Examples with DataSet org.apache.flink.api.java.DataSet used on opensource projects

Example 16 with DataSet

use of org.apache.flink.api.java.DataSet in project flink by apache.

the class JaccardIndex method main.

public static void main(String[] args) throws Exception {
    // Set up the execution environment
    final ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
    env.getConfig().enableObjectReuse();
    ParameterTool parameters = ParameterTool.fromArgs(args);
    env.getConfig().setGlobalJobParameters(parameters);
    int little_parallelism = parameters.getInt("little_parallelism", PARALLELISM_DEFAULT);
    DataSet ji;
    switch(parameters.get("input", "")) {
        case "csv":
            {
                String lineDelimiter = StringEscapeUtils.unescapeJava(parameters.get("input_line_delimiter", CsvOutputFormat.DEFAULT_LINE_DELIMITER));
                String fieldDelimiter = StringEscapeUtils.unescapeJava(parameters.get("input_field_delimiter", CsvOutputFormat.DEFAULT_FIELD_DELIMITER));
                GraphCsvReader reader = Graph.fromCsvReader(parameters.getRequired("input_filename"), env).ignoreCommentsEdges("#").lineDelimiterEdges(lineDelimiter).fieldDelimiterEdges(fieldDelimiter);
                switch(parameters.get("type", "")) {
                    case "integer":
                        {
                            Graph<LongValue, NullValue, NullValue> graph = reader.keyType(LongValue.class);
                            if (parameters.getBoolean("simplify", false)) {
                                graph = graph.run(new org.apache.flink.graph.asm.simple.undirected.Simplify<LongValue, NullValue, NullValue>(false).setParallelism(little_parallelism));
                            }
                            ji = graph.run(new org.apache.flink.graph.library.similarity.JaccardIndex<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        }
                        break;
                    case "string":
                        {
                            Graph<StringValue, NullValue, NullValue> graph = reader.keyType(StringValue.class);
                            if (parameters.getBoolean("simplify", false)) {
                                graph = graph.run(new org.apache.flink.graph.asm.simple.undirected.Simplify<StringValue, NullValue, NullValue>(false).setParallelism(little_parallelism));
                            }
                            ji = graph.run(new org.apache.flink.graph.library.similarity.JaccardIndex<StringValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        }
                        break;
                    default:
                        throw new ProgramParametrizationException(getUsage("invalid CSV type"));
                }
            }
            break;
        case "rmat":
            {
                int scale = parameters.getInt("scale", DEFAULT_SCALE);
                int edgeFactor = parameters.getInt("edge_factor", DEFAULT_EDGE_FACTOR);
                RandomGenerableFactory<JDKRandomGenerator> rnd = new JDKRandomGeneratorFactory();
                long vertexCount = 1L << scale;
                long edgeCount = vertexCount * edgeFactor;
                Graph<LongValue, NullValue, NullValue> graph = new RMatGraph<>(env, rnd, vertexCount, edgeCount).setParallelism(little_parallelism).generate();
                boolean clipAndFlip = parameters.getBoolean("clip_and_flip", DEFAULT_CLIP_AND_FLIP);
                if (scale > 32) {
                    ji = graph.run(new Simplify<LongValue, NullValue, NullValue>(clipAndFlip).setParallelism(little_parallelism)).run(new org.apache.flink.graph.library.similarity.JaccardIndex<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                } else {
                    ji = graph.run(new TranslateGraphIds<LongValue, IntValue, NullValue, NullValue>(new LongValueToUnsignedIntValue()).setParallelism(little_parallelism)).run(new Simplify<IntValue, NullValue, NullValue>(clipAndFlip).setParallelism(little_parallelism)).run(new org.apache.flink.graph.library.similarity.JaccardIndex<IntValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                }
            }
            break;
        default:
            throw new ProgramParametrizationException(getUsage("invalid input type"));
    }
    switch(parameters.get("output", "")) {
        case "print":
            System.out.println();
            for (Object e : ji.collect()) {
                Result result = (Result) e;
                System.out.println(result.toPrintableString());
            }
            break;
        case "hash":
            System.out.println();
            System.out.println(DataSetUtils.checksumHashCode(ji));
            break;
        case "csv":
            String filename = parameters.getRequired("output_filename");
            String lineDelimiter = StringEscapeUtils.unescapeJava(parameters.get("output_line_delimiter", CsvOutputFormat.DEFAULT_LINE_DELIMITER));
            String fieldDelimiter = StringEscapeUtils.unescapeJava(parameters.get("output_field_delimiter", CsvOutputFormat.DEFAULT_FIELD_DELIMITER));
            ji.writeAsCsv(filename, lineDelimiter, fieldDelimiter);
            env.execute("Jaccard Index");
            break;
        default:
            throw new ProgramParametrizationException(getUsage("invalid output type"));
    }
    JobExecutionResult result = env.getLastJobExecutionResult();
    NumberFormat nf = NumberFormat.getInstance();
    System.out.println();
    System.out.println("Execution runtime: " + nf.format(result.getNetRuntime()) + " ms");
}

