Examples with ExternalSortOperatorDescriptor org.apache.hyracks.dataflow.std.sort.ExternalSortOperatorDescriptor used on opensource projects

Example 6 with ExternalSortOperatorDescriptor

use of org.apache.hyracks.dataflow.std.sort.ExternalSortOperatorDescriptor in project asterixdb by apache.

the class DataflowTest method testHDFSReadWriteOperators.

/**
     * Test a job with only HDFS read and writes.
     *
     * @throws Exception
     */
public void testHDFSReadWriteOperators() throws Exception {
    FileInputFormat.setInputPaths(conf, HDFS_INPUT_PATH);
    FileOutputFormat.setOutputPath(conf, new Path(HDFS_OUTPUT_PATH));
    conf.setInputFormat(TextInputFormat.class);
    Scheduler scheduler = new Scheduler(HyracksUtils.CC_HOST, HyracksUtils.TEST_HYRACKS_CC_CLIENT_PORT);
    InputSplit[] splits = conf.getInputFormat().getSplits(conf, numberOfNC * 4);
    String[] readSchedule = scheduler.getLocationConstraints(splits);
    JobSpecification jobSpec = new JobSpecification();
    RecordDescriptor recordDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer() });
    String[] locations = new String[] { HyracksUtils.NC1_ID, HyracksUtils.NC1_ID, HyracksUtils.NC2_ID, HyracksUtils.NC2_ID };
    HDFSReadOperatorDescriptor readOperator = new HDFSReadOperatorDescriptor(jobSpec, recordDesc, conf, splits, readSchedule, new TextKeyValueParserFactory());
    PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, readOperator, locations);
    ExternalSortOperatorDescriptor sortOperator = new ExternalSortOperatorDescriptor(jobSpec, 10, new int[] { 0 }, new IBinaryComparatorFactory[] { RawBinaryComparatorFactory.INSTANCE }, recordDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, sortOperator, locations);
    HDFSWriteOperatorDescriptor writeOperator = new HDFSWriteOperatorDescriptor(jobSpec, conf, new TextTupleWriterFactory());
    PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, writeOperator, HyracksUtils.NC1_ID);
    jobSpec.connect(new OneToOneConnectorDescriptor(jobSpec), readOperator, 0, sortOperator, 0);
    jobSpec.connect(new MToNPartitioningMergingConnectorDescriptor(jobSpec, new FieldHashPartitionComputerFactory(new int[] { 0 }, new IBinaryHashFunctionFactory[] { RawBinaryHashFunctionFactory.INSTANCE }), new int[] { 0 }, new IBinaryComparatorFactory[] { RawBinaryComparatorFactory.INSTANCE }, null), sortOperator, 0, writeOperator, 0);
    jobSpec.addRoot(writeOperator);
    IHyracksClientConnection client = new HyracksConnection(HyracksUtils.CC_HOST, HyracksUtils.TEST_HYRACKS_CC_CLIENT_PORT);
    JobId jobId = client.startJob(jobSpec);
    client.waitForCompletion(jobId);
    Assert.assertEquals(true, checkResults());
}

Example 7 with ExternalSortOperatorDescriptor

use of org.apache.hyracks.dataflow.std.sort.ExternalSortOperatorDescriptor in project asterixdb by apache.

the class DataflowTest method testHDFSReadWriteOperators.

