@Override
public Void visitSemiJoin(SemiJoinNode node, Void context) {
printNode(
node,
"SemiJoin",
format("%s = %s", node.getSourceJoinSymbol(), node.getFilteringSourceJoinSymbol()),
NODE_COLORS.get(NodeType.JOIN));
node.getSource().accept(this, context);
node.getFilteringSource().accept(this, context);
return null;
}
@Override
public SubPlanBuilder visitSemiJoin(SemiJoinNode node, Void context) {
SubPlanBuilder source = node.getSource().accept(this, context);
SubPlanBuilder filteringSource = node.getFilteringSource().accept(this, context);
if (source.isDistributed() || filteringSource.isDistributed()) {
filteringSource.setRoot(
new SinkNode(
idAllocator.getNextId(),
filteringSource.getRoot(),
filteringSource.getRoot().getOutputSymbols()));
source.setRoot(
new SemiJoinNode(
node.getId(),
source.getRoot(),
new ExchangeNode(
idAllocator.getNextId(),
filteringSource.getId(),
filteringSource.getRoot().getOutputSymbols()),
node.getSourceJoinSymbol(),
node.getFilteringSourceJoinSymbol(),
node.getSemiJoinOutput()));
source.addChild(filteringSource.build());
return source;
} else {
SemiJoinNode semiJoinNode =
new SemiJoinNode(
node.getId(),
source.getRoot(),
filteringSource.getRoot(),
node.getSourceJoinSymbol(),
node.getFilteringSourceJoinSymbol(),
node.getSemiJoinOutput());
return createSingleNodePlan(semiJoinNode)
.setChildren(Iterables.concat(source.getChildren(), filteringSource.getChildren()));
}
}
@Override
public PlanNode rewriteSemiJoin(
SemiJoinNode node, Expression inheritedPredicate, PlanRewriter<Expression> planRewriter) {
Expression sourceEffectivePredicate = EffectivePredicateExtractor.extract(node.getSource());
List<Expression> sourceConjuncts = new ArrayList<>();
List<Expression> filteringSourceConjuncts = new ArrayList<>();
List<Expression> postJoinConjuncts = new ArrayList<>();
// TODO: see if there are predicates that can be inferred from the semi join output
// Push inherited and source predicates to filtering source via a contrived join predicate
// (but needs to avoid touching NULL values in the filtering source)
Expression joinPredicate =
equalsExpression(node.getSourceJoinSymbol(), node.getFilteringSourceJoinSymbol());
EqualityInference joinInference =
createEqualityInference(inheritedPredicate, sourceEffectivePredicate, joinPredicate);
for (Expression conjunct :
Iterables.concat(
EqualityInference.nonInferrableConjuncts(inheritedPredicate),
EqualityInference.nonInferrableConjuncts(sourceEffectivePredicate))) {
Expression rewrittenConjunct =
joinInference.rewriteExpression(conjunct, equalTo(node.getFilteringSourceJoinSymbol()));
if (rewrittenConjunct != null && DeterminismEvaluator.isDeterministic(rewrittenConjunct)) {
// Alter conjunct to include an OR filteringSourceJoinSymbol IS NULL disjunct
Expression rewrittenConjunctOrNull =
expressionOrNullSymbols(equalTo(node.getFilteringSourceJoinSymbol()))
.apply(rewrittenConjunct);
filteringSourceConjuncts.add(rewrittenConjunctOrNull);
}
}
EqualityInference.EqualityPartition joinInferenceEqualityPartition =
joinInference.generateEqualitiesPartitionedBy(
equalTo(node.getFilteringSourceJoinSymbol()));
filteringSourceConjuncts.addAll(
ImmutableList.copyOf(
transform(
joinInferenceEqualityPartition.getScopeEqualities(),
expressionOrNullSymbols(equalTo(node.getFilteringSourceJoinSymbol())))));
// Push inheritedPredicates down to the source if they don't involve the semi join output
EqualityInference inheritedInference = createEqualityInference(inheritedPredicate);
for (Expression conjunct : EqualityInference.nonInferrableConjuncts(inheritedPredicate)) {
Expression rewrittenConjunct =
inheritedInference.rewriteExpression(conjunct, in(node.getSource().getOutputSymbols()));
// Since each source row is reflected exactly once in the output, ok to push
// non-deterministic predicates down
if (rewrittenConjunct != null) {
sourceConjuncts.add(rewrittenConjunct);
} else {
postJoinConjuncts.add(conjunct);
}
}
// Add the inherited equality predicates back in
EqualityInference.EqualityPartition equalityPartition =
inheritedInference.generateEqualitiesPartitionedBy(
in(node.getSource().getOutputSymbols()));
sourceConjuncts.addAll(equalityPartition.getScopeEqualities());
postJoinConjuncts.addAll(equalityPartition.getScopeComplementEqualities());
postJoinConjuncts.addAll(equalityPartition.getScopeStraddlingEqualities());
PlanNode rewrittenSource =
planRewriter.rewrite(node.getSource(), combineConjuncts(sourceConjuncts));
PlanNode rewrittenFilteringSource =
planRewriter.rewrite(
node.getFilteringSource(), combineConjuncts(filteringSourceConjuncts));
PlanNode output = node;
if (rewrittenSource != node.getSource()
|| rewrittenFilteringSource != node.getFilteringSource()) {
output =
new SemiJoinNode(
node.getId(),
rewrittenSource,
rewrittenFilteringSource,
node.getSourceJoinSymbol(),
node.getFilteringSourceJoinSymbol(),
node.getSemiJoinOutput());
}
if (!postJoinConjuncts.isEmpty()) {
output =
new FilterNode(idAllocator.getNextId(), output, combineConjuncts(postJoinConjuncts));
}
return output;
}
@Override
public Expression visitSemiJoin(SemiJoinNode node, Void context) {
// Filtering source does not change the effective predicate over the output symbols
return node.getSource().accept(this, context);
}
@Override
public PlanNode visitSemiJoin(SemiJoinNode node, RewriteContext<Set<Symbol>> context) {
ImmutableSet.Builder<Symbol> sourceInputsBuilder = ImmutableSet.builder();
sourceInputsBuilder.addAll(context.get()).add(node.getSourceJoinSymbol());
if (node.getSourceHashSymbol().isPresent()) {
sourceInputsBuilder.add(node.getSourceHashSymbol().get());
}
Set<Symbol> sourceInputs = sourceInputsBuilder.build();
ImmutableSet.Builder<Symbol> filteringSourceInputBuilder = ImmutableSet.builder();
filteringSourceInputBuilder.add(node.getFilteringSourceJoinSymbol());
if (node.getFilteringSourceHashSymbol().isPresent()) {
filteringSourceInputBuilder.add(node.getFilteringSourceHashSymbol().get());
}
Set<Symbol> filteringSourceInputs = filteringSourceInputBuilder.build();
PlanNode source = context.rewrite(node.getSource(), sourceInputs);
PlanNode filteringSource = context.rewrite(node.getFilteringSource(), filteringSourceInputs);
return new SemiJoinNode(
node.getId(),
source,
filteringSource,
node.getSourceJoinSymbol(),
node.getFilteringSourceJoinSymbol(),
node.getSemiJoinOutput(),
node.getSourceHashSymbol(),
node.getFilteringSourceHashSymbol());
}