Example 17 with DataSet

use of org.apache.flink.api.java.DataSet in project flink by apache.

the class PregelCompilerTest method testPregelCompilerWithBroadcastVariable.

@SuppressWarnings("serial")
@Test
public void testPregelCompilerWithBroadcastVariable() {
    try {
        final String BC_VAR_NAME = "borat variable";
        ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
        env.setParallelism(DEFAULT_PARALLELISM);
        // compose test program
        {
            DataSet<Long> bcVar = env.fromElements(1L);
            DataSet<Vertex<Long, Long>> initialVertices = env.fromElements(new Tuple2<>(1L, 1L), new Tuple2<>(2L, 2L)).map(new Tuple2ToVertexMap<Long, Long>());
            DataSet<Edge<Long, NullValue>> edges = env.fromElements(new Tuple2<>(1L, 2L)).map(new MapFunction<Tuple2<Long, Long>, Edge<Long, NullValue>>() {

                public Edge<Long, NullValue> map(Tuple2<Long, Long> edge) {
                    return new Edge<>(edge.f0, edge.f1, NullValue.getInstance());
                }
            });
            Graph<Long, Long, NullValue> graph = Graph.fromDataSet(initialVertices, edges, env);
            VertexCentricConfiguration parameters = new VertexCentricConfiguration();
            parameters.addBroadcastSet(BC_VAR_NAME, bcVar);
            DataSet<Vertex<Long, Long>> result = graph.runVertexCentricIteration(new CCCompute(), null, 100, parameters).getVertices();
            result.output(new DiscardingOutputFormat<Vertex<Long, Long>>());
        }
        Plan p = env.createProgramPlan("Pregel Connected Components");
        OptimizedPlan op = compileNoStats(p);
        // check the sink
        SinkPlanNode sink = op.getDataSinks().iterator().next();
        assertEquals(ShipStrategyType.FORWARD, sink.getInput().getShipStrategy());
        assertEquals(DEFAULT_PARALLELISM, sink.getParallelism());
        // check the iteration
        WorksetIterationPlanNode iteration = (WorksetIterationPlanNode) sink.getInput().getSource();
        assertEquals(DEFAULT_PARALLELISM, iteration.getParallelism());
        // check the solution set delta
        PlanNode ssDelta = iteration.getSolutionSetDeltaPlanNode();
        assertTrue(ssDelta instanceof SingleInputPlanNode);
        SingleInputPlanNode ssFlatMap = (SingleInputPlanNode) ((SingleInputPlanNode) (ssDelta)).getInput().getSource();
        assertEquals(DEFAULT_PARALLELISM, ssFlatMap.getParallelism());
        assertEquals(ShipStrategyType.FORWARD, ssFlatMap.getInput().getShipStrategy());
        // check the computation coGroup
        DualInputPlanNode computationCoGroup = (DualInputPlanNode) (ssFlatMap.getInput().getSource());
        assertEquals(DEFAULT_PARALLELISM, computationCoGroup.getParallelism());
        assertEquals(ShipStrategyType.FORWARD, computationCoGroup.getInput1().getShipStrategy());
        assertEquals(ShipStrategyType.PARTITION_HASH, computationCoGroup.getInput2().getShipStrategy());
        assertTrue(computationCoGroup.getInput2().getTempMode().isCached());
        assertEquals(new FieldList(0), computationCoGroup.getInput2().getShipStrategyKeys());
        // check that the initial partitioning is pushed out of the loop
        assertEquals(ShipStrategyType.PARTITION_HASH, iteration.getInput1().getShipStrategy());
        assertEquals(new FieldList(0), iteration.getInput1().getShipStrategyKeys());
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        fail(e.getMessage());
    }
}