/**
     * Test a job with only HDFS read and writes.
     *
     * @throws Exception
     */
@SuppressWarnings({ "rawtypes", "unchecked" })
public void testHDFSReadWriteOperators() throws Exception {
    FileInputFormat.setInputPaths(conf, HDFS_INPUT_PATH);
    FileOutputFormat.setOutputPath(conf, new Path(HDFS_OUTPUT_PATH));
    conf.setInputFormatClass(TextInputFormat.class);
    Scheduler scheduler = new Scheduler(HyracksUtils.CC_HOST, HyracksUtils.TEST_HYRACKS_CC_CLIENT_PORT);
    InputFormat inputFormat = ReflectionUtils.newInstance(conf.getInputFormatClass(), getConfiguration());
    List<InputSplit> splits = inputFormat.getSplits(conf);
    String[] readSchedule = scheduler.getLocationConstraints(splits);
    JobSpecification jobSpec = new JobSpecification();
    RecordDescriptor recordDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer() });
    String[] locations = new String[] { HyracksUtils.NC1_ID, HyracksUtils.NC1_ID, HyracksUtils.NC2_ID, HyracksUtils.NC2_ID };
    HDFSReadOperatorDescriptor readOperator = new HDFSReadOperatorDescriptor(jobSpec, recordDesc, conf, splits, readSchedule, new TextKeyValueParserFactory());
    PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, readOperator, locations);
    ExternalSortOperatorDescriptor sortOperator = new ExternalSortOperatorDescriptor(jobSpec, 10, new int[] { 0 }, new IBinaryComparatorFactory[] { RawBinaryComparatorFactory.INSTANCE }, recordDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, sortOperator, locations);
    HDFSWriteOperatorDescriptor writeOperator = new HDFSWriteOperatorDescriptor(jobSpec, conf, new TextTupleWriterFactory());
    PartitionConstraintHelper.addAbsoluteLocationConstraint(jobSpec, writeOperator, HyracksUtils.NC1_ID);
    jobSpec.connect(new OneToOneConnectorDescriptor(jobSpec), readOperator, 0, sortOperator, 0);
    jobSpec.connect(new MToNPartitioningMergingConnectorDescriptor(jobSpec, new FieldHashPartitionComputerFactory(new int[] { 0 }, new IBinaryHashFunctionFactory[] { RawBinaryHashFunctionFactory.INSTANCE }), new int[] { 0 }, new IBinaryComparatorFactory[] { RawBinaryComparatorFactory.INSTANCE }, null), sortOperator, 0, writeOperator, 0);
    jobSpec.addRoot(writeOperator);
    IHyracksClientConnection client = new HyracksConnection(HyracksUtils.CC_HOST, HyracksUtils.TEST_HYRACKS_CC_CLIENT_PORT);
    JobId jobId = client.startJob(jobSpec);
    client.waitForCompletion(jobId);
    Assert.assertEquals(true, checkResults());
}

Example 8 with ExternalSortOperatorDescriptor

use of org.apache.hyracks.dataflow.std.sort.ExternalSortOperatorDescriptor in project asterixdb by apache.

the class SortMergeTest method sortMergeTest02.

@Test
public void sortMergeTest02() throws Exception {
    JobSpecification spec = new JobSpecification();
    FileSplit[] ordersSplits = new FileSplit[] { new ManagedFileSplit(NC1_ID, "data" + File.separator + "tpch0.001" + File.separator + "orders-part1.tbl"), new ManagedFileSplit(NC2_ID, "data" + File.separator + "tpch0.001" + File.separator + "orders-part2.tbl") };
    IFileSplitProvider ordersSplitProvider = new ConstantFileSplitProvider(ordersSplits);
    RecordDescriptor ordersDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() });
    FileScanOperatorDescriptor ordScanner = new FileScanOperatorDescriptor(spec, ordersSplitProvider, new DelimitedDataTupleParserFactory(new IValueParserFactory[] { UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE }, '|'), ordersDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, ordScanner, NC1_ID, NC2_ID);
    ExternalSortOperatorDescriptor sorter = new ExternalSortOperatorDescriptor(spec, 4, new int[] { 1, 0 }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY), PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, ordersDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, sorter, NC1_ID, NC2_ID);
    ResultSetId rsId = new ResultSetId(1);
    spec.addResultSetId(rsId);
    IOperatorDescriptor printer = new ResultWriterOperatorDescriptor(spec, rsId, false, false, ResultSerializerFactoryProvider.INSTANCE.getResultSerializerFactoryProvider());
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, printer, NC1_ID);
    spec.connect(new OneToOneConnectorDescriptor(spec), ordScanner, 0, sorter, 0);
    spec.connect(new MToNPartitioningMergingConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 1, 0 }, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY), PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }), new int[] { 1, 0 }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY), PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, new UTF8StringNormalizedKeyComputerFactory()), sorter, 0, printer, 0);
    runTest(spec);
}

Example 9 with ExternalSortOperatorDescriptor

use of org.apache.hyracks.dataflow.std.sort.ExternalSortOperatorDescriptor in project asterixdb by apache.

the class AbstractBTreeOperatorTest method loadPrimaryIndex.

protected void loadPrimaryIndex() throws Exception {
    JobSpecification spec = new JobSpecification();
    FileSplit[] ordersSplits = new FileSplit[] { new ManagedFileSplit(NC1_ID, "data" + File.separator + "tpch0.001" + File.separator + "orders-part1.tbl") };
    IFileSplitProvider ordersSplitProvider = new ConstantFileSplitProvider(ordersSplits);
    RecordDescriptor ordersDesc = inputRecordDesc;
    FileScanOperatorDescriptor ordScanner = new FileScanOperatorDescriptor(spec, ordersSplitProvider, new DelimitedDataTupleParserFactory(inputParserFactories, '|'), ordersDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, ordScanner, NC1_ID);
    ExternalSortOperatorDescriptor sorter = new ExternalSortOperatorDescriptor(spec, 1000, new int[] { 0 }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, ordersDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, sorter, NC1_ID);
    int[] fieldPermutation = { 0, 1, 2, 4, 5, 7 };
    TreeIndexBulkLoadOperatorDescriptor primaryBtreeBulkLoad = new TreeIndexBulkLoadOperatorDescriptor(spec, primaryRecDesc, fieldPermutation, 0.7f, true, 1000L, true, primaryHelperFactory);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, primaryBtreeBulkLoad, NC1_ID);
    NullSinkOperatorDescriptor nsOpDesc = new NullSinkOperatorDescriptor(spec);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, nsOpDesc, NC1_ID);
    spec.connect(new OneToOneConnectorDescriptor(spec), ordScanner, 0, sorter, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), sorter, 0, primaryBtreeBulkLoad, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), primaryBtreeBulkLoad, 0, nsOpDesc, 0);
    spec.addRoot(nsOpDesc);
    runTest(spec);
}

