Examples with DataSourceScanOperator org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator used on opensource projects

Example 6 with DataSourceScanOperator

use of org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator in project asterixdb by apache.

the class PushProjectIntoDataSourceScanRule method rewritePost.

@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context) throws AlgebricksException {
    AbstractLogicalOperator op = (AbstractLogicalOperator) opRef.getValue();
    if (op.getInputs().size() <= 0)
        return false;
    AbstractLogicalOperator project = (AbstractLogicalOperator) op.getInputs().get(0).getValue();
    if (project.getOperatorTag() != LogicalOperatorTag.PROJECT)
        return false;
    AbstractLogicalOperator exchange = (AbstractLogicalOperator) project.getInputs().get(0).getValue();
    if (exchange.getOperatorTag() != LogicalOperatorTag.EXCHANGE)
        return false;
    AbstractLogicalOperator inputOp = (AbstractLogicalOperator) exchange.getInputs().get(0).getValue();
    if (inputOp.getOperatorTag() != LogicalOperatorTag.DATASOURCESCAN)
        return false;
    DataSourceScanOperator scanOp = (DataSourceScanOperator) inputOp;
    ProjectOperator projectOp = (ProjectOperator) project;
    scanOp.addProjectVariables(projectOp.getVariables());
    if (op.getOperatorTag() != LogicalOperatorTag.EXCHANGE) {
        op.getInputs().set(0, project.getInputs().get(0));
    } else {
        op.getInputs().set(0, exchange.getInputs().get(0));
    }
    return true;
}

Example 7 with DataSourceScanOperator

use of org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator in project asterixdb by apache.

the class PushFieldAccessRule method propagateFieldAccessRec.