Example 18 with DataSet

use of org.apache.flink.api.java.DataSet in project flink by apache.

the class PregelTranslationTest method testTranslationPlainEdges.

@Test
public void testTranslationPlainEdges() {
    try {
        final String ITERATION_NAME = "Test Name";
        final String AGGREGATOR_NAME = "AggregatorName";
        final String BC_SET_NAME = "borat messages";
        final int NUM_ITERATIONS = 13;
        final int ITERATION_parallelism = 77;
        ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
        DataSet<Long> bcVar = env.fromElements(1L);
        DataSet<Vertex<String, Double>> result;
        // ------------ construct the test program ------------------
        {
            DataSet<Tuple2<String, Double>> initialVertices = env.fromElements(new Tuple2<>("abc", 3.44));
            DataSet<Tuple2<String, String>> edges = env.fromElements(new Tuple2<>("a", "c"));
            Graph<String, Double, NullValue> graph = Graph.fromTupleDataSet(initialVertices, edges.map(new MapFunction<Tuple2<String, String>, Tuple3<String, String, NullValue>>() {

                public Tuple3<String, String, NullValue> map(Tuple2<String, String> edge) {
                    return new Tuple3<>(edge.f0, edge.f1, NullValue.getInstance());
                }
            }), env);
            VertexCentricConfiguration parameters = new VertexCentricConfiguration();
            parameters.addBroadcastSet(BC_SET_NAME, bcVar);
            parameters.setName(ITERATION_NAME);
            parameters.setParallelism(ITERATION_parallelism);
            parameters.registerAggregator(AGGREGATOR_NAME, new LongSumAggregator());
            result = graph.runVertexCentricIteration(new MyCompute(), null, NUM_ITERATIONS, parameters).getVertices();
            result.output(new DiscardingOutputFormat<Vertex<String, Double>>());
        }
        // ------------- validate the java program ----------------
        assertTrue(result instanceof DeltaIterationResultSet);
        DeltaIterationResultSet<?, ?> resultSet = (DeltaIterationResultSet<?, ?>) result;
        DeltaIteration<?, ?> iteration = resultSet.getIterationHead();
        // check the basic iteration properties
        assertEquals(NUM_ITERATIONS, resultSet.getMaxIterations());
        assertArrayEquals(new int[] { 0 }, resultSet.getKeyPositions());
        assertEquals(ITERATION_parallelism, iteration.getParallelism());
        assertEquals(ITERATION_NAME, iteration.getName());
        assertEquals(AGGREGATOR_NAME, iteration.getAggregators().getAllRegisteredAggregators().iterator().next().getName());
        TwoInputUdfOperator<?, ?, ?, ?> computationCoGroup = (TwoInputUdfOperator<?, ?, ?, ?>) ((SingleInputUdfOperator<?, ?, ?>) resultSet.getNextWorkset()).getInput();
        // validate that the broadcast sets are forwarded
        assertEquals(bcVar, computationCoGroup.getBroadcastSets().get(BC_SET_NAME));
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        fail(e.getMessage());
    }
}

Example 19 with DataSet

use of org.apache.flink.api.java.DataSet in project flink by apache.

the class SpargelCompilerTest method testSpargelCompilerWithBroadcastVariable.

