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

Example 91 with RecordDescriptor

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

the class ConnectorDescriptorWithMessagingTest method testMessageLargerThanEmptyFrame.

@Test
public void testMessageLargerThanEmptyFrame() throws Exception {
    try {
        List<Integer> routing = Arrays.asList(0, 1, 2, 3, 4);
        IConnectorDescriptorRegistry connDescRegistry = Mockito.mock(IConnectorDescriptorRegistry.class);
        ITuplePartitionComputerFactory partitionComputerFactory = new TestPartitionComputerFactory(routing);
        MToNPartitioningWithMessageConnectorDescriptor connector = new MToNPartitioningWithMessageConnectorDescriptor(connDescRegistry, partitionComputerFactory);
        IHyracksTaskContext ctx = TestUtils.create(DEFAULT_FRAME_SIZE);
        VSizeFrame message = new VSizeFrame(ctx);
        VSizeFrame tempBuffer = new VSizeFrame(ctx);
        TaskUtil.putInSharedMap(HyracksConstants.KEY_MESSAGE, message, ctx);
        writeRandomMessage(message, MessagingFrameTupleAppender.MARKER_MESSAGE, DEFAULT_FRAME_SIZE + 1);
        ISerializerDeserializer<?>[] serdes = new ISerializerDeserializer<?>[] { Integer64SerializerDeserializer.INSTANCE, DoubleSerializerDeserializer.INSTANCE, BooleanSerializerDeserializer.INSTANCE, new UTF8StringSerializerDeserializer() };
        RecordDescriptor rDesc = new RecordDescriptor(serdes);
        TestPartitionWriterFactory partitionWriterFactory = new TestPartitionWriterFactory();
        IFrameWriter partitioner = connector.createPartitioner(ctx, rDesc, partitionWriterFactory, CURRENT_PRODUCER, NUMBER_OF_CONSUMERS, NUMBER_OF_CONSUMERS);
        partitioner.open();
        FrameTupleAccessor fta = new FrameTupleAccessor(rDesc);
        List<TestFrameWriter> recipients = new ArrayList<>();
        for (IFrameWriter writer : partitionWriterFactory.getWriters().values()) {
            recipients.add((TestFrameWriter) writer);
        }
        partitioner.flush();
        for (TestFrameWriter writer : recipients) {
            Assert.assertEquals(writer.nextFrameCount(), 1);
            fta.reset(writer.getLastFrame());
            Assert.assertEquals(fta.getTupleCount(), 1);
            FeedUtils.processFeedMessage(writer.getLastFrame(), tempBuffer, fta);
            Assert.assertEquals(MessagingFrameTupleAppender.MARKER_MESSAGE, MessagingFrameTupleAppender.getMessageType(tempBuffer));
        }
        message.getBuffer().clear();
        message.getBuffer().put(MessagingFrameTupleAppender.ACK_REQ_FEED_MESSAGE);
        message.getBuffer().flip();
        partitioner.flush();
        for (TestFrameWriter writer : recipients) {
            Assert.assertEquals(writer.nextFrameCount(), 2);
            fta.reset(writer.getLastFrame());
            Assert.assertEquals(fta.getTupleCount(), 1);
            FeedUtils.processFeedMessage(writer.getLastFrame(), tempBuffer, fta);
            Assert.assertEquals(MessagingFrameTupleAppender.ACK_REQ_FEED_MESSAGE, MessagingFrameTupleAppender.getMessageType(tempBuffer));
        }
        message.getBuffer().clear();
        message.getBuffer().put(MessagingFrameTupleAppender.NULL_FEED_MESSAGE);
        message.getBuffer().flip();
        partitioner.flush();
        for (TestFrameWriter writer : recipients) {
            Assert.assertEquals(writer.nextFrameCount(), 3);
            fta.reset(writer.getLastFrame());
            Assert.assertEquals(fta.getTupleCount(), 1);
            FeedUtils.processFeedMessage(writer.getLastFrame(), tempBuffer, fta);
            Assert.assertEquals(MessagingFrameTupleAppender.NULL_FEED_MESSAGE, MessagingFrameTupleAppender.getMessageType(tempBuffer));
        }
        partitioner.close();
        for (TestFrameWriter writer : recipients) {
            Assert.assertEquals(writer.nextFrameCount(), 4);
            Assert.assertEquals(writer.closeCount(), 1);
        }
    } catch (Throwable th) {
        th.printStackTrace();
        throw th;
    }
}