@SuppressWarnings("unchecked")
private boolean propagateFieldAccessRec(Mutable<ILogicalOperator> opRef, IOptimizationContext context, String finalAnnot) throws AlgebricksException {
    AssignOperator access = (AssignOperator) opRef.getValue();
    Mutable<ILogicalOperator> opRef2 = access.getInputs().get(0);
    AbstractLogicalOperator op2 = (AbstractLogicalOperator) opRef2.getValue();
    // rewritten into index search.
    if (op2.getOperatorTag() == LogicalOperatorTag.PROJECT || context.checkAndAddToAlreadyCompared(access, op2) && !(op2.getOperatorTag() == LogicalOperatorTag.SELECT && isAccessToIndexedField(access, context))) {
        return false;
    }
    Object annotation = op2.getAnnotations().get(IS_MOVABLE);
    if (annotation != null && !((Boolean) annotation)) {
        return false;
    }
    if (tryingToPushThroughSelectionWithSameDataSource(access, op2)) {
        return false;
    }
    if (testAndModifyRedundantOp(access, op2)) {
        propagateFieldAccessRec(opRef2, context, finalAnnot);
        return true;
    }
    List<LogicalVariable> usedInAccess = new LinkedList<>();
    VariableUtilities.getUsedVariables(access, usedInAccess);
    List<LogicalVariable> produced2 = new LinkedList<>();
    if (op2.getOperatorTag() == LogicalOperatorTag.GROUP) {
        VariableUtilities.getLiveVariables(op2, produced2);
    } else {
        VariableUtilities.getProducedVariables(op2, produced2);
    }
    boolean pushItDown = false;
    List<LogicalVariable> inter = new ArrayList<>(usedInAccess);
    if (inter.isEmpty()) {
        // ground value
        return false;
    }
    inter.retainAll(produced2);
    if (inter.isEmpty()) {
        pushItDown = true;
    } else if (op2.getOperatorTag() == LogicalOperatorTag.GROUP) {
        GroupByOperator g = (GroupByOperator) op2;
        List<Pair<LogicalVariable, LogicalVariable>> varMappings = new ArrayList<>();
        for (Pair<LogicalVariable, Mutable<ILogicalExpression>> p : g.getDecorList()) {
            ILogicalExpression e = p.second.getValue();
            if (e.getExpressionTag() == LogicalExpressionTag.VARIABLE) {
                LogicalVariable decorVar = GroupByOperator.getDecorVariable(p);
                if (inter.contains(decorVar)) {
                    inter.remove(decorVar);
                    LogicalVariable v1 = ((VariableReferenceExpression) e).getVariableReference();
                    varMappings.add(new Pair<>(decorVar, v1));
                }
            }
        }
        if (inter.isEmpty()) {
            boolean changed = false;
            for (Pair<LogicalVariable, LogicalVariable> m : varMappings) {
                LogicalVariable v2 = context.newVar();
                LogicalVariable oldVar = access.getVariables().get(0);
                g.getDecorList().add(new Pair<LogicalVariable, Mutable<ILogicalExpression>>(oldVar, new MutableObject<ILogicalExpression>(new VariableReferenceExpression(v2))));
                changed = true;
                access.getVariables().set(0, v2);
                VariableUtilities.substituteVariables(access, m.first, m.second, context);
            }
            if (changed) {
                context.computeAndSetTypeEnvironmentForOperator(g);
            }
            usedInAccess.clear();
            VariableUtilities.getUsedVariables(access, usedInAccess);
            pushItDown = true;
        }
    }
    if (pushItDown) {
        if (op2.getOperatorTag() == LogicalOperatorTag.NESTEDTUPLESOURCE) {
            Mutable<ILogicalOperator> childOfSubplan = ((NestedTupleSourceOperator) op2).getDataSourceReference().getValue().getInputs().get(0);
            pushAccessDown(opRef, op2, childOfSubplan, context, finalAnnot);
            return true;
        }
        if (op2.getInputs().size() == 1 && !op2.hasNestedPlans()) {
            pushAccessDown(opRef, op2, op2.getInputs().get(0), context, finalAnnot);
            return true;
        } else {
            for (Mutable<ILogicalOperator> inp : op2.getInputs()) {
                HashSet<LogicalVariable> v2 = new HashSet<>();
                VariableUtilities.getLiveVariables(inp.getValue(), v2);
                if (v2.containsAll(usedInAccess)) {
                    pushAccessDown(opRef, op2, inp, context, finalAnnot);
                    return true;
                }
            }
        }
        if (op2.hasNestedPlans()) {
            AbstractOperatorWithNestedPlans nestedOp = (AbstractOperatorWithNestedPlans) op2;
            for (ILogicalPlan plan : nestedOp.getNestedPlans()) {
                for (Mutable<ILogicalOperator> root : plan.getRoots()) {
                    HashSet<LogicalVariable> v2 = new HashSet<>();
                    VariableUtilities.getLiveVariables(root.getValue(), v2);
                    if (v2.containsAll(usedInAccess)) {
                        pushAccessDown(opRef, op2, root, context, finalAnnot);
                        return true;
                    }
                }
            }
        }
        throw new AlgebricksException("Field access " + access.getExpressions().get(0).getValue() + " does not correspond to any input of operator " + op2);
    } else {
        // fields. If yes, we can equate the two variables.
        if (op2.getOperatorTag() == LogicalOperatorTag.DATASOURCESCAN) {
            DataSourceScanOperator scan = (DataSourceScanOperator) op2;
            int n = scan.getVariables().size();
            LogicalVariable scanRecordVar = scan.getVariables().get(n - 1);
            AbstractFunctionCallExpression accessFun = (AbstractFunctionCallExpression) access.getExpressions().get(0).getValue();
            ILogicalExpression e0 = accessFun.getArguments().get(0).getValue();
            LogicalExpressionTag tag = e0.getExpressionTag();
            if (tag == LogicalExpressionTag.VARIABLE) {
                VariableReferenceExpression varRef = (VariableReferenceExpression) e0;
                if (varRef.getVariableReference() == scanRecordVar) {
                    ILogicalExpression e1 = accessFun.getArguments().get(1).getValue();
                    if (e1.getExpressionTag() == LogicalExpressionTag.CONSTANT) {
                        IDataSource<DataSourceId> dataSource = (IDataSource<DataSourceId>) scan.getDataSource();
                        byte dsType = ((DataSource) dataSource).getDatasourceType();
                        if (dsType == DataSource.Type.FEED || dsType == DataSource.Type.LOADABLE) {
                            return false;
                        }
                        DataSourceId asid = dataSource.getId();
                        MetadataProvider mp = (MetadataProvider) context.getMetadataProvider();
                        Dataset dataset = mp.findDataset(asid.getDataverseName(), asid.getDatasourceName());
                        if (dataset == null) {
                            throw new AlgebricksException("Dataset " + asid.getDatasourceName() + " not found.");
                        }
                        if (dataset.getDatasetType() != DatasetType.INTERNAL) {
                            setAsFinal(access, context, finalAnnot);
                            return false;
                        }
                        ConstantExpression ce = (ConstantExpression) e1;
                        IAObject obj = ((AsterixConstantValue) ce.getValue()).getObject();
                        String fldName;
                        if (obj.getType().getTypeTag() == ATypeTag.STRING) {
                            fldName = ((AString) obj).getStringValue();
                        } else {
                            int pos = ((AInt32) obj).getIntegerValue();
                            String tName = dataset.getItemTypeName();
                            IAType t = mp.findType(dataset.getItemTypeDataverseName(), tName);
                            if (t.getTypeTag() != ATypeTag.OBJECT) {
                                return false;
                            }
                            ARecordType rt = (ARecordType) t;
                            if (pos >= rt.getFieldNames().length) {
                                setAsFinal(access, context, finalAnnot);
                                return false;
                            }
                            fldName = rt.getFieldNames()[pos];
                        }
                        int p = DatasetUtil.getPositionOfPartitioningKeyField(dataset, fldName);
                        if (p < 0) {
                            // not one of the partitioning fields
                            setAsFinal(access, context, finalAnnot);
                            return false;
                        }
                        LogicalVariable keyVar = scan.getVariables().get(p);
                        access.getExpressions().get(0).setValue(new VariableReferenceExpression(keyVar));
                        return true;
                    }
                }
            }
        }
        setAsFinal(access, context, finalAnnot);
        return false;
    }
}

