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

Example 16 with GroupByOperator

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

the class RemoveUnusedAssignAndAggregateRule method removeFromAssigns.

private int removeFromAssigns(AbstractLogicalOperator op, Set<LogicalVariable> toRemove, IOptimizationContext context) throws AlgebricksException {
    switch(op.getOperatorTag()) {
        case ASSIGN:
            AssignOperator assign = (AssignOperator) op;
            if (removeUnusedVarsAndExprs(toRemove, assign.getVariables(), assign.getExpressions())) {
                context.computeAndSetTypeEnvironmentForOperator(assign);
                isTransformed = true;
            }
            return assign.getVariables().size();
        case AGGREGATE:
            AggregateOperator agg = (AggregateOperator) op;
            if (removeUnusedVarsAndExprs(toRemove, agg.getVariables(), agg.getExpressions())) {
                context.computeAndSetTypeEnvironmentForOperator(agg);
                isTransformed = true;
            }
            return agg.getVariables().size();
        case UNNEST:
            UnnestOperator uOp = (UnnestOperator) op;
            LogicalVariable pVar = uOp.getPositionalVariable();
            if (pVar != null && toRemove != null && toRemove.contains(pVar)) {
                uOp.setPositionalVariable(null);
                assignedVarSet.remove(pVar);
                isTransformed = true;
            }
            break;
        case UNIONALL:
            UnionAllOperator unionOp = (UnionAllOperator) op;
            if (removeUnusedVarsFromUnionAll(unionOp, toRemove)) {
                context.computeAndSetTypeEnvironmentForOperator(unionOp);
                isTransformed = true;
            }
            return unionOp.getVariableMappings().size();
        case GROUP:
            GroupByOperator groupByOp = (GroupByOperator) op;
            if (removeUnusedVarsFromGroupBy(groupByOp, toRemove)) {
                context.computeAndSetTypeEnvironmentForOperator(groupByOp);
                isTransformed = true;
            }
            return groupByOp.getGroupByList().size() + groupByOp.getNestedPlans().size() + groupByOp.getDecorList().size();
        default:
            break;
    }
    return -1;
}

Example 17 with GroupByOperator

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

the class PushNestedOrderByUnderPreSortedGroupByRule method rewritePost.

@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context) throws AlgebricksException {
    AbstractLogicalOperator op = (AbstractLogicalOperator) opRef.getValue();
    if (op.getOperatorTag() != LogicalOperatorTag.GROUP) {
        return false;
    }
    if (op.getPhysicalOperator() == null) {
        return false;
    }
    AbstractPhysicalOperator pOp = (AbstractPhysicalOperator) op.getPhysicalOperator();
    if (pOp.getOperatorTag() != PhysicalOperatorTag.PRE_CLUSTERED_GROUP_BY) {
        return false;
    }
    GroupByOperator gby = (GroupByOperator) op;
    ILogicalPlan plan = gby.getNestedPlans().get(0);
    AbstractLogicalOperator op1 = (AbstractLogicalOperator) plan.getRoots().get(0).getValue();
    if (op1.getOperatorTag() != LogicalOperatorTag.AGGREGATE) {
        return false;
    }
    Mutable<ILogicalOperator> opRef2 = op1.getInputs().get(0);
    AbstractLogicalOperator op2 = (AbstractLogicalOperator) opRef2.getValue();
    if (op2.getOperatorTag() != LogicalOperatorTag.ORDER) {
        return false;
    }
    OrderOperator order1 = (OrderOperator) op2;
    if (!isIndependentFromChildren(order1)) {
        return false;
    }
    AbstractPhysicalOperator pOrder1 = (AbstractPhysicalOperator) op2.getPhysicalOperator();
    if (pOrder1.getOperatorTag() != PhysicalOperatorTag.STABLE_SORT && pOrder1.getOperatorTag() != PhysicalOperatorTag.IN_MEMORY_STABLE_SORT) {
        return false;
    }
    // StableSortPOperator sort1 = (StableSortPOperator) pOrder1;
    AbstractLogicalOperator op3 = (AbstractLogicalOperator) op.getInputs().get(0).getValue();
    if (op3.getOperatorTag() != LogicalOperatorTag.ORDER) {
        return false;
    }
    AbstractPhysicalOperator pOp3 = (AbstractPhysicalOperator) op3.getPhysicalOperator();
    if (pOp3.getOperatorTag() != PhysicalOperatorTag.STABLE_SORT) {
        return false;
    }
    OrderOperator order2 = (OrderOperator) op3;
    StableSortPOperator sort2 = (StableSortPOperator) pOp3;
    // int k = 0;
    for (Pair<IOrder, Mutable<ILogicalExpression>> oe : order1.getOrderExpressions()) {
        order2.getOrderExpressions().add(oe);
    // sortColumns[n2 + k] = sort1.getSortColumns()[k];
    // ++k;
    }
    // sort2.setSortColumns(sortColumns);
    sort2.computeDeliveredProperties(order2, null);
    // remove order1
    ILogicalOperator underOrder1 = order1.getInputs().get(0).getValue();
    opRef2.setValue(underOrder1);
    return true;
}