Example 92 with RecordDescriptor

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

the class ConnectorDescriptorWithMessagingTest method testMessageLargerThanSome.

@Test
public void testMessageLargerThanSome() throws Exception {
    try {
        // Routing will be to 1, 3, and 4 only. 0 and 2 will receive no tuples
        List<Integer> routing = Arrays.asList(1, 3, 4);
        IConnectorDescriptorRegistry connDescRegistry = Mockito.mock(IConnectorDescriptorRegistry.class);
        ITuplePartitionComputerFactory partitionComputerFactory = new TestPartitionComputerFactory(routing);
        MToNPartitioningWithMessageConnectorDescriptor connector = new MToNPartitioningWithMessageConnectorDescriptor(connDescRegistry, partitionComputerFactory);
        IHyracksTaskContext ctx = TestUtils.create(DEFAULT_FRAME_SIZE);
        VSizeFrame message = new VSizeFrame(ctx);
        VSizeFrame tempBuffer = new VSizeFrame(ctx);
        TaskUtil.putInSharedMap(HyracksConstants.KEY_MESSAGE, message, ctx);
        message.getBuffer().clear();
        writeRandomMessage(message, MessagingFrameTupleAppender.MARKER_MESSAGE, DEFAULT_FRAME_SIZE);
        ISerializerDeserializer<?>[] serdes = new ISerializerDeserializer<?>[] { Integer64SerializerDeserializer.INSTANCE, DoubleSerializerDeserializer.INSTANCE, BooleanSerializerDeserializer.INSTANCE, new UTF8StringSerializerDeserializer() };
        FieldType[] types = { FieldType.Integer64, FieldType.Double, FieldType.Boolean, FieldType.String };
        RecordDescriptor rDesc = new RecordDescriptor(serdes);
        TestPartitionWriterFactory partitionWriterFactory = new TestPartitionWriterFactory();
        PartitionWithMessageDataWriter partitioner = (PartitionWithMessageDataWriter) connector.createPartitioner(ctx, rDesc, partitionWriterFactory, CURRENT_PRODUCER, NUMBER_OF_CONSUMERS, NUMBER_OF_CONSUMERS);
        partitioner.open();
        FrameTupleAccessor fta = new FrameTupleAccessor(rDesc);
        List<TestFrameWriter> recipients = new ArrayList<>();
        for (int i = 0; i < partitionWriterFactory.getWriters().values().size(); i++) {
            recipients.add(partitionWriterFactory.getWriters().get(i));
        }
        TestTupleGenerator ttg = new TestTupleGenerator(types, STRING_FIELD_SIZES, true);
        VSizeFrame frame = new VSizeFrame(ctx);
        FrameTupleAppender appender = new FrameTupleAppender(frame);
        ITupleReference tuple = ttg.next();
        while (appender.append(tuple)) {
            tuple = ttg.next();
        }
        partitioner.nextFrame(frame.getBuffer());
        partitioner.flush();
        Assert.assertEquals(1, partitionWriterFactory.getWriters().get(0).nextFrameCount());
        Assert.assertEquals(2, partitionWriterFactory.getWriters().get(1).nextFrameCount());
        Assert.assertEquals(1, partitionWriterFactory.getWriters().get(2).nextFrameCount());
        Assert.assertEquals(2, partitionWriterFactory.getWriters().get(3).nextFrameCount());
        Assert.assertEquals(2, partitionWriterFactory.getWriters().get(4).nextFrameCount());
        for (TestFrameWriter writer : recipients) {
            fta.reset(writer.getLastFrame());
            Assert.assertEquals(fta.getTupleCount(), 1);
            FeedUtils.processFeedMessage(writer.getLastFrame(), tempBuffer, fta);
            Assert.assertEquals(MessagingFrameTupleAppender.MARKER_MESSAGE, MessagingFrameTupleAppender.getMessageType(tempBuffer));
        }
        partitioner.close();
    } catch (Throwable th) {
        th.printStackTrace();
        throw th;
    }
}

