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

Example 11 with DataSet

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

the class PythonPlanBinder method createTextSink.

@SuppressWarnings("unchecked")
private void createTextSink(PythonOperationInfo info) throws IOException {
    DataSet parent = (DataSet) sets.get(info.parentID);
    parent.map(new StringDeserializerMap()).setParallelism(getParallelism(info)).writeAsText(info.path, info.writeMode).setParallelism(getParallelism(info)).name("TextSink");
}

Example 12 with DataSet

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

the class PythonPlanBinder method createRebalanceOperation.

private void createRebalanceOperation(PythonOperationInfo info) throws IOException {
    DataSet op = (DataSet) sets.get(info.parentID);
    sets.put(info.setID, op.rebalance().setParallelism(getParallelism(info)).name("Rebalance"));
}

Example 13 with DataSet

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

the class DeltaIterationTranslationTest method testCorrectTranslation.

@Test
public void testCorrectTranslation() {
    try {
        final String JOB_NAME = "Test JobName";
        final String ITERATION_NAME = "Test Name";
        final String BEFORE_NEXT_WORKSET_MAP = "Some Mapper";
        final String AGGREGATOR_NAME = "AggregatorName";
        final int[] ITERATION_KEYS = new int[] { 2 };
        final int NUM_ITERATIONS = 13;
        final int DEFAULT_parallelism = 133;
        final int ITERATION_parallelism = 77;
        ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
        // ------------ construct the test program ------------------
        {
            env.setParallelism(DEFAULT_parallelism);
            @SuppressWarnings("unchecked") DataSet<Tuple3<Double, Long, String>> initialSolutionSet = env.fromElements(new Tuple3<Double, Long, String>(3.44, 5L, "abc"));
            @SuppressWarnings("unchecked") DataSet<Tuple2<Double, String>> initialWorkSet = env.fromElements(new Tuple2<Double, String>(1.23, "abc"));
            DeltaIteration<Tuple3<Double, Long, String>, Tuple2<Double, String>> iteration = initialSolutionSet.iterateDelta(initialWorkSet, NUM_ITERATIONS, ITERATION_KEYS);
            iteration.name(ITERATION_NAME).parallelism(ITERATION_parallelism);
            iteration.registerAggregator(AGGREGATOR_NAME, new LongSumAggregator());
            // test that multiple workset consumers are supported
            DataSet<Tuple2<Double, String>> worksetSelfJoin = iteration.getWorkset().map(new IdentityMapper<Tuple2<Double, String>>()).join(iteration.getWorkset()).where(1).equalTo(1).projectFirst(0, 1);
            DataSet<Tuple3<Double, Long, String>> joined = worksetSelfJoin.join(iteration.getSolutionSet()).where(1).equalTo(2).with(new SolutionWorksetJoin());
            DataSet<Tuple3<Double, Long, String>> result = iteration.closeWith(joined, joined.map(new NextWorksetMapper()).name(BEFORE_NEXT_WORKSET_MAP));
            result.output(new DiscardingOutputFormat<Tuple3<Double, Long, String>>());
            result.writeAsText("/dev/null");
        }
        Plan p = env.createProgramPlan(JOB_NAME);
        // ------------- validate the plan ----------------
        assertEquals(JOB_NAME, p.getJobName());
        assertEquals(DEFAULT_parallelism, p.getDefaultParallelism());
        // validate the iteration
        GenericDataSinkBase<?> sink1, sink2;
        {
            Iterator<? extends GenericDataSinkBase<?>> sinks = p.getDataSinks().iterator();
            sink1 = sinks.next();
            sink2 = sinks.next();
        }
        DeltaIterationBase<?, ?> iteration = (DeltaIterationBase<?, ?>) sink1.getInput();
        // check that multi consumer translation works for iterations
        assertEquals(iteration, sink2.getInput());
        // check the basic iteration properties
        assertEquals(NUM_ITERATIONS, iteration.getMaximumNumberOfIterations());
        assertArrayEquals(ITERATION_KEYS, iteration.getSolutionSetKeyFields());
        assertEquals(ITERATION_parallelism, iteration.getParallelism());
        assertEquals(ITERATION_NAME, iteration.getName());
        MapOperatorBase<?, ?, ?> nextWorksetMapper = (MapOperatorBase<?, ?, ?>) iteration.getNextWorkset();
        InnerJoinOperatorBase<?, ?, ?, ?> solutionSetJoin = (InnerJoinOperatorBase<?, ?, ?, ?>) iteration.getSolutionSetDelta();
        InnerJoinOperatorBase<?, ?, ?, ?> worksetSelfJoin = (InnerJoinOperatorBase<?, ?, ?, ?>) solutionSetJoin.getFirstInput();
        MapOperatorBase<?, ?, ?> worksetMapper = (MapOperatorBase<?, ?, ?>) worksetSelfJoin.getFirstInput();
        assertEquals(IdentityMapper.class, worksetMapper.getUserCodeWrapper().getUserCodeClass());
        assertEquals(NextWorksetMapper.class, nextWorksetMapper.getUserCodeWrapper().getUserCodeClass());
        if (solutionSetJoin.getUserCodeWrapper().getUserCodeObject() instanceof WrappingFunction) {
            WrappingFunction<?> wf = (WrappingFunction<?>) solutionSetJoin.getUserCodeWrapper().getUserCodeObject();
            assertEquals(SolutionWorksetJoin.class, wf.getWrappedFunction().getClass());
        } else {
            assertEquals(SolutionWorksetJoin.class, solutionSetJoin.getUserCodeWrapper().getUserCodeClass());
        }
        assertEquals(BEFORE_NEXT_WORKSET_MAP, nextWorksetMapper.getName());
        assertEquals(AGGREGATOR_NAME, iteration.getAggregators().getAllRegisteredAggregators().iterator().next().getName());
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        fail(e.getMessage());
    }
}