Example 8 with DataSourceScanOperator

use of org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator in project asterixdb by apache.

the class UnnestToDataScanRule method handleFunction.

protected boolean handleFunction(Mutable<ILogicalOperator> opRef, IOptimizationContext context, UnnestOperator unnest, AbstractFunctionCallExpression f) throws AlgebricksException {
    FunctionIdentifier fid = f.getFunctionIdentifier();
    if (fid.equals(BuiltinFunctions.DATASET)) {
        if (unnest.getPositionalVariable() != null) {
            // TODO remove this after enabling the support of positional variables in data scan
            throw new AlgebricksException("No positional variables are allowed over datasets.");
        }
        ILogicalExpression expr = f.getArguments().get(0).getValue();
        if (expr.getExpressionTag() != LogicalExpressionTag.CONSTANT) {
            return false;
        }
        ConstantExpression ce = (ConstantExpression) expr;
        IAlgebricksConstantValue acv = ce.getValue();
        if (!(acv instanceof AsterixConstantValue)) {
            return false;
        }
        AsterixConstantValue acv2 = (AsterixConstantValue) acv;
        if (acv2.getObject().getType().getTypeTag() != ATypeTag.STRING) {
            return false;
        }
        String datasetArg = ((AString) acv2.getObject()).getStringValue();
        MetadataProvider metadataProvider = (MetadataProvider) context.getMetadataProvider();
        Pair<String, String> datasetReference = parseDatasetReference(metadataProvider, datasetArg);
        String dataverseName = datasetReference.first;
        String datasetName = datasetReference.second;
        Dataset dataset = metadataProvider.findDataset(dataverseName, datasetName);
        if (dataset == null) {
            throw new AlgebricksException("Could not find dataset " + datasetName + " in dataverse " + dataverseName);
        }
        DataSourceId asid = new DataSourceId(dataverseName, datasetName);
        List<LogicalVariable> variables = new ArrayList<>();
        if (dataset.getDatasetType() == DatasetType.INTERNAL) {
            int numPrimaryKeys = dataset.getPrimaryKeys().size();
            for (int i = 0; i < numPrimaryKeys; i++) {
                variables.add(context.newVar());
            }
        }
        variables.add(unnest.getVariable());
        DataSource dataSource = metadataProvider.findDataSource(asid);
        boolean hasMeta = dataSource.hasMeta();
        if (hasMeta) {
            variables.add(context.newVar());
        }
        DataSourceScanOperator scan = new DataSourceScanOperator(variables, dataSource);
        List<Mutable<ILogicalOperator>> scanInpList = scan.getInputs();
        scanInpList.addAll(unnest.getInputs());
        opRef.setValue(scan);
        addPrimaryKey(variables, dataSource, context);
        context.computeAndSetTypeEnvironmentForOperator(scan);
        // Adds equivalence classes --- one equivalent class between a primary key
        // variable and a record field-access expression.
        IAType[] schemaTypes = dataSource.getSchemaTypes();
        ARecordType recordType = (ARecordType) (hasMeta ? schemaTypes[schemaTypes.length - 2] : schemaTypes[schemaTypes.length - 1]);
        ARecordType metaRecordType = (ARecordType) (hasMeta ? schemaTypes[schemaTypes.length - 1] : null);
        EquivalenceClassUtils.addEquivalenceClassesForPrimaryIndexAccess(scan, variables, recordType, metaRecordType, dataset, context);
        return true;
    } else if (fid.equals(BuiltinFunctions.FEED_COLLECT)) {
        if (unnest.getPositionalVariable() != null) {
            throw new AlgebricksException("No positional variables are allowed over feeds.");
        }
        String dataverse = ConstantExpressionUtil.getStringArgument(f, 0);
        String sourceFeedName = ConstantExpressionUtil.getStringArgument(f, 1);
        String getTargetFeed = ConstantExpressionUtil.getStringArgument(f, 2);
        String subscriptionLocation = ConstantExpressionUtil.getStringArgument(f, 3);
        String targetDataset = ConstantExpressionUtil.getStringArgument(f, 4);
        String outputType = ConstantExpressionUtil.getStringArgument(f, 5);
        MetadataProvider metadataProvider = (MetadataProvider) context.getMetadataProvider();
        DataSourceId asid = new DataSourceId(dataverse, getTargetFeed);
        String policyName = metadataProvider.getConfig().get(FeedActivityDetails.FEED_POLICY_NAME);
        FeedPolicyEntity policy = metadataProvider.findFeedPolicy(dataverse, policyName);
        if (policy == null) {
            policy = BuiltinFeedPolicies.getFeedPolicy(policyName);
            if (policy == null) {
                throw new AlgebricksException("Unknown feed policy:" + policyName);
            }
        }
        ArrayList<LogicalVariable> feedDataScanOutputVariables = new ArrayList<>();
        String csLocations = metadataProvider.getConfig().get(FeedActivityDetails.COLLECT_LOCATIONS);
        List<LogicalVariable> pkVars = new ArrayList<>();
        FeedDataSource ds = createFeedDataSource(asid, targetDataset, sourceFeedName, subscriptionLocation, metadataProvider, policy, outputType, csLocations, unnest.getVariable(), context, pkVars);
        // The order for feeds is <Record-Meta-PK>
        feedDataScanOutputVariables.add(unnest.getVariable());
        // Does it produce meta?
        if (ds.hasMeta()) {
            feedDataScanOutputVariables.add(context.newVar());
        }
        // Does it produce pk?
        if (ds.isChange()) {
            feedDataScanOutputVariables.addAll(pkVars);
        }
        DataSourceScanOperator scan = new DataSourceScanOperator(feedDataScanOutputVariables, ds);
        List<Mutable<ILogicalOperator>> scanInpList = scan.getInputs();
        scanInpList.addAll(unnest.getInputs());
        opRef.setValue(scan);
        context.computeAndSetTypeEnvironmentForOperator(scan);
        return true;
    }
    return false;
}