Example 93 with RecordDescriptor

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

the class SecondaryRTreeOperationsHelper method setSecondaryRecDescAndComparators.

@Override
protected void setSecondaryRecDescAndComparators() throws AlgebricksException {
    List<List<String>> secondaryKeyFields = index.getKeyFieldNames();
    int numSecondaryKeys = secondaryKeyFields.size();
    boolean isEnforcingKeyTypes = index.isEnforcingKeyFileds();
    if (numSecondaryKeys != 1) {
        throw new AsterixException("Cannot use " + numSecondaryKeys + " fields as a key for the R-tree index. " + "There can be only one field as a key for the R-tree index.");
    }
    Pair<IAType, Boolean> spatialTypePair = Index.getNonNullableOpenFieldType(index.getKeyFieldTypes().get(0), secondaryKeyFields.get(0), itemType);
    IAType spatialType = spatialTypePair.first;
    anySecondaryKeyIsNullable = spatialTypePair.second;
    if (spatialType == null) {
        throw new AsterixException("Could not find field " + secondaryKeyFields.get(0) + " in the schema.");
    }
    isPointMBR = spatialType.getTypeTag() == ATypeTag.POINT || spatialType.getTypeTag() == ATypeTag.POINT3D;
    int numDimensions = NonTaggedFormatUtil.getNumDimensions(spatialType.getTypeTag());
    numNestedSecondaryKeyFields = numDimensions * 2;
    int recordColumn = dataset.getDatasetType() == DatasetType.INTERNAL ? numPrimaryKeys : 0;
    secondaryFieldAccessEvalFactories = metadataProvider.getFormat().createMBRFactory(isEnforcingKeyTypes ? enforcedItemType : itemType, secondaryKeyFields.get(0), recordColumn, numDimensions, filterFieldName, isPointMBR);
    secondaryComparatorFactories = new IBinaryComparatorFactory[numNestedSecondaryKeyFields];
    valueProviderFactories = new IPrimitiveValueProviderFactory[numNestedSecondaryKeyFields];
    ISerializerDeserializer[] secondaryRecFields = new ISerializerDeserializer[numPrimaryKeys + numNestedSecondaryKeyFields + numFilterFields];
    ISerializerDeserializer[] enforcedRecFields = new ISerializerDeserializer[1 + numPrimaryKeys + numFilterFields];
    secondaryTypeTraits = new ITypeTraits[numNestedSecondaryKeyFields + numPrimaryKeys];
    ITypeTraits[] enforcedTypeTraits = new ITypeTraits[1 + numPrimaryKeys];
    IAType nestedKeyType = NonTaggedFormatUtil.getNestedSpatialType(spatialType.getTypeTag());
    keyType = nestedKeyType.getTypeTag();
    for (int i = 0; i < numNestedSecondaryKeyFields; i++) {
        ISerializerDeserializer keySerde = SerializerDeserializerProvider.INSTANCE.getSerializerDeserializer(nestedKeyType);
        secondaryRecFields[i] = keySerde;
        secondaryComparatorFactories[i] = BinaryComparatorFactoryProvider.INSTANCE.getBinaryComparatorFactory(nestedKeyType, true);
        secondaryTypeTraits[i] = TypeTraitProvider.INSTANCE.getTypeTrait(nestedKeyType);