@SuppressWarnings("serial")
@Test
public void testSpargelCompilerWithBroadcastVariable() {
    try {
        final String BC_VAR_NAME = "borat variable";
        ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
        env.setParallelism(DEFAULT_PARALLELISM);
        // compose test program
        {
            DataSet<Long> bcVar = env.fromElements(1L);
            DataSet<Vertex<Long, Long>> initialVertices = env.fromElements(new Tuple2<>(1L, 1L), new Tuple2<>(2L, 2L)).map(new Tuple2ToVertexMap<Long, Long>());
            DataSet<Edge<Long, NullValue>> edges = env.fromElements(new Tuple2<>(1L, 2L)).map(new MapFunction<Tuple2<Long, Long>, Edge<Long, NullValue>>() {

                public Edge<Long, NullValue> map(Tuple2<Long, Long> edge) {
                    return new Edge<>(edge.f0, edge.f1, NullValue.getInstance());
                }
            });
            Graph<Long, Long, NullValue> graph = Graph.fromDataSet(initialVertices, edges, env);
            ScatterGatherConfiguration parameters = new ScatterGatherConfiguration();
            parameters.addBroadcastSetForScatterFunction(BC_VAR_NAME, bcVar);
            parameters.addBroadcastSetForGatherFunction(BC_VAR_NAME, bcVar);
            DataSet<Vertex<Long, Long>> result = graph.runScatterGatherIteration(new ConnectedComponents.CCMessenger<Long, Long>(BasicTypeInfo.LONG_TYPE_INFO), new ConnectedComponents.CCUpdater<Long, Long>(), 100).getVertices();
            result.output(new DiscardingOutputFormat<Vertex<Long, Long>>());
        }
        Plan p = env.createProgramPlan("Spargel Connected Components");
        OptimizedPlan op = compileNoStats(p);
        // check the sink
        SinkPlanNode sink = op.getDataSinks().iterator().next();
        assertEquals(ShipStrategyType.FORWARD, sink.getInput().getShipStrategy());
        assertEquals(DEFAULT_PARALLELISM, sink.getParallelism());
        // check the iteration
        WorksetIterationPlanNode iteration = (WorksetIterationPlanNode) sink.getInput().getSource();
        assertEquals(DEFAULT_PARALLELISM, iteration.getParallelism());
        // check the solution set join and the delta
        PlanNode ssDelta = iteration.getSolutionSetDeltaPlanNode();
        // this is only true if the update functions preserves the partitioning
        assertTrue(ssDelta instanceof DualInputPlanNode);
        DualInputPlanNode ssJoin = (DualInputPlanNode) ssDelta;
        assertEquals(DEFAULT_PARALLELISM, ssJoin.getParallelism());
        assertEquals(ShipStrategyType.PARTITION_HASH, ssJoin.getInput1().getShipStrategy());
        assertEquals(new FieldList(0), ssJoin.getInput1().getShipStrategyKeys());
        // check the workset set join
        DualInputPlanNode edgeJoin = (DualInputPlanNode) ssJoin.getInput1().getSource();
        assertEquals(DEFAULT_PARALLELISM, edgeJoin.getParallelism());
        assertEquals(ShipStrategyType.PARTITION_HASH, edgeJoin.getInput1().getShipStrategy());
        assertEquals(ShipStrategyType.FORWARD, edgeJoin.getInput2().getShipStrategy());
        assertTrue(edgeJoin.getInput1().getTempMode().isCached());
        assertEquals(new FieldList(0), edgeJoin.getInput1().getShipStrategyKeys());
        // check that the initial partitioning is pushed out of the loop
        assertEquals(ShipStrategyType.PARTITION_HASH, iteration.getInput1().getShipStrategy());
        assertEquals(ShipStrategyType.PARTITION_HASH, iteration.getInput2().getShipStrategy());
        assertEquals(new FieldList(0), iteration.getInput1().getShipStrategyKeys());
        assertEquals(new FieldList(0), iteration.getInput2().getShipStrategyKeys());
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        fail(e.getMessage());
    }
}

Example 20 with DataSet

use of org.apache.flink.api.java.DataSet in project flink by apache.

the class SpargelTranslationTest method testTranslationPlainEdgesWithForkedBroadcastVariable.

