Fix null input in map_keys/values

[cht42]

Which issue does this PR close?

Closes #14400.

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

[cht42]

cannot use slt tests as we can't cast null to Map
see https://github.com/apache/arrow-rs/blob/a6648fe53bc386eb8f491e68dca3d2bdee6c549c/arrow-schema/src/datatype_parse.rs#L683

[alamb]

I think you can test it using slt with this trick (create a column with non null value and then insert a NULL into it)

DataFusion CLI v44.0.0
> create table foo as values (MAP {'foo': 'bar'});
0 row(s) fetched.
Elapsed 0.037 seconds.

> insert into foo values (NULL);
+-------+
| count |
+-------+
| 1     |
+-------+
1 row(s) fetched.
Elapsed 0.016 seconds.

> select * from foo;
+------------+
| column1    |
+------------+
| {foo: bar} |
| NULL       |
+------------+
2 row(s) fetched.
Elapsed 0.004 seconds.

> select column1, arrow_typeof(column1) from foo;
+------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| column1    | arrow_typeof(foo.column1)                                                                                                                                                                                                                                                                                                                            |
+------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| {foo: bar} | Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, false) |
| NULL       | Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, false) |
+------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2 row(s) fetched.
Elapsed 0.006 seconds.

[cht42]

sweet, will update

[alamb]

Thank you very much @cht42 -- this makes sense to me 🙏

I think it would be really nice to update the tests to use slt (I left a suggestion on how to do so ) but not strictly required.

[alamb]

I think you can test it using slt with this trick (create a column with non null value and then insert a NULL into it)

DataFusion CLI v44.0.0
> create table foo as values (MAP {'foo': 'bar'});
0 row(s) fetched.
Elapsed 0.037 seconds.

> insert into foo values (NULL);
+-------+
| count |
+-------+
| 1     |
+-------+
1 row(s) fetched.
Elapsed 0.016 seconds.

> select * from foo;
+------------+
| column1    |
+------------+
| {foo: bar} |
| NULL       |
+------------+
2 row(s) fetched.
Elapsed 0.004 seconds.

> select column1, arrow_typeof(column1) from foo;
+------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| column1    | arrow_typeof(foo.column1)                                                                                                                                                                                                                                                                                                                            |
+------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| {foo: bar} | Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, false) |
| NULL       | Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, false) |
+------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2 row(s) fetched.
Elapsed 0.006 seconds.

[xudong963]

Thanks @cht42