This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/datafusion.git
The following commit(s) were added to refs/heads/main by this push:
new e6849945bf fix: `cardinality()` of an empty array should be zero
(#20533)
e6849945bf is described below
commit e6849945bf280e7dcc5837e048e0759bde388d78
Author: Neil Conway <[email protected]>
AuthorDate: Wed Feb 25 09:02:43 2026 -0500
fix: `cardinality()` of an empty array should be zero (#20533)
## Which issue does this PR close?
- Closes #20526.
## Rationale for this change
Per Postgres and the SQL spec, `cardinality()` of an empty array should
be zero; we previously returned `NULL`.
Along the way, fix another bug: we previously returned `0` for the
cardinality of an untyped `NULL` and `NULL` for the cardinality of a
typed null (e.g., `NULL::int[]`). We should return `NULL` in both cases.
## What changes are included in this PR?
Bug fixes, update SLT.
## Are these changes tested?
Yes.
## Are there any user-facing changes?
Yes: the behavior of `cardinality` has changed, albeit the previous
behavior was incorrect.
---
datafusion/functions-nested/src/cardinality.rs | 13 +++++++++----
datafusion/sqllogictest/test_files/array.slt | 15 +++++++++++++--
2 files changed, 22 insertions(+), 6 deletions(-)
diff --git a/datafusion/functions-nested/src/cardinality.rs
b/datafusion/functions-nested/src/cardinality.rs
index c467686b86..8953a8568f 100644
--- a/datafusion/functions-nested/src/cardinality.rs
+++ b/datafusion/functions-nested/src/cardinality.rs
@@ -120,7 +120,7 @@ impl ScalarUDFImpl for Cardinality {
fn cardinality_inner(args: &[ArrayRef]) -> Result<ArrayRef> {
let [array] = take_function_args("cardinality", args)?;
match array.data_type() {
- Null => Ok(Arc::new(UInt64Array::from_value(0, array.len()))),
+ Null => Ok(Arc::new(UInt64Array::new_null(array.len()))),
List(_) => {
let list_array = as_list_array(array)?;
generic_list_cardinality::<i32>(list_array)
@@ -152,9 +152,14 @@ fn generic_list_cardinality<O: OffsetSizeTrait>(
) -> Result<ArrayRef> {
let result = array
.iter()
- .map(|arr| match crate::utils::compute_array_dims(arr)? {
- Some(vector) => Ok(Some(vector.iter().map(|x|
x.unwrap()).product::<u64>())),
- None => Ok(None),
+ .map(|arr| match arr {
+ Some(arr) if arr.is_empty() => Ok(Some(0u64)),
+ arr => match crate::utils::compute_array_dims(arr)? {
+ Some(vector) => {
+ Ok(Some(vector.iter().map(|x|
x.unwrap()).product::<u64>()))
+ }
+ None => Ok(None),
+ },
})
.collect::<Result<UInt64Array>>()?;
Ok(Arc::new(result) as ArrayRef)
diff --git a/datafusion/sqllogictest/test_files/array.slt
b/datafusion/sqllogictest/test_files/array.slt
index 66503c957c..8eb351bb77 100644
--- a/datafusion/sqllogictest/test_files/array.slt
+++ b/datafusion/sqllogictest/test_files/array.slt
@@ -5181,12 +5181,17 @@ select cardinality(arrow_cast([[1, 2], [3, 4], [5, 6]],
'FixedSizeList(3, List(I
query II
select cardinality(make_array()), cardinality(make_array(make_array()))
----
-NULL 0
+0 0
+
+query II
+select cardinality([]), cardinality([]::int[]) as with_cast
+----
+0 0
query II
select cardinality(arrow_cast(make_array(), 'LargeList(Int64)')),
cardinality(arrow_cast(make_array(make_array()), 'LargeList(List(Int64))'))
----
-NULL 0
+0 0
#TODO
#https://github.com/apache/datafusion/issues/9158
@@ -5195,6 +5200,12 @@ NULL 0
#----
#NULL 0
+# cardinality of NULL arrays should return NULL
+query II
+select cardinality(NULL), cardinality(arrow_cast(NULL, 'LargeList(Int64)'))
+----
+NULL NULL
+
# cardinality with columns
query III
select cardinality(column1), cardinality(column2), cardinality(column3) from
arrays;
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]