@Test
public void testTranslationPlainEdgesWithForkedBroadcastVariable() {
    try {
        final String ITERATION_NAME = "Test Name";
        final String AGGREGATOR_NAME = "AggregatorName";
        final String BC_SET_MESSAGES_NAME = "borat messages";
        final String BC_SET_UPDATES_NAME = "borat updates";
        final int NUM_ITERATIONS = 13;
        final int ITERATION_parallelism = 77;
        ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
        DataSet<Long> bcVar = env.fromElements(1L);
        DataSet<Vertex<String, Double>> result;
        // ------------ construct the test program ------------------
        {
            DataSet<Tuple2<String, Double>> initialVertices = env.fromElements(new Tuple2<>("abc", 3.44));
            DataSet<Tuple2<String, String>> edges = env.fromElements(new Tuple2<>("a", "c"));
            Graph<String, Double, NullValue> graph = Graph.fromTupleDataSet(initialVertices, edges.map(new MapFunction<Tuple2<String, String>, Tuple3<String, String, NullValue>>() {

                public Tuple3<String, String, NullValue> map(Tuple2<String, String> edge) {
                    return new Tuple3<>(edge.f0, edge.f1, NullValue.getInstance());
                }
            }), env);
            ScatterGatherConfiguration parameters = new ScatterGatherConfiguration();
            parameters.addBroadcastSetForScatterFunction(BC_SET_MESSAGES_NAME, bcVar);
            parameters.addBroadcastSetForGatherFunction(BC_SET_UPDATES_NAME, bcVar);
            parameters.setName(ITERATION_NAME);
            parameters.setParallelism(ITERATION_parallelism);
            parameters.registerAggregator(AGGREGATOR_NAME, new LongSumAggregator());
            result = graph.runScatterGatherIteration(new MessageFunctionNoEdgeValue(), new UpdateFunction(), NUM_ITERATIONS, parameters).getVertices();
            result.output(new DiscardingOutputFormat<Vertex<String, Double>>());
        }
        // ------------- validate the java program ----------------
        assertTrue(result instanceof DeltaIterationResultSet);
        DeltaIterationResultSet<?, ?> resultSet = (DeltaIterationResultSet<?, ?>) result;
        DeltaIteration<?, ?> iteration = resultSet.getIterationHead();
        // check the basic iteration properties
        assertEquals(NUM_ITERATIONS, resultSet.getMaxIterations());
        assertArrayEquals(new int[] { 0 }, resultSet.getKeyPositions());
        assertEquals(ITERATION_parallelism, iteration.getParallelism());
        assertEquals(ITERATION_NAME, iteration.getName());
        assertEquals(AGGREGATOR_NAME, iteration.getAggregators().getAllRegisteredAggregators().iterator().next().getName());
        // validate that the semantic properties are set as they should
        TwoInputUdfOperator<?, ?, ?, ?> solutionSetJoin = (TwoInputUdfOperator<?, ?, ?, ?>) resultSet.getNextWorkset();
        assertTrue(solutionSetJoin.getSemanticProperties().getForwardingTargetFields(0, 0).contains(0));
        assertTrue(solutionSetJoin.getSemanticProperties().getForwardingTargetFields(1, 0).contains(0));
        TwoInputUdfOperator<?, ?, ?, ?> edgesJoin = (TwoInputUdfOperator<?, ?, ?, ?>) solutionSetJoin.getInput1();
        // validate that the broadcast sets are forwarded
        assertEquals(bcVar, solutionSetJoin.getBroadcastSets().get(BC_SET_UPDATES_NAME));
        assertEquals(bcVar, edgesJoin.getBroadcastSets().get(BC_SET_MESSAGES_NAME));
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        fail(e.getMessage());
    }
}