        valueProviderFactories[i] = metadataProvider.getStorageComponentProvider().getPrimitiveValueProviderFactory();
    }
    // Add serializers and comparators for primary index fields.
    if (dataset.getDatasetType() == DatasetType.INTERNAL) {
        for (int i = 0; i < numPrimaryKeys; i++) {
            secondaryRecFields[numNestedSecondaryKeyFields + i] = primaryRecDesc.getFields()[i];
            secondaryTypeTraits[numNestedSecondaryKeyFields + i] = primaryRecDesc.getTypeTraits()[i];
            enforcedRecFields[i] = primaryRecDesc.getFields()[i];
            enforcedTypeTraits[i] = primaryRecDesc.getTypeTraits()[i];
        }
    } else {
        for (int i = 0; i < numPrimaryKeys; i++) {
            secondaryRecFields[numNestedSecondaryKeyFields + i] = IndexingConstants.getSerializerDeserializer(i);
            secondaryTypeTraits[numNestedSecondaryKeyFields + i] = IndexingConstants.getTypeTraits(i);
            enforcedRecFields[i] = IndexingConstants.getSerializerDeserializer(i);
            enforcedTypeTraits[i] = IndexingConstants.getTypeTraits(i);
        }
    }
    enforcedRecFields[numPrimaryKeys] = SerializerDeserializerProvider.INSTANCE.getSerializerDeserializer(itemType);
    enforcedRecDesc = new RecordDescriptor(enforcedRecFields, enforcedTypeTraits);
    if (numFilterFields > 0) {
        rtreeFields = new int[numNestedSecondaryKeyFields + numPrimaryKeys];
        for (int i = 0; i < rtreeFields.length; i++) {
            rtreeFields[i] = i;
        }
        Pair<IAType, Boolean> typePair = Index.getNonNullableKeyFieldType(filterFieldName, itemType);
        IAType type = typePair.first;
        ISerializerDeserializer serde = SerializerDeserializerProvider.INSTANCE.getSerializerDeserializer(type);
        secondaryRecFields[numPrimaryKeys + numNestedSecondaryKeyFields] = serde;
    }
    secondaryRecDesc = new RecordDescriptor(secondaryRecFields);
    primaryKeyFields = new int[numPrimaryKeys];
    for (int i = 0; i < primaryKeyFields.length; i++) {
        primaryKeyFields[i] = i + numNestedSecondaryKeyFields;
    }
    if (isPointMBR) {
        int numNestedSecondaryKeyFieldForPointMBR = numNestedSecondaryKeyFields / 2;
        ISerializerDeserializer[] recFieldsForPointMBR = new ISerializerDeserializer[numPrimaryKeys + numNestedSecondaryKeyFieldForPointMBR + numFilterFields];
        int idx = 0;
        for (int i = 0; i < numNestedSecondaryKeyFieldForPointMBR; i++) {
            recFieldsForPointMBR[idx++] = secondaryRecFields[i];
        }
        for (int i = 0; i < numPrimaryKeys + numFilterFields; i++) {
            recFieldsForPointMBR[idx++] = secondaryRecFields[numNestedSecondaryKeyFields + i];
        }
        secondaryRecDescForPointMBR = new RecordDescriptor(recFieldsForPointMBR);
    }
}