Example 14 with DataSet

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

the class ClusteringCoefficient 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);
    if (!parameters.has("directed")) {
        throw new ProgramParametrizationException(getUsage("must declare execution mode as '--directed true' or '--directed false'"));
    }
    boolean directedAlgorithm = parameters.getBoolean("directed");
    int little_parallelism = parameters.getInt("little_parallelism", PARALLELISM_DEFAULT);
    // global and local clustering coefficient results
    GraphAnalytic gcc;
    GraphAnalytic acc;
    DataSet lcc;
    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.get("input_filename"), env).ignoreCommentsEdges("#").lineDelimiterEdges(lineDelimiter).fieldDelimiterEdges(fieldDelimiter);
                switch(parameters.get("type", "")) {
                    case "integer":
                        {
                            Graph<LongValue, NullValue, NullValue> graph = reader.keyType(LongValue.class);
                            if (directedAlgorithm) {
                                if (parameters.getBoolean("simplify", false)) {
                                    graph = graph.run(new org.apache.flink.graph.asm.simple.directed.Simplify<LongValue, NullValue, NullValue>().setParallelism(little_parallelism));
                                }
                                gcc = graph.run(new org.apache.flink.graph.library.clustering.directed.GlobalClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                acc = graph.run(new org.apache.flink.graph.library.clustering.directed.AverageClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                lcc = graph.run(new org.apache.flink.graph.library.clustering.directed.LocalClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                            } else {
                                if (parameters.getBoolean("simplify", false)) {
                                    graph = graph.run(new org.apache.flink.graph.asm.simple.undirected.Simplify<LongValue, NullValue, NullValue>(false).setParallelism(little_parallelism));
                                }
                                gcc = graph.run(new org.apache.flink.graph.library.clustering.undirected.GlobalClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                acc = graph.run(new org.apache.flink.graph.library.clustering.undirected.AverageClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                lcc = graph.run(new org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                            }
                        }
                        break;
                    case "string":
                        {
                            Graph<StringValue, NullValue, NullValue> graph = reader.keyType(StringValue.class);
                            if (directedAlgorithm) {
                                if (parameters.getBoolean("simplify", false)) {
                                    graph = graph.run(new org.apache.flink.graph.asm.simple.directed.Simplify<StringValue, NullValue, NullValue>().setParallelism(little_parallelism));
                                }
                                gcc = graph.run(new org.apache.flink.graph.library.clustering.directed.GlobalClusteringCoefficient<StringValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                acc = graph.run(new org.apache.flink.graph.library.clustering.directed.AverageClusteringCoefficient<StringValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                lcc = graph.run(new org.apache.flink.graph.library.clustering.directed.LocalClusteringCoefficient<StringValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                            } else {
                                if (parameters.getBoolean("simplify", false)) {
                                    graph = graph.run(new org.apache.flink.graph.asm.simple.undirected.Simplify<StringValue, NullValue, NullValue>(false).setParallelism(little_parallelism));
                                }
                                gcc = graph.run(new org.apache.flink.graph.library.clustering.undirected.GlobalClusteringCoefficient<StringValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                acc = graph.run(new org.apache.flink.graph.library.clustering.undirected.AverageClusteringCoefficient<StringValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                                lcc = graph.run(new org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient<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();
                if (directedAlgorithm) {
                    if (scale > 32) {
                        Graph<LongValue, NullValue, NullValue> newGraph = graph.run(new org.apache.flink.graph.asm.simple.directed.Simplify<LongValue, NullValue, NullValue>().setParallelism(little_parallelism));
