Examples with IBinaryHashFunctionFactory org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory used on opensource projects

Example 21 with IBinaryHashFunctionFactory

use of org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory in project asterixdb by apache.

the class TPCHCustomerOrderHashJoinTest method customerOrderCIDJoinMultiMaterialized.

@Test
public void customerOrderCIDJoinMultiMaterialized() throws Exception {
    JobSpecification spec = new JobSpecification();
    FileSplit[] custSplits = new FileSplit[] { new ManagedFileSplit(NC1_ID, "data" + File.separator + "tpch0.001" + File.separator + "customer-part1.tbl"), new ManagedFileSplit(NC2_ID, "data" + File.separator + "tpch0.001" + File.separator + "customer-part2.tbl") };
    IFileSplitProvider custSplitsProvider = new ConstantFileSplitProvider(custSplits);
    RecordDescriptor custDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() });
    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 ordersSplitsProvider = 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() });
    RecordDescriptor custOrderJoinDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() });
    FileScanOperatorDescriptor ordScanner = new FileScanOperatorDescriptor(spec, ordersSplitsProvider, 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);
    FileScanOperatorDescriptor custScanner = new FileScanOperatorDescriptor(spec, custSplitsProvider, new DelimitedDataTupleParserFactory(new IValueParserFactory[] { UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE }, '|'), custDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, custScanner, NC1_ID, NC2_ID);
    MaterializingOperatorDescriptor ordMat = new MaterializingOperatorDescriptor(spec, ordersDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, ordMat, NC1_ID, NC2_ID);
    MaterializingOperatorDescriptor custMat = new MaterializingOperatorDescriptor(spec, custDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, custMat, NC1_ID, NC2_ID);
    InMemoryHashJoinOperatorDescriptor join = new InMemoryHashJoinOperatorDescriptor(spec, new int[] { 1 }, new int[] { 0 }, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, custOrderJoinDesc, 128, null, 128);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, join, 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);
    IConnectorDescriptor ordPartConn = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 1 }, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }));
    spec.connect(ordPartConn, ordScanner, 0, ordMat, 0);
    IConnectorDescriptor custPartConn = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 0 }, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }));
    spec.connect(custPartConn, custScanner, 0, custMat, 0);
    IConnectorDescriptor ordJoinConn = new OneToOneConnectorDescriptor(spec);
    spec.connect(ordJoinConn, ordMat, 0, join, 0);
    IConnectorDescriptor custJoinConn = new OneToOneConnectorDescriptor(spec);
    spec.connect(custJoinConn, custMat, 0, join, 1);
    IConnectorDescriptor joinPrinterConn = new MToNBroadcastConnectorDescriptor(spec);
    spec.connect(joinPrinterConn, join, 0, printer, 0);
    spec.addRoot(printer);
    runTest(spec);
}

Example 22 with IBinaryHashFunctionFactory

use of org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory in project asterixdb by apache.

the class HashPartitionExchangePOperator method createConnectorDescriptor.

@Override
public Pair<IConnectorDescriptor, TargetConstraint> createConnectorDescriptor(IConnectorDescriptorRegistry spec, ILogicalOperator op, IOperatorSchema opSchema, JobGenContext context) throws AlgebricksException {
    int[] keys = new int[hashFields.size()];
    IBinaryHashFunctionFactory[] hashFunctionFactories = new IBinaryHashFunctionFactory[hashFields.size()];
    int i = 0;
    IBinaryHashFunctionFactoryProvider hashFunProvider = context.getBinaryHashFunctionFactoryProvider();
    IVariableTypeEnvironment env = context.getTypeEnvironment(op);
    for (LogicalVariable v : hashFields) {
        keys[i] = opSchema.findVariable(v);
        hashFunctionFactories[i] = hashFunProvider.getBinaryHashFunctionFactory(env.getVarType(v));
        ++i;
    }
    ITuplePartitionComputerFactory tpcf = new FieldHashPartitionComputerFactory(keys, hashFunctionFactories);
    IConnectorDescriptor conn = new MToNPartitioningConnectorDescriptor(spec, tpcf);
    return new Pair<>(conn, null);
}

Example 23 with IBinaryHashFunctionFactory

use of org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory in project asterixdb by apache.

the class HashPartitionMergeExchangePOperator method createConnectorDescriptor.

