rluvaton commented on code in PR #18152:
URL: https://github.com/apache/datafusion/pull/18152#discussion_r2466229121
##########
datafusion/physical-expr/src/expressions/case.rs:
##########
@@ -283,70 +717,81 @@ impl CaseExpr {
/// END
fn case_when_no_expr(&self, batch: &RecordBatch) -> Result<ColumnarValue> {
let return_type = self.data_type(&batch.schema())?;
+ let mut result_builder = ResultBuilder::new(&return_type,
batch.num_rows());
- // start with nulls as default output
- let mut current_value = new_null_array(&return_type, batch.num_rows());
- let mut remainder = BooleanArray::from(vec![true; batch.num_rows()]);
- let mut remainder_count = batch.num_rows();
- for i in 0..self.when_then_expr.len() {
- // If there are no rows left to process, break out of the loop
early
- if remainder_count == 0 {
- break;
- }
+ // `remainder_rows` contains the indices of the rows that need to be
evaluated
+ let mut remainder_rows: ArrayRef =
+ Arc::new(UInt32Array::from_iter(0..batch.num_rows() as u32));
+ // `remainder_batch` contains the rows themselves that need to be
evaluated
+ let mut remainder_batch = Cow::Borrowed(batch);
+ for i in 0..self.when_then_expr.len() {
+ // Evaluate the 'when' predicate for the remainder batch
+ // This results in a boolean array with the same length as the
remaining number of rows
let when_predicate = &self.when_then_expr[i].0;
- let when_value = when_predicate.evaluate_selection(batch,
&remainder)?;
- let when_value = when_value.into_array(batch.num_rows())?;
+ let when_value = when_predicate
+ .evaluate(&remainder_batch)?
+ .into_array(remainder_batch.num_rows())?;
let when_value = as_boolean_array(&when_value).map_err(|_| {
internal_datafusion_err!("WHEN expression did not return a
BooleanArray")
})?;
- // Treat 'NULL' as false value
- let when_value = match when_value.null_count() {
- 0 => Cow::Borrowed(when_value),
- _ => Cow::Owned(prep_null_mask_filter(when_value)),
- };
- // Make sure we only consider rows that have not been matched yet
- let when_value = and(&when_value, &remainder)?;
- // If the predicate did not match any rows, continue to the next
branch immediately
let when_match_count = when_value.true_count();
Review Comment:
Please add comment that `true_count` does not count nulls as true even when
the underlying value is true, so we can delay the preparation of the null_mask
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
