Examples with TypeProvider io.trino.sql.planner.TypeProvider used on opensource projects

Example 1 with TypeProvider

use of io.trino.sql.planner.TypeProvider in project trino by trinodb.

the class InlineProjections method extractInliningTargets.

private static Set<Symbol> extractInliningTargets(PlannerContext plannerContext, ProjectNode parent, ProjectNode child, Session session, TypeAnalyzer typeAnalyzer, TypeProvider types) {
    // candidates for inlining are
    // 1. references to simple constants or symbol references
    // 2. references to complex expressions that
    // a. are not inputs to try() expressions
    // b. appear only once across all expressions
    // c. are not identity projections
    // which come from the child, as opposed to an enclosing scope.
    Set<Symbol> childOutputSet = ImmutableSet.copyOf(child.getOutputSymbols());
    Map<Symbol, Long> dependencies = parent.getAssignments().getExpressions().stream().flatMap(expression -> SymbolsExtractor.extractAll(expression).stream()).filter(childOutputSet::contains).collect(Collectors.groupingBy(Function.identity(), Collectors.counting()));
    // find references to simple constants or symbol references
    Set<Symbol> basicReferences = dependencies.keySet().stream().filter(input -> isEffectivelyLiteral(plannerContext, session, child.getAssignments().get(input)) || child.getAssignments().get(input) instanceof SymbolReference).filter(// skip identities, otherwise, this rule will keep firing forever
    input -> !child.getAssignments().isIdentity(input)).collect(toSet());
    // exclude any complex inputs to TRY expressions. Inlining them would potentially
    // change the semantics of those expressions
    Set<Symbol> tryArguments = parent.getAssignments().getExpressions().stream().flatMap(expression -> extractTryArguments(expression).stream()).collect(toSet());
    Set<Symbol> singletons = dependencies.entrySet().stream().filter(// reference appears just once across all expressions in parent project node
    entry -> entry.getValue() == 1).filter(// they are not inputs to TRY. Otherwise, inlining might change semantics
    entry -> !tryArguments.contains(entry.getKey())).filter(// skip identities, otherwise, this rule will keep firing forever
    entry -> !child.getAssignments().isIdentity(entry.getKey())).filter(entry -> {
        // skip dereferences, otherwise, inlining can cause conflicts with PushdownDereferences
        Expression assignment = child.getAssignments().get(entry.getKey());
        if (assignment instanceof SubscriptExpression) {
            if (typeAnalyzer.getType(session, types, ((SubscriptExpression) assignment).getBase()) instanceof RowType) {
                return false;
            }
        }
        return true;
    }).map(Map.Entry::getKey).collect(toSet());
    return Sets.union(singletons, basicReferences);
}

Example 2 with TypeProvider

use of io.trino.sql.planner.TypeProvider in project trino by trinodb.

the class DetermineSemiJoinDistributionType method getSemiJoinNodeWithCost.

private PlanNodeWithCost getSemiJoinNodeWithCost(SemiJoinNode possibleJoinNode, Context context) {
    TypeProvider types = context.getSymbolAllocator().getTypes();
    StatsProvider stats = context.getStatsProvider();
    boolean replicated = possibleJoinNode.getDistributionType().get() == REPLICATED;
    /*
         *   HACK!
         *
         *   Currently cost model always has to compute the total cost of an operation.
         *   For SEMI-JOIN the total cost consist of 4 parts:
         *     - Cost of exchanges that have to be introduced to execute a JOIN
         *     - Cost of building a hash table
         *     - Cost of probing a hash table
         *     - Cost of building an output for matched rows
         *
         *   When output size for a SEMI-JOIN cannot be estimated the cost model returns
         *   UNKNOWN cost for the join.
         *
         *   However assuming the cost of SEMI-JOIN output is always the same, we can still make
         *   cost based decisions based on the input cost for different types of SEMI-JOINs.
         *
         *   TODO Decision about the distribution should be based on LocalCostEstimate only when PlanCostEstimate cannot be calculated. Otherwise cost comparator cannot take query.max-memory into account.
         */
    int estimatedSourceDistributedTaskCount = taskCountEstimator.estimateSourceDistributedTaskCount(context.getSession());
    LocalCostEstimate cost = calculateJoinCostWithoutOutput(possibleJoinNode.getSource(), possibleJoinNode.getFilteringSource(), stats, types, replicated, estimatedSourceDistributedTaskCount);
    return new PlanNodeWithCost(cost.toPlanCost(), possibleJoinNode);
}

Example 3 with TypeProvider

use of io.trino.sql.planner.TypeProvider in project trino by trinodb.

the class PushJoinIntoTableScan method getJoinStatistics.

private JoinStatistics getJoinStatistics(JoinNode join, TableScanNode left, TableScanNode right, Context context) {
    return new JoinStatistics() {

        @Override
        public Optional<BasicRelationStatistics> getLeftStatistics() {
            return getBasicRelationStats(left, left.getOutputSymbols(), context);
        }

        @Override
        public Optional<BasicRelationStatistics> getRightStatistics() {
            return getBasicRelationStats(right, right.getOutputSymbols(), context);
        }

