use of nl.knaw.huygens.timbuctoo.v5.rml.RdfDataSourceFactory in project timbuctoo by HuygensING.
the class RmlRdfCreator method sendQuads.
@Override
public void sendQuads(RdfSerializer saver, DataSet dataSet, Consumer<String> status) throws LogStorageFailedException {
RdfDataSourceFactory dataSourceFactory = dataSet.getDataSource();
final Model model = ModelFactory.createDefaultModel();
try {
model.read(new ByteArrayInputStream(rdfData.getBytes(StandardCharsets.UTF_8)), null, "JSON-LD");
} catch (Exception e) {
throw new LogStorageFailedException(e);
}
final RmlMappingDocument rmlMappingDocument = rmlBuilder.fromRdf(model, dataSourceFactory::apply);
if (rmlMappingDocument.getErrors().size() > 0) {
throw new LogStorageFailedException("failure: " + String.join("\nfailure: ", rmlMappingDocument.getErrors()) + "\n");
}
// FIXME: trigger onprefix for all rml prefixes
// FIXME: store rml and retrieve it from tripleStore when mapping
Stream<Quad> triples = rmlMappingDocument.execute(new ReportingErrorHandler(status));
Iterator<Quad> iterator = triples.iterator();
while (iterator.hasNext()) {
Quad triple = iterator.next();
saver.onQuad(triple.getSubject().getUri().get(), triple.getPredicate().getUri().get(), triple.getObject().getContent(), triple.getObject().getLiteralType().orElse(null), triple.getObject().getLiteralLanguage().orElse(null), baseUri);
}
}
use of nl.knaw.huygens.timbuctoo.v5.rml.RdfDataSourceFactory in project timbuctoo by HuygensING.
the class DataSet method dataSet.
public static DataSet dataSet(DataSetMetaData metadata, ExecutorService executorService, String rdfPrefix, BdbEnvironmentCreator dataStoreFactory, Runnable onUpdated, DataSetStorage dataSetStorage) throws IOException, DataStoreCreationException {
String userId = metadata.getOwnerId();
String dataSetId = metadata.getDataSetId();
File descriptionFile = dataSetStorage.getResourceSyncDescriptionFile();
FileStorage fileStorage = dataSetStorage.getFileStorage();
ImportManager importManager = new ImportManager(dataSetStorage.getLogList(), fileStorage, fileStorage, dataSetStorage.getLogStorage(), executorService, dataSetStorage.getRdfIo(), onUpdated);
try {
importManager.subscribeToRdf(new RdfDescriptionSaver(descriptionFile, metadata.getBaseUri(), importManager.getImportStatus()));
} catch (ParserConfigurationException | SAXException e) {
LOG.error("Could not construct import manager of data set", e);
}
final TupleBinding<String> stringBinding = TupleBinding.getPrimitiveBinding(String.class);
try {
StringStringIsCleanHandler stringStringIsCleanHandler = new StringStringIsCleanHandler();
BdbTripleStore quadStore = new BdbTripleStore(dataStoreFactory.getDatabase(userId, dataSetId, "rdfData", true, stringBinding, stringBinding, stringStringIsCleanHandler));
final BdbTypeNameStore typeNameStore = new BdbTypeNameStore(new BdbBackedData(dataStoreFactory.getDatabase(userId, dataSetId, "typenames", false, stringBinding, stringBinding, stringStringIsCleanHandler)), rdfPrefix);
final BdbSchemaStore schema = new BdbSchemaStore(new BdbBackedData(dataStoreFactory.getDatabase(userId, dataSetId, "schema", false, stringBinding, stringBinding, stringStringIsCleanHandler)), importManager.getImportStatus());
final BdbTruePatchStore truePatchStore = new BdbTruePatchStore(dataStoreFactory.getDatabase(userId, dataSetId, "truePatch", true, stringBinding, stringBinding, stringStringIsCleanHandler));
final TupleBinding<Integer> integerBinding = TupleBinding.getPrimitiveBinding(Integer.class);
final UpdatedPerPatchStore updatedPerPatchStore = new UpdatedPerPatchStore(dataStoreFactory.getDatabase(userId, dataSetId, "updatedPerPatch", true, integerBinding, stringBinding, new IsCleanHandler<Integer, String>() {
@Override
public Integer getKey() {
return Integer.MAX_VALUE;
}
@Override
public String getValue() {
return "isClean";
}
}));
final BdbRmlDataSourceStore rmlDataSourceStore = new BdbRmlDataSourceStore(dataStoreFactory.getDatabase(userId, dataSetId, "rmlSource", true, stringBinding, stringBinding, stringStringIsCleanHandler), importManager.getImportStatus());
VersionStore versionStore = new VersionStore(dataStoreFactory.getDatabase(userId, dataSetId, "versions", false, stringBinding, integerBinding, new IsCleanHandler<String, Integer>() {
@Override
public String getKey() {
return "isClean";
}
@Override
public Integer getValue() {
return Integer.MAX_VALUE;
}
}));
final StoreUpdater storeUpdater = new StoreUpdater(dataStoreFactory, quadStore, typeNameStore, truePatchStore, updatedPerPatchStore, Lists.newArrayList(schema, rmlDataSourceStore), versionStore, importManager.getImportStatus());
importManager.subscribeToRdf(storeUpdater);
ImmutableDataSet dataSet = ImmutableDataSet.builder().ownerId(userId).dataSetName(dataSetId).bdbEnvironmentCreator(dataStoreFactory).metadata(metadata).quadStore(quadStore).typeNameStore(typeNameStore).schemaStore(schema).dataSource(new RdfDataSourceFactory(rmlDataSourceStore)).schemaStore(schema).importManager(importManager).dataSetStorage(dataSetStorage).build();
importManager.init(dataSet);
if (!quadStore.isClean() || !typeNameStore.isClean() || !schema.isClean() || !truePatchStore.isClean() || !updatedPerPatchStore.isClean() || !rmlDataSourceStore.isClean() || !versionStore.isClean()) {
LOG.error("Data set '{}__{}' data is corrupted, starting to reimport.", userId, dataSetId);
quadStore.empty();
typeNameStore.empty();
schema.empty();
truePatchStore.empty();
updatedPerPatchStore.empty();
rmlDataSourceStore.empty();
versionStore.empty();
importManager.reprocessLogs();
} else {
// process unprocessed logs
importManager.processLogs();
}
return dataSet;
} catch (BdbDbCreationException e) {
throw new DataStoreCreationException(e.getCause());
}
}
Aggregations