@Override
public Pair<IConnectorDescriptor, TargetConstraint> createConnectorDescriptor(IConnectorDescriptorRegistry spec, ILogicalOperator op, IOperatorSchema opSchema, JobGenContext context) throws AlgebricksException {
    int[] keys = new int[partitionFields.size()];
    IBinaryHashFunctionFactory[] hashFunctionFactories = new IBinaryHashFunctionFactory[partitionFields.size()];
    IVariableTypeEnvironment env = context.getTypeEnvironment(op);
    {
        int i = 0;
        IBinaryHashFunctionFactoryProvider hashFunProvider = context.getBinaryHashFunctionFactoryProvider();
        for (LogicalVariable v : partitionFields) {
            keys[i] = opSchema.findVariable(v);
            hashFunctionFactories[i] = hashFunProvider.getBinaryHashFunctionFactory(env.getVarType(v));
            ++i;
        }
    }
    ITuplePartitionComputerFactory tpcf = new FieldHashPartitionComputerFactory(keys, hashFunctionFactories);
    int n = orderColumns.size();
    int[] sortFields = new int[n];
    IBinaryComparatorFactory[] comparatorFactories = new IBinaryComparatorFactory[n];
    INormalizedKeyComputerFactoryProvider nkcfProvider = context.getNormalizedKeyComputerFactoryProvider();
    INormalizedKeyComputerFactory nkcf = null;
    int j = 0;
    for (OrderColumn oc : orderColumns) {
        LogicalVariable var = oc.getColumn();
        sortFields[j] = opSchema.findVariable(var);
        Object type = env.getVarType(var);
        IBinaryComparatorFactoryProvider bcfp = context.getBinaryComparatorFactoryProvider();
        comparatorFactories[j] = bcfp.getBinaryComparatorFactory(type, oc.getOrder() == OrderKind.ASC);
        if (j == 0 && nkcfProvider != null && type != null) {
            nkcf = nkcfProvider.getNormalizedKeyComputerFactory(type, oc.getOrder() == OrderKind.ASC);
        }
        j++;
    }
    IConnectorDescriptor conn = new MToNPartitioningMergingConnectorDescriptor(spec, tpcf, sortFields, comparatorFactories, nkcf);
    return new Pair<IConnectorDescriptor, TargetConstraint>(conn, null);
}

Example 24 with IBinaryHashFunctionFactory

use of org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory in project asterixdb by apache.

the class InMemoryHashJoinPOperator method contributeRuntimeOperator.

@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema propagatedSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
    int[] keysLeft = JobGenHelper.variablesToFieldIndexes(keysLeftBranch, inputSchemas[0]);
    int[] keysRight = JobGenHelper.variablesToFieldIndexes(keysRightBranch, inputSchemas[1]);
    IVariableTypeEnvironment env = context.getTypeEnvironment(op);
    IBinaryHashFunctionFactory[] hashFunFactories = JobGenHelper.variablesToBinaryHashFunctionFactories(keysLeftBranch, env, context);
    IBinaryComparatorFactory[] comparatorFactories = new IBinaryComparatorFactory[keysLeft.length];
    int i = 0;
    IBinaryComparatorFactoryProvider bcfp = context.getBinaryComparatorFactoryProvider();
    for (LogicalVariable v : keysLeftBranch) {
        Object t = env.getVarType(v);
        comparatorFactories[i++] = bcfp.getBinaryComparatorFactory(t, true);
    }
    IPredicateEvaluatorFactoryProvider predEvaluatorFactoryProvider = context.getPredicateEvaluatorFactoryProvider();
    IPredicateEvaluatorFactory predEvaluatorFactory = (predEvaluatorFactoryProvider == null ? null : predEvaluatorFactoryProvider.getPredicateEvaluatorFactory(keysLeft, keysRight));
    RecordDescriptor recDescriptor = JobGenHelper.mkRecordDescriptor(context.getTypeEnvironment(op), propagatedSchema, context);
    IOperatorDescriptorRegistry spec = builder.getJobSpec();
    IOperatorDescriptor opDesc = null;
    switch(kind) {
        case INNER:
            {
                opDesc = new InMemoryHashJoinOperatorDescriptor(spec, keysLeft, keysRight, hashFunFactories, comparatorFactories, recDescriptor, tableSize, predEvaluatorFactory, memSizeInFrames);
                break;
            }
        case LEFT_OUTER:
            {
                IMissingWriterFactory[] nonMatchWriterFactories = new IMissingWriterFactory[inputSchemas[1].getSize()];
                for (int j = 0; j < nonMatchWriterFactories.length; j++) {
                    nonMatchWriterFactories[j] = context.getMissingWriterFactory();
                }
                opDesc = new InMemoryHashJoinOperatorDescriptor(spec, keysLeft, keysRight, hashFunFactories, comparatorFactories, predEvaluatorFactory, recDescriptor, true, nonMatchWriterFactories, tableSize, memSizeInFrames);
                break;
            }
        default:
            {
                throw new NotImplementedException();
            }
    }
    contributeOpDesc(builder, (AbstractLogicalOperator) op, opDesc);
    ILogicalOperator src1 = op.getInputs().get(0).getValue();
    builder.contributeGraphEdge(src1, 0, op, 0);
    ILogicalOperator src2 = op.getInputs().get(1).getValue();
    builder.contributeGraphEdge(src2, 0, op, 1);
}

