Search in sources :

Example 1 with JexlRowFactory

use of nl.knaw.huygens.timbuctoo.rml.datasource.jexl.JexlRowFactory in project timbuctoo by HuygensING.

the class BdbRmlDataSourceStoreTest method itWorks.

@Test
public void itWorks() throws Exception {
    BdbNonPersistentEnvironmentCreator dbCreator = new BdbNonPersistentEnvironmentCreator();
    DataSetMetaData dataSetMetadata = new BasicDataSetMetaData("userid", "datasetid", "http://timbuctoo.huygens.knaw.nl/v5/userid/datasetid", "http://example.org/prefix/", false, false);
    final RmlDataSourceStore rmlDataSourceStore = new BdbRmlDataSourceStore(dbCreator.getDatabase("userid", "datasetid", "rmlSource", true, TupleBinding.getPrimitiveBinding(String.class), TupleBinding.getPrimitiveBinding(String.class), new StringStringIsCleanHandler()), new ImportStatus(new LogList()));
    RdfSerializer rdfSerializer = new RmlDataSourceRdfSerializer(rmlDataSourceStore);
    RawUploadRdfSaver rawUploadRdfSaver = new RawUploadRdfSaver(dataSetMetadata, "fileName", APPLICATION_OCTET_STREAM_TYPE, rdfSerializer, "origFileName", Clock.systemUTC());
    final String inputCol1 = rawUploadRdfSaver.addCollection("collection1");
    ImportPropertyDescriptions importPropertyDescriptions = new ImportPropertyDescriptions();
    importPropertyDescriptions.getOrCreate(1).setPropertyName("propName1");
    importPropertyDescriptions.getOrCreate(2).setPropertyName("propName2");
    rawUploadRdfSaver.addPropertyDescriptions(inputCol1, importPropertyDescriptions);
    rawUploadRdfSaver.addEntity(inputCol1, ImmutableMap.of("propName1", "value1", "propName2", "val2"));
    rawUploadRdfSaver.addEntity(inputCol1, ImmutableMap.of("propName1", "entVal1", "propName2", "entVal2"));
    final String inputCol2 = rawUploadRdfSaver.addCollection("collection2");
    ImportPropertyDescriptions importPropertyDescriptions1 = new ImportPropertyDescriptions();
    importPropertyDescriptions1.getOrCreate(1).setPropertyName("prop3");
    importPropertyDescriptions1.getOrCreate(2).setPropertyName("prop4");
    rawUploadRdfSaver.addPropertyDescriptions(inputCol2, importPropertyDescriptions1);
    rawUploadRdfSaver.addEntity(inputCol2, ImmutableMap.of("prop3", "value1", "prop4", "val2"));
    rawUploadRdfSaver.addEntity(inputCol2, ImmutableMap.of("prop3", "entVal1", "prop4", "entVal2"));
    rdfSerializer.close();
    RdfDataSource rdfDataSource = new RdfDataSource(rmlDataSourceStore, inputCol1, new JexlRowFactory(ImmutableMap.of(), new HashMapBasedJoinHandler()));
    RdfDataSource rdfDataSource2 = new RdfDataSource(rmlDataSourceStore, inputCol2, new JexlRowFactory(ImmutableMap.of(), new HashMapBasedJoinHandler()));
    final List<String> collection1;
    final List<String> collection2;
    try (Stream<Row> stream = rdfDataSource.getRows(new ThrowingErrorHandler())) {
        collection1 = stream.map(x -> x.getRawValue("propName1") + ":" + x.getRawValue("propName2")).collect(toList());
    }
    try (Stream<Row> stream = rdfDataSource2.getRows(new ThrowingErrorHandler())) {
        collection2 = stream.map(x -> x.getRawValue("prop3") + ":" + x.getRawValue("prop4")).collect(toList());
    }
    assertThat(collection1, contains("value1:val2", "entVal1:entVal2"));
    assertThat(collection2, contains("value1:val2", "entVal1:entVal2"));
    dbCreator.close();
}
Also used : RawUploadRdfSaver(nl.knaw.huygens.timbuctoo.v5.bulkupload.RawUploadRdfSaver) ImportPropertyDescriptions(nl.knaw.huygens.timbuctoo.bulkupload.parsingstatemachine.ImportPropertyDescriptions) StringStringIsCleanHandler(nl.knaw.huygens.timbuctoo.v5.berkeleydb.isclean.StringStringIsCleanHandler) RmlDataSourceStore(nl.knaw.huygens.timbuctoo.v5.datastores.rmldatasource.RmlDataSourceStore) BdbNonPersistentEnvironmentCreator(nl.knaw.huygens.timbuctoo.v5.dropwizard.BdbNonPersistentEnvironmentCreator) RdfSerializer(nl.knaw.huygens.timbuctoo.v5.rdfio.RdfSerializer) JexlRowFactory(nl.knaw.huygens.timbuctoo.rml.datasource.jexl.JexlRowFactory) LogList(nl.knaw.huygens.timbuctoo.v5.dataset.dto.LogList) ImportStatus(nl.knaw.huygens.timbuctoo.v5.dataset.ImportStatus) HashMapBasedJoinHandler(nl.knaw.huygens.timbuctoo.rml.datasource.joinhandlers.HashMapBasedJoinHandler) BasicDataSetMetaData(nl.knaw.huygens.timbuctoo.v5.dataset.dto.BasicDataSetMetaData) ThrowingErrorHandler(nl.knaw.huygens.timbuctoo.rml.ThrowingErrorHandler) Row(nl.knaw.huygens.timbuctoo.rml.Row) DataSetMetaData(nl.knaw.huygens.timbuctoo.v5.dataset.dto.DataSetMetaData) BasicDataSetMetaData(nl.knaw.huygens.timbuctoo.v5.dataset.dto.BasicDataSetMetaData) RdfDataSource(nl.knaw.huygens.timbuctoo.v5.rml.RdfDataSource) Test(org.junit.Test)