                        gcc = newGraph.run(new org.apache.flink.graph.library.clustering.directed.GlobalClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        acc = newGraph.run(new org.apache.flink.graph.library.clustering.directed.AverageClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        lcc = newGraph.run(new org.apache.flink.graph.library.clustering.directed.LocalClusteringCoefficient<LongValue, NullValue, NullValue>().setIncludeZeroDegreeVertices(false).setLittleParallelism(little_parallelism));
                    } else {
                        Graph<IntValue, NullValue, NullValue> newGraph = graph.run(new TranslateGraphIds<LongValue, IntValue, NullValue, NullValue>(new LongValueToUnsignedIntValue()).setParallelism(little_parallelism)).run(new org.apache.flink.graph.asm.simple.directed.Simplify<IntValue, NullValue, NullValue>().setParallelism(little_parallelism));
                        gcc = newGraph.run(new org.apache.flink.graph.library.clustering.directed.GlobalClusteringCoefficient<IntValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        acc = newGraph.run(new org.apache.flink.graph.library.clustering.directed.AverageClusteringCoefficient<IntValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        lcc = newGraph.run(new org.apache.flink.graph.library.clustering.directed.LocalClusteringCoefficient<IntValue, NullValue, NullValue>().setIncludeZeroDegreeVertices(false).setLittleParallelism(little_parallelism));
                    }
                } else {
                    boolean clipAndFlip = parameters.getBoolean("clip_and_flip", DEFAULT_CLIP_AND_FLIP);
                    if (scale > 32) {
                        Graph<LongValue, NullValue, NullValue> newGraph = graph.run(new org.apache.flink.graph.asm.simple.undirected.Simplify<LongValue, NullValue, NullValue>(clipAndFlip).setParallelism(little_parallelism));
                        gcc = newGraph.run(new org.apache.flink.graph.library.clustering.undirected.GlobalClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        acc = newGraph.run(new org.apache.flink.graph.library.clustering.undirected.AverageClusteringCoefficient<LongValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        lcc = newGraph.run(new org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient<LongValue, NullValue, NullValue>().setIncludeZeroDegreeVertices(false).setLittleParallelism(little_parallelism));
                    } else {
                        Graph<IntValue, NullValue, NullValue> newGraph = graph.run(new TranslateGraphIds<LongValue, IntValue, NullValue, NullValue>(new LongValueToUnsignedIntValue()).setParallelism(little_parallelism)).run(new org.apache.flink.graph.asm.simple.undirected.Simplify<IntValue, NullValue, NullValue>(clipAndFlip).setParallelism(little_parallelism));
                        gcc = newGraph.run(new org.apache.flink.graph.library.clustering.undirected.GlobalClusteringCoefficient<IntValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        acc = newGraph.run(new org.apache.flink.graph.library.clustering.undirected.AverageClusteringCoefficient<IntValue, NullValue, NullValue>().setLittleParallelism(little_parallelism));
                        lcc = newGraph.run(new org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient<IntValue, NullValue, NullValue>().setIncludeZeroDegreeVertices(false).setLittleParallelism(little_parallelism));
                    }
                }
            }
            break;
        default:
            throw new ProgramParametrizationException(getUsage("invalid input type"));
    }
    switch(parameters.get("output", "")) {
        case "print":
            if (directedAlgorithm) {
                for (Object e : lcc.collect()) {
                    org.apache.flink.graph.library.clustering.directed.LocalClusteringCoefficient.Result result = (org.apache.flink.graph.library.clustering.directed.LocalClusteringCoefficient.Result) e;
                    System.out.println(result.toPrintableString());
                }
            } else {
                for (Object e : lcc.collect()) {
                    org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient.Result result = (org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient.Result) e;
                    System.out.println(result.toPrintableString());
                }
            }
            break;
        case "hash":
            System.out.println(DataSetUtils.checksumHashCode(lcc));
            break;
        case "csv":
            String filename = parameters.get("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));
            lcc.writeAsCsv(filename, lineDelimiter, fieldDelimiter);
            env.execute("Clustering Coefficient");
            break;
        default:
            throw new ProgramParametrizationException(getUsage("invalid output type"));
    }
    System.out.println(gcc.getResult());
    System.out.println(acc.getResult());
    JobExecutionResult result = env.getLastJobExecutionResult();
    NumberFormat nf = NumberFormat.getInstance();
    System.out.println("Execution runtime: " + nf.format(result.getNetRuntime()) + " ms");
}