Example 25 with IBinaryHashFunctionFactory

use of org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory in project asterixdb by apache.

the class WordCountMain method createJob.

private static JobSpecification createJob(FileSplit[] inSplits, FileSplit[] outSplits, String algo, int htSize, int frameLimit, String format, int frameSize) {
    JobSpecification spec = new JobSpecification(frameSize);
    IFileSplitProvider splitsProvider = new ConstantFileSplitProvider(inSplits);
    RecordDescriptor wordDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer() });
    FileScanOperatorDescriptor wordScanner = new FileScanOperatorDescriptor(spec, splitsProvider, new WordTupleParserFactory(), wordDesc);
    createPartitionConstraint(spec, wordScanner, inSplits);
    RecordDescriptor groupResultDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer(), IntegerSerializerDeserializer.INSTANCE });
    IOperatorDescriptor gBy;
    int[] keys = new int[] { 0 };
    if ("hash".equalsIgnoreCase(algo)) {
        gBy = new ExternalGroupOperatorDescriptor(spec, htSize, fileSize, keys, frameLimit, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, new UTF8StringNormalizedKeyComputerFactory(), new MultiFieldsAggregatorFactory(new IFieldAggregateDescriptorFactory[] { new IntSumFieldAggregatorFactory(1, false), new IntSumFieldAggregatorFactory(3, false), new FloatSumFieldAggregatorFactory(5, false) }), new MultiFieldsAggregatorFactory(new IFieldAggregateDescriptorFactory[] { new IntSumFieldAggregatorFactory(1, false), new IntSumFieldAggregatorFactory(2, false), new FloatSumFieldAggregatorFactory(3, false) }), groupResultDesc, groupResultDesc, new HashSpillableTableFactory(new IBinaryHashFunctionFamily[] { UTF8StringBinaryHashFunctionFamily.INSTANCE }));
        createPartitionConstraint(spec, gBy, outSplits);
        IConnectorDescriptor scanGroupConn = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(keys, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }));
        spec.connect(scanGroupConn, wordScanner, 0, gBy, 0);
    } else {
        IBinaryComparatorFactory[] cfs = new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) };
        IOperatorDescriptor sorter = "memsort".equalsIgnoreCase(algo) ? new InMemorySortOperatorDescriptor(spec, keys, new UTF8StringNormalizedKeyComputerFactory(), cfs, wordDesc) : new ExternalSortOperatorDescriptor(spec, frameLimit, keys, new UTF8StringNormalizedKeyComputerFactory(), cfs, wordDesc);
        createPartitionConstraint(spec, sorter, outSplits);
        IConnectorDescriptor scanSortConn = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(keys, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }));
        spec.connect(scanSortConn, wordScanner, 0, sorter, 0);
        gBy = new PreclusteredGroupOperatorDescriptor(spec, keys, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, new MultiFieldsAggregatorFactory(new IFieldAggregateDescriptorFactory[] { new CountFieldAggregatorFactory(true) }), groupResultDesc);
        createPartitionConstraint(spec, gBy, outSplits);
        OneToOneConnectorDescriptor sortGroupConn = new OneToOneConnectorDescriptor(spec);
        spec.connect(sortGroupConn, sorter, 0, gBy, 0);
    }
    IFileSplitProvider outSplitProvider = new ConstantFileSplitProvider(outSplits);
    IOperatorDescriptor writer = "text".equalsIgnoreCase(format) ? new PlainFileWriterOperatorDescriptor(spec, outSplitProvider, ",") : new FrameFileWriterOperatorDescriptor(spec, outSplitProvider);
    createPartitionConstraint(spec, writer, outSplits);
    IConnectorDescriptor gbyPrinterConn = new OneToOneConnectorDescriptor(spec);
    spec.connect(gbyPrinterConn, gBy, 0, writer, 0);
    spec.addRoot(writer);
    return spec;
}