Dandandan commented on code in PR #15697:
URL: https://github.com/apache/datafusion/pull/15697#discussion_r2042641574
##########
datafusion/physical-plan/src/topk/mod.rs:
##########
@@ -202,27 +204,99 @@ impl TopK {
})
.collect::<Result<Vec<_>>>()?;
+ // selected indices
+ let mut selected_rows = None;
+
+ // If the heap doesn't have k elements yet, we can't create thresholds
+ if let Some(max_row) = self.heap.max() {
+ // Get the batch that contains the max row
+ let batch_entry = match self.heap.store.get(max_row.batch_id) {
+ Some(entry) => entry,
+ None => return internal_err!("Invalid batch ID in TopKRow"),
+ };
+
+ // Extract threshold values for each sort expression
+ // TODO: create a filter for each key that respects lexical
ordering
+ // in the form of col0 < threshold0 || col0 == threshold0 && (col1
< threshold1 || ...)
+ // This could use BinaryExpr to benefit from short circuiting and
early evaluation
+ // https://github.com/apache/datafusion/issues/15698
+ // Extract the value for this column from the max row
+ let expr = Arc::clone(&self.expr[0].expr);
+ let value = expr.evaluate(&batch_entry.batch.slice(max_row.index,
1))?;
+
+ // Convert to scalar value - should be a single value since we're
evaluating on a single row batch
+ let threshold = Scalar::new(value.to_array(1)?);
+
+ // Create a filter for each sort key
+ let is_multi_col = self.expr.len() > 1;
+ let filter = match (is_multi_col, self.expr[0].options.descending)
{
Review Comment:
Nulls are skipped for now
--
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