Example 94 with RecordDescriptor

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

the class SortGroupByPOperator method contributeRuntimeOperator.

@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema opSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
    List<LogicalVariable> gbyCols = getGbyColumns();
    int[] keys = JobGenHelper.variablesToFieldIndexes(gbyCols, inputSchemas[0]);
    GroupByOperator gby = (GroupByOperator) op;
    int numFds = gby.getDecorList().size();
    int[] fdColumns = new int[numFds];
    int j = 0;
    for (Pair<LogicalVariable, Mutable<ILogicalExpression>> p : gby.getDecorList()) {
        ILogicalExpression expr = p.second.getValue();
        if (expr.getExpressionTag() != LogicalExpressionTag.VARIABLE) {
            throw new AlgebricksException("Sort group-by expects variable references.");
        }
        VariableReferenceExpression v = (VariableReferenceExpression) expr;
        LogicalVariable decor = v.getVariableReference();
        fdColumns[j++] = inputSchemas[0].findVariable(decor);
    }
    if (gby.getNestedPlans().size() != 1) {
        throw new AlgebricksException("Sort group-by currently works only for one nested plan with one root containing" + "an aggregate and a nested-tuple-source.");
    }
    ILogicalPlan p0 = gby.getNestedPlans().get(0);
    if (p0.getRoots().size() != 1) {
        throw new AlgebricksException("Sort group-by currently works only for one nested plan with one root containing" + "an aggregate and a nested-tuple-source.");
    }
    Mutable<ILogicalOperator> r0 = p0.getRoots().get(0);
    AggregateOperator aggOp = (AggregateOperator) r0.getValue();
    IPartialAggregationTypeComputer partialAggregationTypeComputer = context.getPartialAggregationTypeComputer();
    List<Object> intermediateTypes = new ArrayList<Object>();
    int n = aggOp.getExpressions().size();
    IAggregateEvaluatorFactory[] aff = new IAggregateEvaluatorFactory[n];
    int i = 0;
    IExpressionRuntimeProvider expressionRuntimeProvider = context.getExpressionRuntimeProvider();
    IVariableTypeEnvironment aggOpInputEnv = context.getTypeEnvironment(aggOp.getInputs().get(0).getValue());
    IVariableTypeEnvironment outputEnv = context.getTypeEnvironment(op);
    for (Mutable<ILogicalExpression> exprRef : aggOp.getExpressions()) {
        AggregateFunctionCallExpression aggFun = (AggregateFunctionCallExpression) exprRef.getValue();
        aff[i++] = expressionRuntimeProvider.createAggregateFunctionFactory(aggFun, aggOpInputEnv, inputSchemas, context);
        intermediateTypes.add(partialAggregationTypeComputer.getType(aggFun, aggOpInputEnv, context.getMetadataProvider()));
    }
    int[] keyAndDecFields = new int[keys.length + fdColumns.length];
    for (i = 0; i < keys.length; ++i) {
        keyAndDecFields[i] = keys[i];
    }
    for (i = 0; i < fdColumns.length; i++) {
        keyAndDecFields[keys.length + i] = fdColumns[i];
    }
    List<LogicalVariable> keyAndDecVariables = new ArrayList<LogicalVariable>();
    for (Pair<LogicalVariable, Mutable<ILogicalExpression>> p : gby.getGroupByList()) {
        keyAndDecVariables.add(p.first);
    }
    for (Pair<LogicalVariable, Mutable<ILogicalExpression>> p : gby.getDecorList()) {
        keyAndDecVariables.add(GroupByOperator.getDecorVariable(p));
    }
    for (LogicalVariable var : keyAndDecVariables) {
        aggOpInputEnv.setVarType(var, outputEnv.getVarType(var));
    }
    compileSubplans(inputSchemas[0], gby, opSchema, context);
    IOperatorDescriptorRegistry spec = builder.getJobSpec();
    IBinaryComparatorFactory[] compFactories = new IBinaryComparatorFactory[gbyCols.size()];
    IBinaryComparatorFactoryProvider bcfProvider = context.getBinaryComparatorFactoryProvider();
    i = 0;
    for (LogicalVariable v : gbyCols) {
        Object type = aggOpInputEnv.getVarType(v);
        if (orderColumns[i].getOrder() == OrderKind.ASC) {
