Search in sources :

Example 1 with LoadableDataSource

use of org.apache.asterix.metadata.declared.LoadableDataSource in project asterixdb by apache.

the class LangExpressionToPlanTranslator method translateLoad.

@Override
public ILogicalPlan translateLoad(ICompiledDmlStatement stmt) throws AlgebricksException {
    CompiledLoadFromFileStatement clffs = (CompiledLoadFromFileStatement) stmt;
    Dataset dataset = metadataProvider.findDataset(clffs.getDataverseName(), clffs.getDatasetName());
    if (dataset == null) {
        // This would never happen since we check for this in AqlTranslator
        throw new AlgebricksException("Unable to load dataset " + clffs.getDatasetName() + " since it does not exist");
    }
    IAType itemType = metadataProvider.findType(dataset.getItemTypeDataverseName(), dataset.getItemTypeName());
    IAType metaItemType = metadataProvider.findType(dataset.getMetaItemTypeDataverseName(), dataset.getMetaItemTypeName());
    DatasetDataSource targetDatasource = validateDatasetInfo(metadataProvider, stmt.getDataverseName(), stmt.getDatasetName());
    List<List<String>> partitionKeys = targetDatasource.getDataset().getPrimaryKeys();
    if (dataset.hasMetaPart()) {
        throw new AlgebricksException(dataset.getDatasetName() + ": load dataset is not supported on Datasets with Meta records");
    }
    LoadableDataSource lds;
    try {
        lds = new LoadableDataSource(dataset, itemType, metaItemType, clffs.getAdapter(), clffs.getProperties());
    } catch (IOException e) {
        throw new AlgebricksException(e);
    }
    // etsOp is a dummy input operator used to keep the compiler happy. it
    // could be removed but would result in
    // the need to fix many rewrite rules that assume that datasourcescan
    // operators always have input.
    ILogicalOperator etsOp = new EmptyTupleSourceOperator();
    // Add a logical variable for the record.
    List<LogicalVariable> payloadVars = new ArrayList<>();
    payloadVars.add(context.newVar());
    // Create a scan operator and make the empty tuple source its input
    DataSourceScanOperator dssOp = new DataSourceScanOperator(payloadVars, lds);
    dssOp.getInputs().add(new MutableObject<>(etsOp));
    ILogicalExpression payloadExpr = new VariableReferenceExpression(payloadVars.get(0));
    Mutable<ILogicalExpression> payloadRef = new MutableObject<>(payloadExpr);
    // Creating the assign to extract the PK out of the record
    ArrayList<LogicalVariable> pkVars = new ArrayList<>();
    ArrayList<Mutable<ILogicalExpression>> pkExprs = new ArrayList<>();
    List<Mutable<ILogicalExpression>> varRefsForLoading = new ArrayList<>();
    LogicalVariable payloadVar = payloadVars.get(0);
    for (List<String> keyFieldName : partitionKeys) {
        PlanTranslationUtil.prepareVarAndExpression(keyFieldName, payloadVar, pkVars, pkExprs, varRefsForLoading, context);
    }
    AssignOperator assign = new AssignOperator(pkVars, pkExprs);
    assign.getInputs().add(new MutableObject<>(dssOp));
    // If the input is pre-sorted, we set the ordering property explicitly in the assign
    if (clffs.alreadySorted()) {
        List<OrderColumn> orderColumns = new ArrayList<>();
        for (int i = 0; i < pkVars.size(); ++i) {
            orderColumns.add(new OrderColumn(pkVars.get(i), OrderKind.ASC));
        }
        assign.setExplicitOrderingProperty(new LocalOrderProperty(orderColumns));
    }
    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, payloadVar, additionalFilteringVars, additionalFilteringAssignExpressions, additionalFilteringExpressions, context);
        additionalFilteringAssign = new AssignOperator(additionalFilteringVars, additionalFilteringAssignExpressions);
    }
    InsertDeleteUpsertOperator insertOp = new InsertDeleteUpsertOperator(targetDatasource, payloadRef, varRefsForLoading, InsertDeleteUpsertOperator.Kind.INSERT, true);
    insertOp.setAdditionalFilteringExpressions(additionalFilteringExpressions);
    if (additionalFilteringAssign != null) {
        additionalFilteringAssign.getInputs().add(new MutableObject<>(assign));
        insertOp.getInputs().add(new MutableObject<>(additionalFilteringAssign));
    } else {
        insertOp.getInputs().add(new MutableObject<>(assign));
    }
    ILogicalOperator leafOperator = new SinkOperator();
    leafOperator.getInputs().add(new MutableObject<>(insertOp));
    return new ALogicalPlanImpl(new MutableObject<>(leafOperator));
}
Also used : CompiledLoadFromFileStatement(org.apache.asterix.translator.CompiledStatements.CompiledLoadFromFileStatement) ArrayList(java.util.ArrayList) DatasetDataSource(org.apache.asterix.metadata.declared.DatasetDataSource) AString(org.apache.asterix.om.base.AString) DataSourceScanOperator(org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator) ALogicalPlanImpl(org.apache.hyracks.algebricks.core.algebra.plan.ALogicalPlanImpl) ArrayList(java.util.ArrayList) List(java.util.List) LoadableDataSource(org.apache.asterix.metadata.declared.LoadableDataSource) EmptyTupleSourceOperator(org.apache.hyracks.algebricks.core.algebra.operators.logical.EmptyTupleSourceOperator) MutableObject(org.apache.commons.lang3.mutable.MutableObject) LogicalVariable(org.apache.hyracks.algebricks.core.algebra.base.LogicalVariable) SinkOperator(org.apache.hyracks.algebricks.core.algebra.operators.logical.SinkOperator) Dataset(org.apache.asterix.metadata.entities.Dataset) ILogicalOperator(org.apache.hyracks.algebricks.core.algebra.base.ILogicalOperator) OrderColumn(org.apache.hyracks.algebricks.core.algebra.properties.OrderColumn) AlgebricksException(org.apache.hyracks.algebricks.common.exceptions.AlgebricksException) IOException(java.io.IOException) AssignOperator(org.apache.hyracks.algebricks.core.algebra.operators.logical.AssignOperator) Mutable(org.apache.commons.lang3.mutable.Mutable) ILogicalExpression(org.apache.hyracks.algebricks.core.algebra.base.ILogicalExpression) InsertDeleteUpsertOperator(org.apache.hyracks.algebricks.core.algebra.operators.logical.InsertDeleteUpsertOperator) VariableReferenceExpression(org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression) LocalOrderProperty(org.apache.hyracks.algebricks.core.algebra.properties.LocalOrderProperty) IAType(org.apache.asterix.om.types.IAType)