Example 18 with GroupByOperator

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

the class CountVarToCountOneRule method rewritePost.

// It is only for a group-by having just one aggregate which is a count.
@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context) throws AlgebricksException {
    AbstractLogicalOperator op1 = (AbstractLogicalOperator) opRef.getValue();
    if (op1.getOperatorTag() != LogicalOperatorTag.GROUP) {
        return false;
    }
    GroupByOperator g = (GroupByOperator) op1;
    if (g.getNestedPlans().size() != 1) {
        return false;
    }
    ILogicalPlan p = g.getNestedPlans().get(0);
    if (p.getRoots().size() != 1) {
        return false;
    }
    AbstractLogicalOperator op2 = (AbstractLogicalOperator) p.getRoots().get(0).getValue();
    if (op2.getOperatorTag() != LogicalOperatorTag.AGGREGATE) {
        return false;
    }
    AggregateOperator agg = (AggregateOperator) op2;
    if (agg.getExpressions().size() != 1) {
        return false;
    }
    ILogicalExpression exp2 = agg.getExpressions().get(0).getValue();
    if (exp2.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
        return false;
    }
    AbstractFunctionCallExpression fun = (AbstractFunctionCallExpression) exp2;
    if (fun.getFunctionIdentifier() != BuiltinFunctions.COUNT) {
        return false;
    }
    ILogicalExpression exp3 = fun.getArguments().get(0).getValue();
    if (exp3.getExpressionTag() != LogicalExpressionTag.VARIABLE) {
        return false;
    }
    if (((AbstractLogicalOperator) agg.getInputs().get(0).getValue()).getOperatorTag() != LogicalOperatorTag.NESTEDTUPLESOURCE) {
        return false;
    }
    fun.getArguments().get(0).setValue(new ConstantExpression(new AsterixConstantValue(new AInt64(1L))));
    return true;
}

Example 19 with GroupByOperator

use of org.apache.hyracks.algebricks.core.algebra.operators.logical.GroupByOperator 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 20 with GroupByOperator

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

the class RemoveLeftOuterUnnestForLeftOuterJoinRule method rewritePre.

@Override
public boolean rewritePre(Mutable<ILogicalOperator> opRef, IOptimizationContext context) throws AlgebricksException {
    ILogicalOperator op1 = opRef.getValue();
    // Checks the plan pattern.
    if (!checkOperatorPattern(op1)) {
        return false;
    }
    LeftOuterUnnestOperator outerUnnest = (LeftOuterUnnestOperator) op1;
    GroupByOperator gbyOperator = (GroupByOperator) outerUnnest.getInputs().get(0).getValue();
    LeftOuterJoinOperator lojOperator = (LeftOuterJoinOperator) gbyOperator.getInputs().get(0).getValue();
    // Checks whether the left outer unnest and the group-by operator are qualified for rewriting.
    Triple<Boolean, ILogicalExpression, ILogicalExpression> checkGbyResult = checkUnnestAndGby(outerUnnest, gbyOperator);
    // The argument for listify and not(is-missing(...)) check should be variables.
    if (!isVariableReference(checkGbyResult.second) || !isVariableReference(checkGbyResult.third)) {
        return false;
    }
    // Checks whether both the listify variable and the condition test variable are from the right input
    // branch of the left outer join.
    LogicalVariable listifyVar = ((VariableReferenceExpression) checkGbyResult.second).getVariableReference();
    LogicalVariable conditionTestVar = ((VariableReferenceExpression) checkGbyResult.third).getVariableReference();
    if (!checkListifyAndConditionVar(lojOperator, listifyVar, conditionTestVar)) {
        return false;
    }
    // Does the rewrite.
    removeGroupByAndOuterUnnest(opRef, context, outerUnnest, gbyOperator, lojOperator, listifyVar);
    return true;
}