Examples with SparkContext org.apache.spark.SparkContext used on opensource projects

Example 1 with SparkContext

use of org.apache.spark.SparkContext in project zeppelin by apache.

the class SparkSqlInterpreter method cancel.

@Override
public void cancel(InterpreterContext context) {
    SparkInterpreter sparkInterpreter = getSparkInterpreter();
    SQLContext sqlc = sparkInterpreter.getSQLContext();
    SparkContext sc = sqlc.sparkContext();
    sc.cancelJobGroup(Utils.buildJobGroupId(context));
}

Example 2 with SparkContext

use of org.apache.spark.SparkContext in project zeppelin by apache.

the class SparkInterpreterTest method testListener.

@Test
public void testListener() {
    SparkContext sc = repl.getSparkContext();
    assertNotNull(SparkInterpreter.setupListeners(sc));
}

Example 3 with SparkContext

use of org.apache.spark.SparkContext in project Gaffer by gchq.

the class GetRDDOfElementsHandlerTest method checkGetCorrectElementsInRDDForEntitySeed.

@Test
public void checkGetCorrectElementsInRDDForEntitySeed() throws OperationException, IOException {
    final Graph graph1 = new Graph.Builder().addSchema(getClass().getResourceAsStream("/schema/dataSchema.json")).addSchema(getClass().getResourceAsStream("/schema/dataTypes.json")).addSchema(getClass().getResourceAsStream("/schema/storeTypes.json")).storeProperties(getClass().getResourceAsStream("/store.properties")).build();
    final List<Element> elements = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        final Entity entity = new Entity(ENTITY_GROUP);
        entity.setVertex("" + i);
        final Edge edge1 = new Edge(EDGE_GROUP);
        edge1.setSource("" + i);
        edge1.setDestination("B");
        edge1.setDirected(false);
        edge1.putProperty("count", 2);
        final Edge edge2 = new Edge(EDGE_GROUP);
        edge2.setSource("" + i);
        edge2.setDestination("C");
        edge2.setDirected(false);
        edge2.putProperty("count", 4);
        elements.add(edge1);
        elements.add(edge2);
        elements.add(entity);
    }
    final User user = new User();
    graph1.execute(new AddElements(elements), user);
    final SparkConf sparkConf = new SparkConf().setMaster("local").setAppName("testCheckGetCorrectElementsInRDDForEntitySeed").set("spark.serializer", "org.apache.spark.serializer.KryoSerializer").set("spark.kryo.registrator", "uk.gov.gchq.gaffer.spark.serialisation.kryo.Registrator").set("spark.driver.allowMultipleContexts", "true");
    final SparkContext sparkContext = new SparkContext(sparkConf);
    // Create Hadoop configuration and serialise to a string
    final Configuration configuration = new Configuration();
    final ByteArrayOutputStream baos = new ByteArrayOutputStream();
    configuration.write(new DataOutputStream(baos));
    final String configurationString = new String(baos.toByteArray(), CommonConstants.UTF_8);
    // Check get correct edges for "1"
    GetRDDOfElements<EntitySeed> rddQuery = new GetRDDOfElements.Builder<EntitySeed>().sparkContext(sparkContext).seeds(Collections.singleton(new EntitySeed("1"))).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    RDD<Element> rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    Set<Element> results = new HashSet<>();
    // NB: IDE suggests the cast in the following line is unnecessary but compilation fails without it
    Element[] returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    final Set<Element> expectedElements = new HashSet<>();
    final Entity entity1 = new Entity(ENTITY_GROUP);
    entity1.setVertex("1");
    final Edge edge1B = new Edge(EDGE_GROUP);
    edge1B.setSource("1");
    edge1B.setDestination("B");
    edge1B.setDirected(false);
    edge1B.putProperty("count", 2);
    final Edge edge1C = new Edge(EDGE_GROUP);
    edge1C.setSource("1");
    edge1C.setDestination("C");
    edge1C.setDirected(false);
    edge1C.putProperty("count", 4);
    expectedElements.add(entity1);
    expectedElements.add(edge1B);
    expectedElements.add(edge1C);
    assertEquals(expectedElements, results);
    // Check get correct edges for "1" when specify entities only
    rddQuery = new GetRDDOfElements.Builder<EntitySeed>().sparkContext(sparkContext).seeds(Collections.singleton(new EntitySeed("1"))).view(new View.Builder().entity(ENTITY_GROUP).build()).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    results.clear();
    returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    expectedElements.clear();
    expectedElements.add(entity1);
    assertEquals(expectedElements, results);
    // Check get correct edges for "1" when specify edges only
    rddQuery = new GetRDDOfElements.Builder<EntitySeed>().sparkContext(sparkContext).seeds(Collections.singleton(new EntitySeed("1"))).view(new View.Builder().edge(EDGE_GROUP).build()).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    results.clear();
    returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    expectedElements.clear();
    expectedElements.add(edge1B);
    expectedElements.add(edge1C);
    assertEquals(expectedElements, results);
    // Check get correct edges for "1" and "5"
    Set<EntitySeed> seeds = new HashSet<>();
    seeds.add(new EntitySeed("1"));
    seeds.add(new EntitySeed("5"));
    rddQuery = new GetRDDOfElements.Builder<EntitySeed>().sparkContext(sparkContext).seeds(seeds).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    results.clear();
    returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    final Entity entity5 = new Entity(ENTITY_GROUP);
    entity5.setVertex("5");
    final Edge edge5B = new Edge(EDGE_GROUP);
    edge5B.setSource("5");
    edge5B.setDestination("B");
    edge5B.setDirected(false);
    edge5B.putProperty("count", 2);
    final Edge edge5C = new Edge(EDGE_GROUP);
    edge5C.setSource("5");
    edge5C.setDestination("C");
    edge5C.setDirected(false);
    edge5C.putProperty("count", 4);
    expectedElements.clear();
    expectedElements.add(entity1);
    expectedElements.add(edge1B);
    expectedElements.add(edge1C);
    expectedElements.add(entity5);
    expectedElements.add(edge5B);
    expectedElements.add(edge5C);
    assertEquals(expectedElements, results);
    sparkContext.stop();
}