            compFactories[i] = bcfProvider.getBinaryComparatorFactory(type, true);
        } else {
            compFactories[i] = bcfProvider.getBinaryComparatorFactory(type, false);
        }
        i++;
    }
    RecordDescriptor recordDescriptor = JobGenHelper.mkRecordDescriptor(context.getTypeEnvironment(op), opSchema, context);
    IAggregateEvaluatorFactory[] merges = new IAggregateEvaluatorFactory[n];
    List<LogicalVariable> usedVars = new ArrayList<LogicalVariable>();
    IOperatorSchema[] localInputSchemas = new IOperatorSchema[1];
    localInputSchemas[0] = new OperatorSchemaImpl();
    for (i = 0; i < n; i++) {
        AggregateFunctionCallExpression aggFun = (AggregateFunctionCallExpression) aggOp.getMergeExpressions().get(i).getValue();
        aggFun.getUsedVariables(usedVars);
    }
    i = 0;
    for (Object type : intermediateTypes) {
        aggOpInputEnv.setVarType(usedVars.get(i++), type);
    }
    for (LogicalVariable keyVar : keyAndDecVariables) {
        localInputSchemas[0].addVariable(keyVar);
    }
    for (LogicalVariable usedVar : usedVars) {
        localInputSchemas[0].addVariable(usedVar);
    }
    for (i = 0; i < n; i++) {
        AggregateFunctionCallExpression mergeFun = (AggregateFunctionCallExpression) aggOp.getMergeExpressions().get(i).getValue();
        merges[i] = expressionRuntimeProvider.createAggregateFunctionFactory(mergeFun, aggOpInputEnv, localInputSchemas, context);
    }
    RecordDescriptor partialAggRecordDescriptor = JobGenHelper.mkRecordDescriptor(context.getTypeEnvironment(op), localInputSchemas[0], context);
    IAggregatorDescriptorFactory aggregatorFactory = new SimpleAlgebricksAccumulatingAggregatorFactory(aff, keyAndDecFields);
    IAggregatorDescriptorFactory mergeFactory = new SimpleAlgebricksAccumulatingAggregatorFactory(merges, keyAndDecFields);
    INormalizedKeyComputerFactory normalizedKeyFactory = null;
    INormalizedKeyComputerFactoryProvider nkcfProvider = context.getNormalizedKeyComputerFactoryProvider();
    if (nkcfProvider == null) {
        normalizedKeyFactory = null;
    }
    Object type = aggOpInputEnv.getVarType(gbyCols.get(0));
    normalizedKeyFactory = orderColumns[0].getOrder() == OrderKind.ASC ? nkcfProvider.getNormalizedKeyComputerFactory(type, true) : nkcfProvider.getNormalizedKeyComputerFactory(type, false);
    SortGroupByOperatorDescriptor gbyOpDesc = new SortGroupByOperatorDescriptor(spec, frameLimit, keys, keyAndDecFields, normalizedKeyFactory, compFactories, aggregatorFactory, mergeFactory, partialAggRecordDescriptor, recordDescriptor, false);
    contributeOpDesc(builder, gby, gbyOpDesc);
    ILogicalOperator src = op.getInputs().get(0).getValue();
    builder.contributeGraphEdge(src, 0, op, 0);
}

Example 95 with RecordDescriptor

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

the class StreamSelectPOperator method contributeRuntimeOperator.

@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema opSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
    SelectOperator select = (SelectOperator) op;
    IExpressionRuntimeProvider expressionRuntimeProvider = context.getExpressionRuntimeProvider();
    IScalarEvaluatorFactory cond = expressionRuntimeProvider.createEvaluatorFactory(select.getCondition().getValue(), context.getTypeEnvironment(op), inputSchemas, context);
    StreamSelectRuntimeFactory runtime = new StreamSelectRuntimeFactory(cond, null, context.getBinaryBooleanInspectorFactory(), select.getRetainMissing(), inputSchemas[0].findVariable(select.getMissingPlaceholderVariable()), context.getMissingWriterFactory());
    // contribute one Asterix framewriter
    RecordDescriptor recDesc = JobGenHelper.mkRecordDescriptor(context.getTypeEnvironment(op), opSchema, context);
    builder.contributeMicroOperator(select, runtime, recDesc);
    // and contribute one edge from its child
    ILogicalOperator src = select.getInputs().get(0).getValue();
    builder.contributeGraphEdge(src, 0, select, 0);
}