Aggregations

IOException (java.io.IOException)1 ArrayList (java.util.ArrayList)1 List (java.util.List)1 DatasetDataSource (org.apache.asterix.metadata.declared.DatasetDataSource)1 LoadableDataSource (org.apache.asterix.metadata.declared.LoadableDataSource)1 Dataset (org.apache.asterix.metadata.entities.Dataset)1 AString (org.apache.asterix.om.base.AString)1 IAType (org.apache.asterix.om.types.IAType)1 CompiledLoadFromFileStatement (org.apache.asterix.translator.CompiledStatements.CompiledLoadFromFileStatement)1 Mutable (org.apache.commons.lang3.mutable.Mutable)1 MutableObject (org.apache.commons.lang3.mutable.MutableObject)1 AlgebricksException (org.apache.hyracks.algebricks.common.exceptions.AlgebricksException)1 ILogicalExpression (org.apache.hyracks.algebricks.core.algebra.base.ILogicalExpression)1 ILogicalOperator (org.apache.hyracks.algebricks.core.algebra.base.ILogicalOperator)1 LogicalVariable (org.apache.hyracks.algebricks.core.algebra.base.LogicalVariable)1 VariableReferenceExpression (org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression)1 AssignOperator (org.apache.hyracks.algebricks.core.algebra.operators.logical.AssignOperator)1 DataSourceScanOperator (org.apache.hyracks.algebricks.core.algebra.operators.logical.DataSourceScanOperator)1 EmptyTupleSourceOperator (org.apache.hyracks.algebricks.core.algebra.operators.logical.EmptyTupleSourceOperator)1 InsertDeleteUpsertOperator (org.apache.hyracks.algebricks.core.algebra.operators.logical.InsertDeleteUpsertOperator)1