Ted-Jiang commented on code in PR #2591:
URL: https://github.com/apache/arrow-datafusion/pull/2591#discussion_r881201496
##########
datafusion/core/src/physical_plan/hash_join.rs:
##########
@@ -791,6 +826,109 @@ fn build_join_indexes(
}
}
+fn apply_join_filter(
+ left: &RecordBatch,
+ right: &RecordBatch,
+ join_type: JoinType,
+ left_indices: UInt64Array,
+ right_indices: UInt32Array,
+ filter: &JoinFilter,
+) -> Result<(UInt64Array, UInt32Array)> {
+ if left_indices.is_empty() {
+ return Ok((left_indices, right_indices));
+ };
+
+ let (intermediate_batch, _) = build_batch_from_indices(
+ filter.schema(),
+ left,
+ right,
+ PrimitiveArray::from(left_indices.data().clone()),
+ PrimitiveArray::from(right_indices.data().clone()),
+ filter.column_indices(),
+ )?;
+
+ match join_type {
+ JoinType::Inner | JoinType::Left => {
+ // For both INNER and LEFT joins, input arrays contains only
indices for matched data.
+ // Due to this fact it's correct to simply apply filter to
intermediate batch and return
+ // indices for left/right rows satisfying filter predicate
+ let filter_result = filter
+ .expression()
+ .evaluate(&intermediate_batch)?
+ .into_array(intermediate_batch.num_rows());
+ let mask = as_boolean_array(&filter_result);
+
+ let left_filtered = PrimitiveArray::<UInt64Type>::from(
+ compute::filter(&left_indices, mask)?.data().clone(),
Review Comment:
@korowa @alamb ❤️ Thanks your explanations! I think after merge this Pr, we
should add a check in `filter_push_down.rs` to improve this situation.
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