        @Override
        public Optional<BasicRelationStatistics> getJoinStatistics() {
            return getBasicRelationStats(join, join.getOutputSymbols(), context);
        }

        private Optional<BasicRelationStatistics> getBasicRelationStats(PlanNode node, List<Symbol> outputSymbols, Context context) {
            PlanNodeStatsEstimate stats = context.getStatsProvider().getStats(node);
            TypeProvider types = context.getSymbolAllocator().getTypes();
            double outputRowCount = stats.getOutputRowCount();
            double outputSize = stats.getOutputSizeInBytes(outputSymbols, types);
            if (isNaN(outputRowCount) || isNaN(outputSize)) {
                return Optional.empty();
            }
            return Optional.of(new BasicRelationStatistics((long) outputRowCount, (long) outputSize));
        }
    };
}

Example 4 with TypeProvider

use of io.trino.sql.planner.TypeProvider in project trino by trinodb.

the class PushProjectionThroughJoin method inlineProjections.

private static PlanNode inlineProjections(PlannerContext plannerContext, ProjectNode parentProjection, Lookup lookup, Session session, TypeAnalyzer typeAnalyzer, TypeProvider types) {
    PlanNode child = lookup.resolve(parentProjection.getSource());
    if (!(child instanceof ProjectNode)) {
        return parentProjection;
    }
    ProjectNode childProjection = (ProjectNode) child;
    return InlineProjections.inlineProjections(plannerContext, parentProjection, childProjection, session, typeAnalyzer, types).map(node -> inlineProjections(plannerContext, node, lookup, session, typeAnalyzer, types)).orElse(parentProjection);
}

Example 5 with TypeProvider

use of io.trino.sql.planner.TypeProvider in project trino by trinodb.

the class PushProjectionThroughJoin method pushProjectionThroughJoin.

public static Optional<PlanNode> pushProjectionThroughJoin(PlannerContext plannerContext, ProjectNode projectNode, Lookup lookup, PlanNodeIdAllocator planNodeIdAllocator, Session session, TypeAnalyzer typeAnalyzer, TypeProvider types) {
    if (!projectNode.getAssignments().getExpressions().stream().allMatch(expression -> isDeterministic(expression, plannerContext.getMetadata()))) {
        return Optional.empty();
    }
    PlanNode child = lookup.resolve(projectNode.getSource());
    if (!(child instanceof JoinNode)) {
        return Optional.empty();
    }
    JoinNode joinNode = (JoinNode) child;
    PlanNode leftChild = joinNode.getLeft();
    PlanNode rightChild = joinNode.getRight();
    if (joinNode.getType() != INNER) {
        return Optional.empty();
    }
    Assignments.Builder leftAssignmentsBuilder = Assignments.builder();
    Assignments.Builder rightAssignmentsBuilder = Assignments.builder();
    for (Map.Entry<Symbol, Expression> assignment : projectNode.getAssignments().entrySet()) {
        Expression expression = assignment.getValue();
        Set<Symbol> symbols = extractUnique(expression);
        if (leftChild.getOutputSymbols().containsAll(symbols)) {
            // expression is satisfied with left child symbols
            leftAssignmentsBuilder.put(assignment.getKey(), expression);
        } else if (rightChild.getOutputSymbols().containsAll(symbols)) {
            // expression is satisfied with right child symbols
            rightAssignmentsBuilder.put(assignment.getKey(), expression);
        } else {
            // expression is using symbols from both join sides
            return Optional.empty();
        }
    }
    // add projections for symbols required by the join itself
    Set<Symbol> joinRequiredSymbols = getJoinRequiredSymbols(joinNode);
    for (Symbol requiredSymbol : joinRequiredSymbols) {
        if (leftChild.getOutputSymbols().contains(requiredSymbol)) {
            leftAssignmentsBuilder.putIdentity(requiredSymbol);
        } else {
            checkState(rightChild.getOutputSymbols().contains(requiredSymbol));
            rightAssignmentsBuilder.putIdentity(requiredSymbol);
        }
    }
    Assignments leftAssignments = leftAssignmentsBuilder.build();
    Assignments rightAssignments = rightAssignmentsBuilder.build();
    List<Symbol> leftOutputSymbols = leftAssignments.getOutputs().stream().filter(ImmutableSet.copyOf(projectNode.getOutputSymbols())::contains).collect(toImmutableList());
    List<Symbol> rightOutputSymbols = rightAssignments.getOutputs().stream().filter(ImmutableSet.copyOf(projectNode.getOutputSymbols())::contains).collect(toImmutableList());
    return Optional.of(new JoinNode(joinNode.getId(), joinNode.getType(), inlineProjections(plannerContext, new ProjectNode(planNodeIdAllocator.getNextId(), leftChild, leftAssignments), lookup, session, typeAnalyzer, types), inlineProjections(plannerContext, new ProjectNode(planNodeIdAllocator.getNextId(), rightChild, rightAssignments), lookup, session, typeAnalyzer, types), joinNode.getCriteria(), leftOutputSymbols, rightOutputSymbols, joinNode.isMaySkipOutputDuplicates(), joinNode.getFilter(), joinNode.getLeftHashSymbol(), joinNode.getRightHashSymbol(), joinNode.getDistributionType(), joinNode.isSpillable(), joinNode.getDynamicFilters(), joinNode.getReorderJoinStatsAndCost()));
}