Example 15 with DataSet

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

the class HITS 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 iterations = parameters.getInt("iterations", DEFAULT_ITERATIONS);
    DataSet hits;
    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":
                        {
                            hits = reader.keyType(LongValue.class).run(new org.apache.flink.graph.library.link_analysis.HITS<LongValue, NullValue, NullValue>(iterations));
                        }
                        break;
                    case "string":
                        {
                            hits = reader.keyType(StringValue.class).run(new org.apache.flink.graph.library.link_analysis.HITS<StringValue, NullValue, NullValue>(iterations));
                        }
                        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).generate();
                if (scale > 32) {
                    hits = graph.run(new Simplify<LongValue, NullValue, NullValue>()).run(new org.apache.flink.graph.library.link_analysis.HITS<LongValue, NullValue, NullValue>(iterations));
                } else {
                    hits = graph.run(new TranslateGraphIds<LongValue, IntValue, NullValue, NullValue>(new LongValueToUnsignedIntValue())).run(new Simplify<IntValue, NullValue, NullValue>()).run(new org.apache.flink.graph.library.link_analysis.HITS<IntValue, NullValue, NullValue>(iterations));
                }
            }
            break;
        default:
            throw new ProgramParametrizationException(getUsage("invalid input type"));
    }
    switch(parameters.get("output", "")) {
        case "print":
            System.out.println();
            for (Object e : hits.collect()) {
                System.out.println(((Result) e).toPrintableString());
            }
            break;
        case "hash":
            System.out.println();
            System.out.println(DataSetUtils.checksumHashCode(hits));
            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));
            hits.writeAsCsv(filename, lineDelimiter, fieldDelimiter);
            env.execute("HITS");
            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");
}