Example 9 with DataSourceScanOperator

use of org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator in project asterixdb by apache.

the class RemoveUnusedOneToOneEquiJoinRule method gatherProducingDataScans.

private void gatherProducingDataScans(Mutable<ILogicalOperator> opRef, List<LogicalVariable> joinUsedVars, List<DataSourceScanOperator> dataScans) {
    AbstractLogicalOperator op = (AbstractLogicalOperator) opRef.getValue();
    if (op.getOperatorTag() != LogicalOperatorTag.DATASOURCESCAN) {
        for (Mutable<ILogicalOperator> inputOp : op.getInputs()) {
            gatherProducingDataScans(inputOp, joinUsedVars, dataScans);
        }
        return;
    }
    DataSourceScanOperator dataScan = (DataSourceScanOperator) op;
    fillPKVars(dataScan, pkVars);
    // Check if join uses all PK vars.
    if (joinUsedVars.containsAll(pkVars)) {
        dataScans.add(dataScan);
    }
}

Example 10 with DataSourceScanOperator

use of org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator in project asterixdb by apache.

the class DataSourceScanPOperator method contributeRuntimeOperator.

@SuppressWarnings("unchecked")
@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema opSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
    DataSourceScanOperator scan = (DataSourceScanOperator) op;
    IMetadataProvider mp = context.getMetadataProvider();
    IVariableTypeEnvironment typeEnv = context.getTypeEnvironment(op);
    List<LogicalVariable> vars = scan.getVariables();
    List<LogicalVariable> projectVars = scan.getProjectVariables();
    Pair<IOperatorDescriptor, AlgebricksPartitionConstraint> p = mp.getScannerRuntime(dataSource, vars, projectVars, scan.isProjectPushed(), scan.getMinFilterVars(), scan.getMaxFilterVars(), opSchema, typeEnv, context, builder.getJobSpec(), implConfig);
    builder.contributeHyracksOperator(scan, p.first);
    if (p.second != null) {
        builder.contributeAlgebricksPartitionConstraint(p.first, p.second);
    }
    ILogicalOperator srcExchange = scan.getInputs().get(0).getValue();
    builder.contributeGraphEdge(srcExchange, 0, scan, 0);
}