12-2008-2012_i_2008-2012_v01_r01/Art12-2008-2012_SHP").build()).build())).license(DcatLicenseDocumentContainer.builder().license(DcatLicenseDocument.builder().type(Subject.builder().skosConcept(SkosConcept.builder().prefLabel("CC0").build()).build()).build()).build()).build();
    // POJO to XML
    JAXBContext context = null;
    String dcatXml = null;
    try {
        StringWriter sw = new StringWriter();
        context = JAXBContext.newInstance(SkosConcept.class, Dataset.class);
        Marshaller marshaller = context.createMarshaller();
        marshaller.setProperty(Marshaller.JAXB_FRAGMENT, Boolean.TRUE);
        marshaller.setProperty(Marshaller.JAXB_FORMATTED_OUTPUT, Boolean.TRUE);
        marshaller.marshal(dcatDataset, sw);
        dcatXml = sw.toString();
    } catch (JAXBException e) {
        e.printStackTrace();
    }
    org.eclipse.rdf4j.model.Model model = Rio.parse(new ByteArrayInputStream(dcatXml.getBytes()), "", RDFFormat.RDFXML);
    // Rio.write(model, System.out, RDFFormat.RDFXML);
    Rio.write(model, System.out, RDFFormat.TURTLE);
// Rio.write(model, System.out, RDFFormat.JSONLD);
}
Also used : Marshaller(javax.xml.bind.Marshaller) Dataset(org.fao.geonet.index.model.dcat2.Dataset) JAXBException(javax.xml.bind.JAXBException) JAXBContext(javax.xml.bind.JAXBContext) BigDecimal(java.math.BigDecimal) Date(java.util.Date) StringWriter(java.io.StringWriter) ByteArrayInputStream(java.io.ByteArrayInputStream) RdfResource(org.fao.geonet.index.model.dcat2.RdfResource) SkosConcept(org.fao.geonet.index.model.dcat2.SkosConcept) Test(org.junit.Test)

Aggregations

IOException (java.io.IOException)20 DataLabelingServiceClient (com.google.cloud.datalabeling.v1beta1.DataLabelingServiceClient)13 ByteArrayOutputStream (java.io.ByteArrayOutputStream)10 PrintStream (java.io.PrintStream)10 Before (org.junit.Before)10 Dataset (com.google.cloud.datalabeling.v1beta1.Dataset)9 ArrayList (java.util.ArrayList)9 Test (org.junit.Test)9 AutoMlClient (com.google.cloud.automl.v1.AutoMlClient)8 Dataset (com.google.cloud.automl.v1.Dataset)8 ProjectName (com.google.cloud.datalabeling.v1beta1.ProjectName)8 LocationName (com.google.cloud.automl.v1.LocationName)7 ListDatasetsPagedResponse (com.google.cloud.datalabeling.v1beta1.DataLabelingServiceClient.ListDatasetsPagedResponse)7 AbstractMessage (com.google.protobuf.AbstractMessage)7 CreateDatasetOperationMetadata (com.google.cloud.aiplatform.v1.CreateDatasetOperationMetadata)6 Dataset (com.google.cloud.aiplatform.v1.Dataset)6 DatasetServiceClient (com.google.cloud.aiplatform.v1.DatasetServiceClient)6 DatasetServiceSettings (com.google.cloud.aiplatform.v1.DatasetServiceSettings)6 LocationName (com.google.cloud.aiplatform.v1.LocationName)6 OperationMetadata (com.google.cloud.automl.v1.OperationMetadata)6