Example 4 with SparkContext

use of org.apache.spark.SparkContext in project Gaffer by gchq.

the class GetRDDOfElementsHandlerTest method checkGetCorrectElementsInRDDForEdgeSeed.

@Test
public void checkGetCorrectElementsInRDDForEdgeSeed() throws OperationException, IOException {
    final Graph graph1 = new Graph.Builder().addSchema(getClass().getResourceAsStream("/schema/dataSchema.json")).addSchema(getClass().getResourceAsStream("/schema/dataTypes.json")).addSchema(getClass().getResourceAsStream("/schema/storeTypes.json")).storeProperties(getClass().getResourceAsStream("/store.properties")).build();
    final List<Element> elements = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        final Entity entity = new Entity(ENTITY_GROUP);
        entity.setVertex("" + i);
        final Edge edge1 = new Edge(EDGE_GROUP);
        edge1.setSource("" + i);
        edge1.setDestination("B");
        edge1.setDirected(false);
        edge1.putProperty("count", 2);
        final Edge edge2 = new Edge(EDGE_GROUP);
        edge2.setSource("" + i);
        edge2.setDestination("C");
        edge2.setDirected(false);
        edge2.putProperty("count", 4);
        elements.add(edge1);
        elements.add(edge2);
        elements.add(entity);
    }
    final User user = new User();
    graph1.execute(new AddElements(elements), user);
    final SparkConf sparkConf = new SparkConf().setMaster("local").setAppName("testCheckGetCorrectElementsInRDDForEdgeSeed").set("spark.serializer", "org.apache.spark.serializer.KryoSerializer").set("spark.kryo.registrator", "uk.gov.gchq.gaffer.spark.serialisation.kryo.Registrator").set("spark.driver.allowMultipleContexts", "true");
    final SparkContext sparkContext = new SparkContext(sparkConf);
    // Create Hadoop configuration and serialise to a string
    final Configuration configuration = new Configuration();
    final ByteArrayOutputStream baos = new ByteArrayOutputStream();
    configuration.write(new DataOutputStream(baos));
    final String configurationString = new String(baos.toByteArray(), CommonConstants.UTF_8);
    // Check get correct edges for EdgeSeed 1 -> B
    GetRDDOfElements<EdgeSeed> rddQuery = new GetRDDOfElements.Builder<EdgeSeed>().sparkContext(sparkContext).seeds(Collections.singleton(new EdgeSeed("1", "B", false))).includeEdges(GetOperation.IncludeEdgeType.ALL).includeEntities(false).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    RDD<Element> rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    Set<Element> results = new HashSet<>();
    // NB: IDE suggests the cast in the following line is unnecessary but compilation fails without it
    Element[] returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    final Set<Element> expectedElements = new HashSet<>();
    final Edge edge1B = new Edge(EDGE_GROUP);
    edge1B.setSource("1");
    edge1B.setDestination("B");
    edge1B.setDirected(false);
    edge1B.putProperty("count", 2);
    expectedElements.add(edge1B);
    assertEquals(expectedElements, results);
    // Check get entity for 1 when query for 1 -> B and specify entities only
    rddQuery = new GetRDDOfElements.Builder<EdgeSeed>().sparkContext(sparkContext).seeds(Collections.singleton(new EdgeSeed("1", "B", false))).includeEntities(true).includeEdges(GetOperation.IncludeEdgeType.NONE).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    results.clear();
    returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    expectedElements.clear();
    final Entity entity1 = new Entity(ENTITY_GROUP);
    entity1.setVertex("1");
    expectedElements.add(entity1);
    assertEquals(expectedElements, results);
    // Check get correct edges for 1 -> B when specify edges only
    rddQuery = new GetRDDOfElements.Builder<EdgeSeed>().sparkContext(sparkContext).seeds(Collections.singleton(new EdgeSeed("1", "B", false))).view(new View.Builder().edge(EDGE_GROUP).build()).includeEntities(false).includeEdges(GetOperation.IncludeEdgeType.ALL).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    results.clear();
    returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    expectedElements.clear();
    expectedElements.add(edge1B);
    assertEquals(expectedElements, results);
    // Check get correct edges for 1 -> B and 5 -> C
    Set<EdgeSeed> seeds = new HashSet<>();
    seeds.add(new EdgeSeed("1", "B", false));
    seeds.add(new EdgeSeed("5", "C", false));
    rddQuery = new GetRDDOfElements.Builder<EdgeSeed>().sparkContext(sparkContext).includeEntities(false).seeds(seeds).build();
    rddQuery.addOption(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString);
    rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    results.clear();
    returnedElements = (Element[]) rdd.collect();
    for (int i = 0; i < returnedElements.length; i++) {
        results.add(returnedElements[i]);
    }
    final Edge edge5C = new Edge(EDGE_GROUP);
    edge5C.setSource("5");
    edge5C.setDestination("C");
    edge5C.setDirected(false);
    edge5C.putProperty("count", 4);
    expectedElements.clear();
    expectedElements.add(edge1B);
    expectedElements.add(edge5C);
    assertEquals(expectedElements, results);
    sparkContext.stop();
}

