Search in sources :

Example 11 with IExpressionRuntimeProvider

use of org.apache.hyracks.algebricks.core.algebra.expressions.IExpressionRuntimeProvider in project asterixdb by apache.

the class ExternalGroupByPOperator 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("pre-sorted 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("External 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("External 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();
    ISerializedAggregateEvaluatorFactory[] aff = new ISerializedAggregateEvaluatorFactory[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.createSerializableAggregateFunctionFactory(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[] comparatorFactories = JobGenHelper.variablesToAscBinaryComparatorFactories(gbyCols, aggOpInputEnv, context);
    RecordDescriptor recordDescriptor = JobGenHelper.mkRecordDescriptor(context.getTypeEnvironment(op), opSchema, context);
    IBinaryHashFunctionFamily[] hashFunctionFactories = JobGenHelper.variablesToBinaryHashFunctionFamilies(gbyCols, aggOpInputEnv, context);
    ISerializedAggregateEvaluatorFactory[] merges = new ISerializedAggregateEvaluatorFactory[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.createSerializableAggregateFunctionFactory(mergeFun, aggOpInputEnv, localInputSchemas, context);
    }
    IAggregatorDescriptorFactory aggregatorFactory = new SerializableAggregatorDescriptorFactory(aff);
    IAggregatorDescriptorFactory mergeFactory = new SerializableAggregatorDescriptorFactory(merges);
    INormalizedKeyComputerFactory normalizedKeyFactory = JobGenHelper.variablesToAscNormalizedKeyComputerFactory(gbyCols, aggOpInputEnv, context);
    // Calculates the hash table size (# of unique hash values) based on the budget and a tuple size.
    int memoryBudgetInBytes = context.getFrameSize() * frameLimit;
    int groupByColumnsCount = gby.getGroupByList().size() + numFds;
    int hashTableSize = ExternalGroupOperatorDescriptor.calculateGroupByTableCardinality(memoryBudgetInBytes, groupByColumnsCount, context.getFrameSize());
    ExternalGroupOperatorDescriptor gbyOpDesc = new ExternalGroupOperatorDescriptor(spec, hashTableSize, inputSize, keyAndDecFields, frameLimit, comparatorFactories, normalizedKeyFactory, aggregatorFactory, mergeFactory, recordDescriptor, recordDescriptor, new HashSpillableTableFactory(hashFunctionFactories));
    contributeOpDesc(builder, gby, gbyOpDesc);
    ILogicalOperator src = op.getInputs().get(0).getValue();
    builder.contributeGraphEdge(src, 0, op, 0);
}
Also used : RecordDescriptor(org.apache.hyracks.api.dataflow.value.RecordDescriptor) IOperatorSchema(org.apache.hyracks.algebricks.core.algebra.operators.logical.IOperatorSchema) HashSpillableTableFactory(org.apache.hyracks.dataflow.std.group.HashSpillableTableFactory) ArrayList(java.util.ArrayList) ISerializedAggregateEvaluatorFactory(org.apache.hyracks.algebricks.runtime.base.ISerializedAggregateEvaluatorFactory) IExpressionRuntimeProvider(org.apache.hyracks.algebricks.core.algebra.expressions.IExpressionRuntimeProvider) AggregateOperator(org.apache.hyracks.algebricks.core.algebra.operators.logical.AggregateOperator) IPartialAggregationTypeComputer(org.apache.hyracks.algebricks.core.algebra.expressions.IPartialAggregationTypeComputer) IBinaryHashFunctionFamily(org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFamily) LogicalVariable(org.apache.hyracks.algebricks.core.algebra.base.LogicalVariable) AggregateFunctionCallExpression(org.apache.hyracks.algebricks.core.algebra.expressions.AggregateFunctionCallExpression) GroupByOperator(org.apache.hyracks.algebricks.core.algebra.operators.logical.GroupByOperator) SerializableAggregatorDescriptorFactory(org.apache.hyracks.algebricks.runtime.operators.aggreg.SerializableAggregatorDescriptorFactory) ExternalGroupOperatorDescriptor(org.apache.hyracks.dataflow.std.group.external.ExternalGroupOperatorDescriptor) ILogicalOperator(org.apache.hyracks.algebricks.core.algebra.base.ILogicalOperator) AlgebricksException(org.apache.hyracks.algebricks.common.exceptions.AlgebricksException) IBinaryComparatorFactory(org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory) IOperatorDescriptorRegistry(org.apache.hyracks.api.job.IOperatorDescriptorRegistry) OperatorSchemaImpl(org.apache.hyracks.algebricks.core.jobgen.impl.OperatorSchemaImpl) IAggregatorDescriptorFactory(org.apache.hyracks.dataflow.std.group.IAggregatorDescriptorFactory) Mutable(org.apache.commons.lang3.mutable.Mutable) ILogicalExpression(org.apache.hyracks.algebricks.core.algebra.base.ILogicalExpression) INormalizedKeyComputerFactory(org.apache.hyracks.api.dataflow.value.INormalizedKeyComputerFactory) VariableReferenceExpression(org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression) ILogicalPlan(org.apache.hyracks.algebricks.core.algebra.base.ILogicalPlan) IVariableTypeEnvironment(org.apache.hyracks.algebricks.core.algebra.expressions.IVariableTypeEnvironment)

Aggregations

IExpressionRuntimeProvider (org.apache.hyracks.algebricks.core.algebra.expressions.IExpressionRuntimeProvider)11 ILogicalOperator (org.apache.hyracks.algebricks.core.algebra.base.ILogicalOperator)10 RecordDescriptor (org.apache.hyracks.api.dataflow.value.RecordDescriptor)10 IScalarEvaluatorFactory (org.apache.hyracks.algebricks.runtime.base.IScalarEvaluatorFactory)6 Mutable (org.apache.commons.lang3.mutable.Mutable)5 LogicalVariable (org.apache.hyracks.algebricks.core.algebra.base.LogicalVariable)5 ILogicalExpression (org.apache.hyracks.algebricks.core.algebra.base.ILogicalExpression)4 IOperatorDescriptorRegistry (org.apache.hyracks.api.job.IOperatorDescriptorRegistry)4 AlgebricksException (org.apache.hyracks.algebricks.common.exceptions.AlgebricksException)3 AggregateFunctionCallExpression (org.apache.hyracks.algebricks.core.algebra.expressions.AggregateFunctionCallExpression)3 IVariableTypeEnvironment (org.apache.hyracks.algebricks.core.algebra.expressions.IVariableTypeEnvironment)3 AggregateOperator (org.apache.hyracks.algebricks.core.algebra.operators.logical.AggregateOperator)3 IOperatorSchema (org.apache.hyracks.algebricks.core.algebra.operators.logical.IOperatorSchema)3 ArrayList (java.util.ArrayList)2 ILogicalPlan (org.apache.hyracks.algebricks.core.algebra.base.ILogicalPlan)2 IPartialAggregationTypeComputer (org.apache.hyracks.algebricks.core.algebra.expressions.IPartialAggregationTypeComputer)2 VariableReferenceExpression (org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression)2 GroupByOperator (org.apache.hyracks.algebricks.core.algebra.operators.logical.GroupByOperator)2 OperatorSchemaImpl (org.apache.hyracks.algebricks.core.jobgen.impl.OperatorSchemaImpl)2 IAggregateEvaluatorFactory (org.apache.hyracks.algebricks.runtime.base.IAggregateEvaluatorFactory)2