nuno-faria commented on code in PR #17201:
URL: https://github.com/apache/datafusion/pull/17201#discussion_r2281651562
##########
datafusion/sqllogictest/test_files/push_down_filter.slt:
##########
@@ -286,5 +286,37 @@ explain select a from t where CAST(a AS string) = '0123';
physical_plan DataSourceExec: file_groups={1 group:
[[WORKSPACE_ROOT/datafusion/sqllogictest/test_files/scratch/push_down_filter/t.parquet]]},
projection=[a], file_type=parquet, predicate=CAST(a@0 AS Utf8View) = 0123
+# Test dynamic filter pushdown with swapped join inputs (issue #17196)
+# Create tables with different sizes to force join input swapping
+statement ok
+copy (select i as k from generate_series(1, 100) t(i)) to
'test_files/scratch/push_down_filter/small_table.parquet';
+
+statement ok
+copy (select i as k, i as v from generate_series(1, 1000) t(i)) to
'test_files/scratch/push_down_filter/large_table.parquet';
+
+statement ok
+create external table small_table stored as parquet location
'test_files/scratch/push_down_filter/small_table.parquet';
+
+statement ok
+create external table large_table stored as parquet location
'test_files/scratch/push_down_filter/large_table.parquet';
+
+# Test that dynamic filter is applied to the correct table after join input
swapping
+# The small_table should be the build side, large_table should be the probe
side with dynamic filter
+query TT
+explain select * from small_table join large_table on small_table.k =
large_table.k where large_table.v >= 50;
+----
+physical_plan
+01)CoalesceBatchesExec: target_batch_size=8192
+02)--HashJoinExec: mode=CollectLeft, join_type=Inner, on=[(k@0, k@0)]
+03)----DataSourceExec: file_groups={1 group:
[[WORKSPACE_ROOT/datafusion/sqllogictest/test_files/scratch/push_down_filter/small_table.parquet]]},
projection=[k], file_type=parquet
+04)----RepartitionExec: partitioning=RoundRobinBatch(4), input_partitions=1
+05)------DataSourceExec: file_groups={1 group:
[[WORKSPACE_ROOT/datafusion/sqllogictest/test_files/scratch/push_down_filter/large_table.parquet]]},
projection=[k, v], file_type=parquet, predicate=v@1 >= 50 AND
DynamicFilterPhysicalExpr [ true ], pruning_predicate=v_null_count@1 !=
row_count@2 AND v_max@0 >= 50, required_guarantees=[]
Review Comment:
```sql
DataSourceExec small_table.parquet
DataSourceExec large_table.parquet, predicate=v@1 >= 50 AND
DynamicFilterPhysicalExpr [ true ]
```
But in this case `large_table` is filtered by both `v >= 50` and the dynamic
filter. Shouldn't the dynamic filter be applied to `small_table` instead?
If `where v = 50` is executed instead, the dynamic filter now appears in the
`small_table` table:
```sql
DataSourceExec large_table.parquet, predicate=v@1 = 50
DataSourceExec small_table.parquet, predicate=DynamicFilterPhysicalExpr [
k@0 >= 50 AND k@0 <= 50 ]
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org
For additional commands, e-mail: github-h...@datafusion.apache.org