Example 10 with ExternalSortOperatorDescriptor

use of org.apache.hyracks.dataflow.std.sort.ExternalSortOperatorDescriptor in project asterixdb by apache.

the class PrimaryIndexBulkLoadExample method createJob.

private static JobSpecification createJob(Options options) {
    JobSpecification spec = new JobSpecification(options.frameSize);
    String[] splitNCs = options.ncs.split(",");
    // schema of tuples to be generated: 5 fields with string, string, int,
    // int, string
    // we will use field-index 2 as primary key to fill a clustered index
    RecordDescriptor recDesc = new RecordDescriptor(new ISerializerDeserializer[] { // this field will not go into B-Tree
    new UTF8StringSerializerDeserializer(), // we will use this as payload
    new UTF8StringSerializerDeserializer(), // we will use this field as key
    IntegerSerializerDeserializer.INSTANCE, // we will use this as payload
    IntegerSerializerDeserializer.INSTANCE, // we will use this as payload
    new UTF8StringSerializerDeserializer() });
    // generate numRecords records with field 2 being unique, integer values
    // in [0, 100000], and strings with max length of 10 characters, and
    // random seed 50
    DataGenOperatorDescriptor dataGen = new DataGenOperatorDescriptor(spec, recDesc, options.numTuples, 2, 0, 100000, 10, 50);
    // run data generator on first nodecontroller given
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, dataGen, splitNCs[0]);
    // sort the tuples as preparation for bulk load
    // fields to sort on
    int[] sortFields = { 2 };
    // comparators for sort fields
    IBinaryComparatorFactory[] comparatorFactories = new IBinaryComparatorFactory[1];
    comparatorFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    ExternalSortOperatorDescriptor sorter = new ExternalSortOperatorDescriptor(spec, options.sbSize, sortFields, comparatorFactories, recDesc);
    JobHelper.createPartitionConstraint(spec, sorter, splitNCs);
    // tuples to be put into B-Tree shall have 4 fields
    int fieldCount = 4;
    ITypeTraits[] typeTraits = new ITypeTraits[fieldCount];
    typeTraits[0] = IntegerPointable.TYPE_TRAITS;
    typeTraits[1] = UTF8StringPointable.TYPE_TRAITS;
    typeTraits[2] = IntegerPointable.TYPE_TRAITS;
    typeTraits[3] = UTF8StringPointable.TYPE_TRAITS;
    // create providers for B-Tree
    IStorageManager storageManager = BTreeHelperStorageManager.INSTANCE;
    // the B-Tree expects its keyfields to be at the front of its input
    // tuple
    // map field 2 of input tuple
    int[] fieldPermutation = { 2, 1, 3, 4 };
    // to field 0 of B-Tree tuple,
    // etc.
    IFileSplitProvider btreeSplitProvider = JobHelper.createFileSplitProvider(splitNCs, options.btreeName);
    IIndexDataflowHelperFactory dataflowHelperFactory = new IndexDataflowHelperFactory(storageManager, btreeSplitProvider);
    TreeIndexBulkLoadOperatorDescriptor btreeBulkLoad = new TreeIndexBulkLoadOperatorDescriptor(spec, recDesc, fieldPermutation, 0.7f, false, 1000L, true, dataflowHelperFactory);
    JobHelper.createPartitionConstraint(spec, btreeBulkLoad, splitNCs);
    // distribute the records from the datagen via hashing to the bulk load
    // ops
    IBinaryHashFunctionFactory[] hashFactories = new IBinaryHashFunctionFactory[1];
    hashFactories[0] = PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY);
    IConnectorDescriptor hashConn = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 0 }, hashFactories));
    NullSinkOperatorDescriptor nsOpDesc = new NullSinkOperatorDescriptor(spec);
    JobHelper.createPartitionConstraint(spec, nsOpDesc, splitNCs);
    spec.connect(hashConn, dataGen, 0, sorter, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), sorter, 0, btreeBulkLoad, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), btreeBulkLoad, 0, nsOpDesc, 0);
    spec.addRoot(nsOpDesc);
    return spec;
}