Skip to content

branch-4.1: [refactor](fe) Create PreAggInfoContext top-down from aggregate instead of bottom-up from scan in SetPreAggStatus #65237 #65372

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
github-actions wants to merge 1 commit into
base: branch-4.1
Choose a base branch
from
+54 −17
Open

branch-4.1: [refactor](fe) Create PreAggInfoContext top-down from aggregate instead of bottom-up from scan in SetPreAggStatus #65237 #65372

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
46 changes: 31 additions & 15 deletions fe/fe-core/src/main/java/org/apache/doris/nereids/rules/rewrite/SetPreAggStatus.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,10 +54,13 @@
import org.apache.doris.nereids.trees.plans.visitor.DefaultPlanRewriter;
import org.apache.doris.nereids.util.ExpressionUtils;
import org.apache.doris.qe.ConnectContext;
import org.apache.doris.qe.SessionVariable;

import com.google.common.base.Preconditions;
import com.google.common.collect.Lists;
import com.google.common.collect.Sets;
import org.apache.logging.log4j.LogManager;
import org.apache.logging.log4j.Logger;

import java.util.ArrayList;
import java.util.HashMap;
Expand All @@ -75,6 +78,7 @@
*/
public class SetPreAggStatus extends DefaultPlanRewriter<Stack<SetPreAggStatus.PreAggInfoContext>>
implements CustomRewriter {
private static final Logger LOG = LogManager.getLogger(SetPreAggStatus.class);
private Map<RelationId, PreAggInfoContext> olapScanPreAggContexts = new HashMap<>();

/**
Expand Down Expand Up @@ -141,8 +145,11 @@ public Plan rewriteRoot(Plan plan, JobContext jobContext) {

@Override
public Plan visit(Plan plan, Stack<PreAggInfoContext> context) {
// push null sentinel to stop preagg collection for children reached via generic visitor,
// while keeping the aggregate's own frame intact on the stack
context.push(null);
Plan newPlan = super.visit(plan, context);
context.clear();
context.pop();
return newPlan;
}

Expand All @@ -161,10 +168,9 @@ public Plan visitLogicalOlapScan(LogicalOlapScan logicalOlapScan, Stack<PreAggIn
logicalOlapScan.getTable().getName()))) {
return logicalOlapScan.withPreAggStatus(PreAggStatus.on());
} else {
if (context.empty()) {
context.push(new PreAggInfoContext());
if (!context.empty() && context.peek() != null) {
context.peek().addRelationId(logicalOlapScan.getRelationId());
}
context.peek().addRelationId(logicalOlapScan.getRelationId());
return logicalOlapScan;
}
} else {
Expand All @@ -175,7 +181,7 @@ public Plan visitLogicalOlapScan(LogicalOlapScan logicalOlapScan, Stack<PreAggIn
@Override
public Plan visitLogicalFilter(LogicalFilter<? extends Plan> logicalFilter, Stack<PreAggInfoContext> context) {
LogicalFilter plan = (LogicalFilter) super.visit(logicalFilter, context);
if (!context.empty()) {
if (!context.empty() && context.peek() != null) {
context.peek().addFilterConjuncts(plan.getExpressions());
}
return plan;
Expand All @@ -185,7 +191,7 @@ public Plan visitLogicalFilter(LogicalFilter<? extends Plan> logicalFilter, Stac
public Plan visitLogicalJoin(LogicalJoin<? extends Plan, ? extends Plan> logicalJoin,
Stack<PreAggInfoContext> context) {
LogicalJoin plan = (LogicalJoin) super.visit(logicalJoin, context);
if (!context.empty()) {
if (!context.empty() && context.peek() != null) {
context.peek().addJoinInfo(plan);
}
return plan;
Expand All @@ -195,7 +201,7 @@ public Plan visitLogicalJoin(LogicalJoin<? extends Plan, ? extends Plan> logical
public Plan visitLogicalProject(LogicalProject<? extends Plan> logicalProject,
Stack<PreAggInfoContext> context) {
LogicalProject plan = (LogicalProject) super.visit(logicalProject, context);
if (!context.empty()) {
if (!context.empty() && context.peek() != null) {
context.peek().setReplaceMap(plan.getAliasToProducer());
}
return plan;
Expand All @@ -204,23 +210,33 @@ public Plan visitLogicalProject(LogicalProject<? extends Plan> logicalProject,
@Override
public Plan visitLogicalAggregate(LogicalAggregate<? extends Plan> logicalAggregate,
Stack<PreAggInfoContext> context) {
PreAggInfoContext preAggInfoContext = new PreAggInfoContext();
context.push(preAggInfoContext);
Plan plan = super.visit(logicalAggregate, context);
if (!context.isEmpty()) {
PreAggInfoContext preAggInfoContext = context.pop();
preAggInfoContext.olapScanIds.retainAll(logicalAggregate.child().getInputRelations());
preAggInfoContext.addAggregateFunctions(logicalAggregate.getAggregateFunctions());
preAggInfoContext.addGroupByExpresssions(logicalAggregate.getGroupByExpressions());
for (RelationId id : preAggInfoContext.olapScanIds) {
olapScanPreAggContexts.put(id, preAggInfoContext);
PreAggInfoContext popped = context.pop();
if (popped != preAggInfoContext) {
if (SessionVariable.isFeDebug()) {
Preconditions.checkState(popped == preAggInfoContext,
"PreAggInfoContext stack mismatch in visitLogicalAggregate");
} else {
LOG.warn("PreAggInfoContext stack mismatch in visitLogicalAggregate: "
+ "expected {} but got {}. Skipping preagg for this aggregate.",
preAggInfoContext, popped);
return plan;
}
}
popped.addAggregateFunctions(logicalAggregate.getAggregateFunctions());
popped.addGroupByExpresssions(logicalAggregate.getGroupByExpressions());
for (RelationId id : popped.olapScanIds) {
olapScanPreAggContexts.put(id, popped);
}
return plan;
}