Example 5 with SparkContext

use of org.apache.spark.SparkContext in project Gaffer by gchq.

the class ImportKeyValuePairRDDToAccumuloHandlerTest method checkImportRDDOfElements.

@Test
public void checkImportRDDOfElements() throws OperationException, IOException {
    final Graph graph1 = new Graph.Builder().addSchema(getClass().getResourceAsStream("/schema/dataSchema.json")).addSchema(getClass().getResourceAsStream("/schema/dataTypes.json")).addSchema(getClass().getResourceAsStream("/schema/storeTypes.json")).addSchema(getClass().getResourceAsStream("/schema/storeSchema.json")).storeProperties(getClass().getResourceAsStream("/store.properties")).build();
    final ArrayBuffer<Element> elements = new ArrayBuffer<>();
    for (int i = 0; i < 10; i++) {
        final Entity entity = new Entity(TestGroups.ENTITY);
        entity.setVertex("" + i);
        final Edge edge1 = new Edge(TestGroups.EDGE);
        edge1.setSource("" + i);
        edge1.setDestination("B");
        edge1.setDirected(false);
        edge1.putProperty(TestPropertyNames.COUNT, 2);
        final Edge edge2 = new Edge(TestGroups.EDGE);
        edge2.setSource("" + i);
        edge2.setDestination("C");
        edge2.setDirected(false);
        edge2.putProperty(TestPropertyNames.COUNT, 4);
        elements.$plus$eq(edge1);
        elements.$plus$eq(edge2);
        elements.$plus$eq(entity);
    }
    final User user = new User();
    final SparkConf sparkConf = new SparkConf().setMaster("local").setAppName("tests").set("spark.serializer", "org.apache.spark.serializer.KryoSerializer").set("spark.kryo.registrator", "uk.gov.gchq.gaffer.spark.serialisation.kryo.Registrator").set("spark.driver.allowMultipleContexts", "true");
    final SparkContext sparkContext = new SparkContext(sparkConf);
    // Create Hadoop configuration and serialise to a string
    final Configuration configuration = new Configuration();
    final ByteArrayOutputStream baos = new ByteArrayOutputStream();
    configuration.write(new DataOutputStream(baos));
    final String configurationString = new String(baos.toByteArray(), CommonConstants.UTF_8);
    final String outputPath = this.getClass().getResource("/").getPath().toString() + "load";
    final String failurePath = this.getClass().getResource("/").getPath().toString() + "failure";
    final File file = new File(outputPath);
    if (file.exists()) {
        FileUtils.forceDelete(file);
    }
    final ElementConverterFunction func = new ElementConverterFunction(sparkContext.broadcast(new ByteEntityAccumuloElementConverter(graph1.getSchema()), ACCUMULO_ELEMENT_CONVERTER_CLASS_TAG));
    final RDD<Tuple2<Key, Value>> elementRDD = sparkContext.parallelize(elements, 1, ELEMENT_CLASS_TAG).flatMap(func, TUPLE2_CLASS_TAG);
    final ImportKeyValuePairRDDToAccumulo addRdd = new ImportKeyValuePairRDDToAccumulo.Builder().input(elementRDD).outputPath(outputPath).failurePath(failurePath).build();
    graph1.execute(addRdd, user);
    FileUtils.forceDelete(file);
    // Check all elements were added
    final GetRDDOfAllElements rddQuery = new GetRDDOfAllElements.Builder().sparkContext(sparkContext).option(AbstractGetRDDHandler.HADOOP_CONFIGURATION_KEY, configurationString).build();
    final RDD<Element> rdd = graph1.execute(rddQuery, user);
    if (rdd == null) {
        fail("No RDD returned");
    }
    final Set<Element> results = new HashSet<>();
    final Element[] returnedElements = (Element[]) rdd.collect();
    Collections.addAll(results, returnedElements);
    assertEquals(elements.size(), results.size());
    sparkContext.stop();
}