Examples with FileSplit org.apache.hyracks.api.io.FileSplit used on opensource projects

Example 1 with FileSplit

use of org.apache.hyracks.api.io.FileSplit in project asterixdb by apache.

the class LangExpressionToPlanTranslator method translate.

public ILogicalPlan translate(Query expr, String outputDatasetName, ICompiledDmlStatement stmt, ILogicalOperator baseOp) throws AlgebricksException {
    MutableObject<ILogicalOperator> base = new MutableObject<>(new EmptyTupleSourceOperator());
    if (baseOp != null) {
        base = new MutableObject<>(baseOp);
    }
    Pair<ILogicalOperator, LogicalVariable> p = expr.accept(this, base);
    ArrayList<Mutable<ILogicalOperator>> globalPlanRoots = new ArrayList<>();
    ILogicalOperator topOp = p.first;
    List<LogicalVariable> liveVars = new ArrayList<>();
    VariableUtilities.getLiveVariables(topOp, liveVars);
    LogicalVariable unnestVar = liveVars.get(0);
    LogicalVariable resVar = unnestVar;
    if (outputDatasetName == null) {
        FileSplit outputFileSplit = metadataProvider.getOutputFile();
        if (outputFileSplit == null) {
            outputFileSplit = getDefaultOutputFileLocation(metadataProvider.getApplicationContext());
        }
        metadataProvider.setOutputFile(outputFileSplit);
        List<Mutable<ILogicalExpression>> writeExprList = new ArrayList<>(1);
        writeExprList.add(new MutableObject<>(new VariableReferenceExpression(resVar)));
        ResultSetSinkId rssId = new ResultSetSinkId(metadataProvider.getResultSetId());
        ResultSetDataSink sink = new ResultSetDataSink(rssId, null);
        DistributeResultOperator newTop = new DistributeResultOperator(writeExprList, sink);
        newTop.getInputs().add(new MutableObject<>(topOp));
        topOp = newTop;
        // Retrieve the Output RecordType (if any) and store it on
        // the DistributeResultOperator
        IAType outputRecordType = metadataProvider.findOutputRecordType();
        if (outputRecordType != null) {
            topOp.getAnnotations().put("output-record-type", outputRecordType);
        }
    } else {
        /**
             * add the collection-to-sequence right before the project,
             * because dataset only accept non-collection records
             */
        LogicalVariable seqVar = context.newVar();
        /**
             * This assign adds a marker function collection-to-sequence: if the input is a singleton collection, unnest
             * it; otherwise do nothing.
             */
        AssignOperator assignCollectionToSequence = new AssignOperator(seqVar, new MutableObject<>(new ScalarFunctionCallExpression(FunctionUtil.getFunctionInfo(BuiltinFunctions.COLLECTION_TO_SEQUENCE), new MutableObject<>(new VariableReferenceExpression(resVar)))));
        assignCollectionToSequence.getInputs().add(new MutableObject<>(topOp.getInputs().get(0).getValue()));
        topOp.getInputs().get(0).setValue(assignCollectionToSequence);
        ProjectOperator projectOperator = (ProjectOperator) topOp;
        projectOperator.getVariables().set(0, seqVar);
        resVar = seqVar;
        DatasetDataSource targetDatasource = validateDatasetInfo(metadataProvider, stmt.getDataverseName(), stmt.getDatasetName());
        List<Integer> keySourceIndicator = ((InternalDatasetDetails) targetDatasource.getDataset().getDatasetDetails()).getKeySourceIndicator();
        ArrayList<LogicalVariable> vars = new ArrayList<>();
        ArrayList<Mutable<ILogicalExpression>> exprs = new ArrayList<>();
        List<Mutable<ILogicalExpression>> varRefsForLoading = new ArrayList<>();