@Override
public Plan visitLogicalRepeat(LogicalRepeat<? extends Plan> repeat, Stack<PreAggInfoContext> context) {
repeat = (LogicalRepeat<? extends Plan>) super.visit(repeat, context);
if (!context.isEmpty()) {
if (!context.isEmpty() && context.peek() != null) {
context.peek().addGroupingScalarFunctionExpresssion(repeat.getGroupingId().get());
context.peek().addGroupingScalarFunctionExpresssions(
repeat.getOutputExpressions().stream()
Expand Down
25 changes: 23 additions & 2 deletions regression-test/suites/nereids_rules_p0/set_preagg/set_preagg.groovy
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -271,7 +271,7 @@ suite("set_preagg") {
""")
contains "(preagg_t1), PREAGGREGATION: ON"
contains "(preagg_t2), PREAGGREGATION: ON"
contains "(preagg_t3), PREAGGREGATION: OFF"
contains "(preagg_t3), PREAGGREGATION: ON"
}

explain {
Expand All @@ -291,7 +291,7 @@ suite("set_preagg") {
""")
contains "(preagg_t1), PREAGGREGATION: ON"
contains "(preagg_t2), PREAGGREGATION: ON"
contains "(preagg_t3), PREAGGREGATION: OFF"
contains "(preagg_t3), PREAGGREGATION: ON"
}

explain {
Expand Down Expand Up @@ -328,4 +328,25 @@ suite("set_preagg") {
contains "(preagg_t1), PREAGGREGATION: OFF. Reason: No valid aggregate on scan."
contains "(preagg_t1), PREAGGREGATION: ON"
}

// Aggregate over limited subquery: Limit between aggregate and scan goes
// through generic visitor path, which should block preagg collection
// without clearing the aggregate's own frame.
explain {
sql("""
select k1, sum(v7)
from (select k1, v7 from preagg_t1 limit 10) s
group by k1;
""")
notContains "(preagg_t1), PREAGGREGATION: ON"
}

// Aggregate over non-Olap relation (numbers TVF).
// The TVF scan goes through the generic visitor path, which should not
// clear the aggregate's frame and cause EmptyStackException.
explain {
sql("""
select count(*) from (select * from numbers("number"="10")) t;
""")
}
}
Loading