fix: null cast not valid in substrait round trip (#18414)

## Which issue does this PR close?

<!--
We generally require a GitHub issue to be filed for all bug fixes and
enhancements and this helps us generate change logs for our releases.
You can link an issue to this PR using the GitHub syntax. For example
`Closes #123` indicates that this PR will close issue #123.
-->

- Closes #16272.

## Rationale for this change

- Null with null type were throwing invalid casts in substrait round
trips

<!--
Why are you proposing this change? If this is already explained clearly
in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand
your changes and offer better suggestions for fixes.
-->

## What changes are included in this PR?

- Added a null variation const that allows NULL with NULL types to be
casted properly.
- Made this a UserDefined type from the substrait side of things

<!--
There is no need to duplicate the description in the issue here but it
is sometimes worth providing a summary of the individual changes in this
PR.
-->

## Are these changes tested?
- Yes added unit test in producer/types.rs
- Previously failing tests pass

<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code

If tests are not included in your PR, please explain why (for example,
are they covered by existing tests)?
-->

## Are there any user-facing changes?

<!--
If there are user-facing changes then we may require documentation to be
updated before approving the PR.
-->

<!--
If there are any breaking changes to public APIs, please add the `api
change` label.
-->

Author: gene-bordegaray

datafusion/substrait/src/logical_plan/consumer/types.rs

@@ -17,7 +17,6 @@
 
 use super::utils::{from_substrait_precision, next_struct_field_name, DEFAULT_TIMEZONE};
 use super::SubstraitConsumer;
-use crate::variation_const::FLOAT_16_TYPE_NAME;
 #[allow(deprecated)]
 use crate::variation_const::{
     DATE_32_TYPE_VARIATION_REF, DATE_64_TYPE_VARIATION_REF,
@@ -33,6 +32,7 @@ use crate::variation_const::{
     TIME_64_TYPE_VARIATION_REF, UNSIGNED_INTEGER_TYPE_VARIATION_REF,
     VIEW_CONTAINER_TYPE_VARIATION_REF,
 };
+use crate::variation_const::{FLOAT_16_TYPE_NAME, NULL_TYPE_NAME};
 use datafusion::arrow::datatypes::{
     DataType, Field, Fields, IntervalUnit, Schema, TimeUnit,
 };
@@ -253,6 +253,7 @@ pub fn from_substrait_type(
                         // Kept for backwards compatibility, producers should use IntervalCompound instead
                         INTERVAL_MONTH_DAY_NANO_TYPE_NAME => Ok(DataType::Interval(IntervalUnit::MonthDayNano)),
                         FLOAT_16_TYPE_NAME => Ok(DataType::Float16),
+                        NULL_TYPE_NAME => Ok(DataType::Null),
                         _ => not_impl_err!(
                                 "Unsupported Substrait user defined type with ref {} and variation {}",
                                 u.type_reference,

datafusion/substrait/src/logical_plan/producer/types.rs

@@ -23,12 +23,12 @@ use crate::variation_const::{
     DEFAULT_CONTAINER_TYPE_VARIATION_REF, DEFAULT_INTERVAL_DAY_TYPE_VARIATION_REF,
     DEFAULT_MAP_TYPE_VARIATION_REF, DEFAULT_TYPE_VARIATION_REF,
     DICTIONARY_MAP_TYPE_VARIATION_REF, DURATION_INTERVAL_DAY_TYPE_VARIATION_REF,
-    FLOAT_16_TYPE_NAME, LARGE_CONTAINER_TYPE_VARIATION_REF, TIME_32_TYPE_VARIATION_REF,
-    TIME_64_TYPE_VARIATION_REF, UNSIGNED_INTEGER_TYPE_VARIATION_REF,
-    VIEW_CONTAINER_TYPE_VARIATION_REF,
+    FLOAT_16_TYPE_NAME, LARGE_CONTAINER_TYPE_VARIATION_REF, NULL_TYPE_NAME,
+    TIME_32_TYPE_VARIATION_REF, TIME_64_TYPE_VARIATION_REF,
+    UNSIGNED_INTEGER_TYPE_VARIATION_REF, VIEW_CONTAINER_TYPE_VARIATION_REF,
 };
 use datafusion::arrow::datatypes::{DataType, IntervalUnit};
-use datafusion::common::{internal_err, not_impl_err, plan_err, DFSchemaRef};
+use datafusion::common::{not_impl_err, plan_err, DFSchemaRef};
 use substrait::proto::{r#type, NamedStruct};
 
 pub(crate) fn to_substrait_type(
@@ -42,7 +42,17 @@ pub(crate) fn to_substrait_type(
         r#type::Nullability::Required as i32
     };
     match dt {
-        DataType::Null => internal_err!("Null cast is not valid"),
+        DataType::Null => {
+            let type_anchor = producer.register_type(NULL_TYPE_NAME.to_string());
+            Ok(substrait::proto::Type {
+                kind: Some(r#type::Kind::UserDefined(r#type::UserDefined {
+                    type_reference: type_anchor,
+                    type_variation_reference: DEFAULT_TYPE_VARIATION_REF,
+                    nullability,
+                    type_parameters: vec![],
+                })),
+            })
+        }
         DataType::Boolean => Ok(substrait::proto::Type {
             kind: Some(r#type::Kind::Bool(r#type::Boolean {
                 type_variation_reference: DEFAULT_TYPE_VARIATION_REF,
@@ -377,6 +387,7 @@ mod tests {
     use crate::logical_plan::consumer::tests::test_consumer;
     use crate::logical_plan::consumer::{
         from_substrait_named_struct, from_substrait_type_without_names,
+        DefaultSubstraitConsumer,
     };
     use crate::logical_plan::producer::DefaultSubstraitProducer;
     use datafusion::arrow::datatypes::{Field, Fields, Schema, TimeUnit};
@@ -386,6 +397,7 @@ mod tests {
 
     #[test]
     fn round_trip_types() -> Result<()> {
+        round_trip_type(DataType::Null)?;
         round_trip_type(DataType::Boolean)?;
         round_trip_type(DataType::Int8)?;
         round_trip_type(DataType::UInt8)?;
@@ -474,7 +486,12 @@ mod tests {
         // As DataFusion doesn't consider nullability as a property of the type, but field,
         // it doesn't matter if we set nullability to true or false here.
         let substrait = to_substrait_type(&mut producer, &dt, true)?;
-        let consumer = test_consumer();
+
+        // Get the extensions from the producer so the consumer can look up
+        // any registered user-defined types (like "null" or "f16")
+        let extensions = producer.get_extensions();
+        let consumer = DefaultSubstraitConsumer::new(&extensions, &state);
+
         let roundtrip_dt = from_substrait_type_without_names(&consumer, &substrait)?;
         assert_eq!(dt, roundtrip_dt);
         Ok(())

datafusion/substrait/src/variation_const.rs

@@ -125,3 +125,8 @@ pub const INTERVAL_MONTH_DAY_NANO_TYPE_NAME: &str = "interval-month-day-nano";
 
 /// Defined in <https://github.com/apache/arrow/blame/main/format/substrait/extension_types.yaml>
 pub const FLOAT_16_TYPE_NAME: &str = "fp16";
+
+/// For [`DataType::Null`]
+///
+/// [`DataType::Null`]: datafusion::arrow::datatypes::DataType::Null
+pub const NULL_TYPE_NAME: &str = "null";