Example 2 with JexlRowFactory

use of nl.knaw.huygens.timbuctoo.rml.datasource.jexl.JexlRowFactory in project timbuctoo by HuygensING.

the class RdfDataSourceFactory method apply.

public Optional<DataSource> apply(RdfResource rdfResource) {
    for (RdfResource resource : rdfResource.out(NS_RML + "source")) {
        Set<RdfResource> rawCollection = resource.out("http://timbuctoo.huygens.knaw.nl/mapping#rawCollectionUri");
        Set<RdfResource> customFields = resource.out("http://timbuctoo.huygens.knaw.nl/mapping#customField");
        Map<String, String> expressions = new HashMap<>();
        for (RdfResource customField : customFields) {
            Set<RdfResource> fieldNameResource = customField.out("http://timbuctoo.huygens.knaw.nl/mapping#name");
            Set<RdfResource> fieldValueResource = customField.out("http://timbuctoo.huygens.knaw.nl/mapping#expression");
            fieldNameResource.iterator().next().asLiteral().ifPresent(fieldName -> {
                fieldValueResource.iterator().next().asLiteral().ifPresent(fieldValue -> {
                    expressions.put(fieldName.getValue(), fieldValue.getValue());
                });
            });
        }
        if (rawCollection.size() == 1) {
            return rawCollection.iterator().next().asIri().map(collectionIri -> new RdfDataSource(rmlDataSourceStore, collectionIri, new JexlRowFactory(expressions, new HashMapBasedJoinHandler())));
        }
    }
    return Optional.empty();
}
Also used : HashMap(java.util.HashMap) HashMapBasedJoinHandler(nl.knaw.huygens.timbuctoo.rml.datasource.joinhandlers.HashMapBasedJoinHandler) RdfResource(nl.knaw.huygens.timbuctoo.rml.rdfshim.RdfResource) JexlRowFactory(nl.knaw.huygens.timbuctoo.rml.datasource.jexl.JexlRowFactory)

Aggregations

JexlRowFactory (nl.knaw.huygens.timbuctoo.rml.datasource.jexl.JexlRowFactory)2 HashMapBasedJoinHandler (nl.knaw.huygens.timbuctoo.rml.datasource.joinhandlers.HashMapBasedJoinHandler)2 HashMap (java.util.HashMap)1 ImportPropertyDescriptions (nl.knaw.huygens.timbuctoo.bulkupload.parsingstatemachine.ImportPropertyDescriptions)1 Row (nl.knaw.huygens.timbuctoo.rml.Row)1 ThrowingErrorHandler (nl.knaw.huygens.timbuctoo.rml.ThrowingErrorHandler)1 RdfResource (nl.knaw.huygens.timbuctoo.rml.rdfshim.RdfResource)1 StringStringIsCleanHandler (nl.knaw.huygens.timbuctoo.v5.berkeleydb.isclean.StringStringIsCleanHandler)1 RawUploadRdfSaver (nl.knaw.huygens.timbuctoo.v5.bulkupload.RawUploadRdfSaver)1 ImportStatus (nl.knaw.huygens.timbuctoo.v5.dataset.ImportStatus)1 BasicDataSetMetaData (nl.knaw.huygens.timbuctoo.v5.dataset.dto.BasicDataSetMetaData)1 DataSetMetaData (nl.knaw.huygens.timbuctoo.v5.dataset.dto.DataSetMetaData)1 LogList (nl.knaw.huygens.timbuctoo.v5.dataset.dto.LogList)1 RmlDataSourceStore (nl.knaw.huygens.timbuctoo.v5.datastores.rmldatasource.RmlDataSourceStore)1 BdbNonPersistentEnvironmentCreator (nl.knaw.huygens.timbuctoo.v5.dropwizard.BdbNonPersistentEnvironmentCreator)1 RdfSerializer (nl.knaw.huygens.timbuctoo.v5.rdfio.RdfSerializer)1 RdfDataSource (nl.knaw.huygens.timbuctoo.v5.rml.RdfDataSource)1 Test (org.junit.Test)1