        List<List<String>> partitionKeys = targetDatasource.getDataset().getPrimaryKeys();
        int numOfPrimaryKeys = partitionKeys.size();
        for (int i = 0; i < numOfPrimaryKeys; i++) {
            if (keySourceIndicator == null || keySourceIndicator.get(i).intValue() == 0) {
                // record part
                PlanTranslationUtil.prepareVarAndExpression(partitionKeys.get(i), resVar, vars, exprs, varRefsForLoading, context);
            } else {
                // meta part
                PlanTranslationUtil.prepareMetaKeyAccessExpression(partitionKeys.get(i), unnestVar, exprs, vars, varRefsForLoading, context);
            }
        }
        AssignOperator assign = new AssignOperator(vars, exprs);
        List<String> additionalFilteringField = DatasetUtil.getFilterField(targetDatasource.getDataset());
        List<LogicalVariable> additionalFilteringVars;
        List<Mutable<ILogicalExpression>> additionalFilteringAssignExpressions;
        List<Mutable<ILogicalExpression>> additionalFilteringExpressions = null;
        AssignOperator additionalFilteringAssign = null;
        if (additionalFilteringField != null) {
            additionalFilteringVars = new ArrayList<>();
            additionalFilteringAssignExpressions = new ArrayList<>();
            additionalFilteringExpressions = new ArrayList<>();
            PlanTranslationUtil.prepareVarAndExpression(additionalFilteringField, resVar, additionalFilteringVars, additionalFilteringAssignExpressions, additionalFilteringExpressions, context);
            additionalFilteringAssign = new AssignOperator(additionalFilteringVars, additionalFilteringAssignExpressions);
            additionalFilteringAssign.getInputs().add(new MutableObject<>(topOp));
            assign.getInputs().add(new MutableObject<>(additionalFilteringAssign));
        } else {
            assign.getInputs().add(new MutableObject<>(topOp));
        }
        Mutable<ILogicalExpression> varRef = new MutableObject<>(new VariableReferenceExpression(resVar));
        ILogicalOperator leafOperator;
        switch(stmt.getKind()) {
            case Statement.Kind.INSERT:
                leafOperator = translateInsert(targetDatasource, varRef, varRefsForLoading, additionalFilteringExpressions, assign, stmt);
                break;
            case Statement.Kind.UPSERT:
                leafOperator = translateUpsert(targetDatasource, varRef, varRefsForLoading, additionalFilteringExpressions, assign, additionalFilteringField, unnestVar, topOp, exprs, resVar, additionalFilteringAssign, stmt);
                break;
            case Statement.Kind.DELETE:
                leafOperator = translateDelete(targetDatasource, varRef, varRefsForLoading, additionalFilteringExpressions, assign);
                break;
            case Statement.Kind.CONNECT_FEED:
                leafOperator = translateConnectFeed(targetDatasource, varRef, varRefsForLoading, additionalFilteringExpressions, assign);
                break;
            case Statement.Kind.SUBSCRIBE_FEED:
                leafOperator = translateSubscribeFeed((CompiledSubscribeFeedStatement) stmt, targetDatasource, unnestVar, topOp, exprs, resVar, varRefsForLoading, varRef, assign, additionalFilteringField, additionalFilteringAssign, additionalFilteringExpressions);
                break;
            default:
                throw new AlgebricksException("Unsupported statement kind " + stmt.getKind());
        }
        topOp = leafOperator;
    }
    globalPlanRoots.add(new MutableObject<>(topOp));
    ILogicalPlan plan = new ALogicalPlanImpl(globalPlanRoots);
    eliminateSharedOperatorReferenceForPlan(plan);
    return plan;
}

Example 2 with FileSplit

use of org.apache.hyracks.api.io.FileSplit in project asterixdb by apache.

the class AbstractRTreeOperatorTest method insertPipeline.

protected void insertPipeline() throws Exception {
    JobSpecification spec = new JobSpecification();
    FileSplit[] ordersSplits = new FileSplit[] { new ManagedFileSplit(NC1_ID, "data" + File.separator + "orders-with-locations-part2.txt") };
    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(), DoubleSerializerDeserializer.INSTANCE, DoubleSerializerDeserializer.INSTANCE, DoubleSerializerDeserializer.INSTANCE, DoubleSerializerDeserializer.INSTANCE });
    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, DoubleParserFactory.INSTANCE, DoubleParserFactory.INSTANCE, DoubleParserFactory.INSTANCE, DoubleParserFactory.INSTANCE }, '|'), ordersDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, ordScanner, NC1_ID);
    // insert into primary index
    int[] primaryFieldPermutation = { 0, 1, 2, 4, 5, 7, 9, 10, 11, 12 };
    TreeIndexInsertUpdateDeleteOperatorDescriptor primaryInsertOp = new TreeIndexInsertUpdateDeleteOperatorDescriptor(spec, ordersDesc, primaryFieldPermutation, IndexOperation.INSERT, primaryHelperFactory, null, NoOpOperationCallbackFactory.INSTANCE);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, primaryInsertOp, NC1_ID);
    // secondary index
    int[] secondaryFieldPermutation = { 9, 10, 11, 12, 0 };
    TreeIndexInsertUpdateDeleteOperatorDescriptor secondaryInsertOp = new TreeIndexInsertUpdateDeleteOperatorDescriptor(spec, ordersDesc, secondaryFieldPermutation, IndexOperation.INSERT, secondaryHelperFactory, null, NoOpOperationCallbackFactory.INSTANCE);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, secondaryInsertOp, NC1_ID);
    NullSinkOperatorDescriptor nullSink = new NullSinkOperatorDescriptor(spec);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, nullSink, NC1_ID);
    spec.connect(new OneToOneConnectorDescriptor(spec), ordScanner, 0, primaryInsertOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), primaryInsertOp, 0, secondaryInsertOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), secondaryInsertOp, 0, nullSink, 0);
    spec.addRoot(nullSink);
    runTest(spec);
}

Example 3 with FileSplit

use of org.apache.hyracks.api.io.FileSplit in project asterixdb by apache.

the class SleepOperatorDescriptor method jobWithSleepOp.

private JobSpecification jobWithSleepOp() {
    JobSpecification spec = new JobSpecification();
    FileSplit[] ordersSplits = new FileSplit[] { new ManagedFileSplit(ASTERIX_IDS[0], "data" + File.separator + "tpch0.001" + File.separator + "orders-part1.tbl") };
    IFileSplitProvider ordersSplitsProvider = new ConstantFileSplitProvider(ordersSplits);
    RecordDescriptor recordDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() });
    // File scan operator.
    FileScanOperatorDescriptor scanOp = 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 }, '|'), recordDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, scanOp, ASTERIX_IDS[0]);
    // Sleep operator.
    SleepOperatorDescriptor sleepOp = new SleepOperatorDescriptor(spec);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, sleepOp, ASTERIX_IDS);
    // Sink operator.
    SinkOperatorDescriptor sinkOp = new SinkOperatorDescriptor(spec, 1);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, sinkOp, ASTERIX_IDS);
    // Hash-repartitioning connector.
    IConnectorDescriptor conn1 = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 0 }, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }));
    spec.connect(conn1, scanOp, 0, sleepOp, 0);
    // One-to-one connector.
    IConnectorDescriptor conn2 = new OneToOneConnectorDescriptor(spec);
    spec.connect(conn2, sleepOp, 0, sinkOp, 0);
    return spec;
}

Example 4 with FileSplit

use of org.apache.hyracks.api.io.FileSplit in project asterixdb by apache.

the class HeapSortMergeTest method optimizedSortMergeTest01.

@Test
public void optimizedSortMergeTest01() 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);
    // larger than the total record numbers.
    int outputLimit = 5;
    TopKSorterOperatorDescriptor sorter = new TopKSorterOperatorDescriptor(spec, 4, outputLimit, new int[] { 1, 0 }, null, 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);
    FileSplit fs = createFile(nc1);
    IFileSplitProvider outputSplitProvider = new ConstantFileSplitProvider(new FileSplit[] { fs });
    IOperatorDescriptor printer = new PlainFileWriterOperatorDescriptor(spec, outputSplitProvider, "|");
    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);
    System.out.println("Result write into :" + fs.getPath() + " in node: " + fs.getNodeName());
}

Example 5 with FileSplit

use of org.apache.hyracks.api.io.FileSplit in project asterixdb by apache.

the class HeapSortMergeTest method createSortMergeJobSpec.

public static JobSpecification createSortMergeJobSpec() 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);
    int outputLimit = 20;
    TopKSorterOperatorDescriptor sorter = new TopKSorterOperatorDescriptor(spec, 4, outputLimit, new int[] { 1, 0 }, null, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY), PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, ordersDesc);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, sorter, NC1_ID, NC2_ID);
    LimitOperatorDescriptor filter = new LimitOperatorDescriptor(spec, ordersDesc, outputLimit);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, filter, NC1_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, filter, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), filter, 0, printer, 0);
    return spec;
}