fix: spark crc32 custom nullability (#19271)

## 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 #19157 

## Rationale for this change

<!--
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.
-->
The `crc32` UDF was using the default return_type implementation which
does not preserve nullability information [Spark
CRC32](https://github.com/apache/spark/blob/master/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/hash.scala#L213-L240)
* Only returns the data type (Int64)
* Doesn't consider nullability of inputs
* Would always mark output as non-nullable


## What changes are included in this PR?

<!--
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.
-->
* Implemented `return_field_from_args`: Creates a field with Int64 type
and correctly propagates nullability from input fields and scalar
arguments
* Updated `return_type`: Now returns an error directing users to use
return_field_from_args instead
* Added necessary imports: `Field`, `FieldRef`, and `ReturnFieldArgs` to
support the new implementation
* Added comprehensive nullability tests: Verifies that nullable inputs,
non-nullable inputs, and null scalar literals are handled correctly

## Are these changes tested?

<!--
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)?
-->
* Non-nullable Binary input produces non-nullable Int64 output
* Nullable Binary input produces nullable Int64 output
* Null scalar literal (e.g., crc32(NULL)) produces nullable Int64 output
* Data type is correctly set to Int64 in all cases

## 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.
-->
This is a bug fix that corrects schema metadata only, it does not change
the actual computation or introduce any breaking changes to the API.

Author: watanaberin

datafusion/spark/src/function/hash/crc32.rs

@@ -19,7 +19,7 @@ use std::any::Any;
 use std::sync::Arc;
 
 use arrow::array::{ArrayRef, Int64Array};
-use arrow::datatypes::DataType;
+use arrow::datatypes::{DataType, Field, FieldRef};
 use crc32fast::Hasher;
 use datafusion_common::cast::{
     as_binary_array, as_binary_view_array, as_fixed_size_binary_array,
@@ -29,8 +29,8 @@ use datafusion_common::types::{NativeType, logical_string};
 use datafusion_common::utils::take_function_args;
 use datafusion_common::{Result, internal_err};
 use datafusion_expr::{
-    Coercion, ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature,
-    TypeSignatureClass, Volatility,
+    Coercion, ColumnarValue, ReturnFieldArgs, ScalarFunctionArgs, ScalarUDFImpl,
+    Signature, TypeSignatureClass, Volatility,
 };
 use datafusion_functions::utils::make_scalar_function;
 
@@ -75,7 +75,16 @@ impl ScalarUDFImpl for SparkCrc32 {
     }
 
     fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
-        Ok(DataType::Int64)
+        internal_err!("return_field_from_args should be used instead")
+    }
+
+    fn return_field_from_args(&self, args: ReturnFieldArgs) -> Result<FieldRef> {
+        let nullable = args.arg_fields.iter().any(|f| f.is_nullable())
+            || args
+                .scalar_arguments
+                .iter()
+                .any(|scalar| scalar.is_some_and(|s| s.is_null()));
+        Ok(Arc::new(Field::new(self.name(), DataType::Int64, nullable)))
     }
 
     fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
@@ -122,3 +131,43 @@ fn spark_crc32(args: &[ArrayRef]) -> Result<ArrayRef> {
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use datafusion_common::ScalarValue;
+
+    #[test]
+    fn test_crc32_nullability() -> Result<()> {
+        let crc32_func = SparkCrc32::new();
+
+        // non-nullable field should produce non-nullable output
+        let field_not_null = Arc::new(Field::new("data", DataType::Binary, false));
+        let result = crc32_func.return_field_from_args(ReturnFieldArgs {
+            arg_fields: std::slice::from_ref(&field_not_null),
+            scalar_arguments: &[None],
+        })?;
+        assert!(!result.is_nullable());
+        assert_eq!(result.data_type(), &DataType::Int64);
+
+        // nullable field should produce nullable output
+        let field_nullable = Arc::new(Field::new("data", DataType::Binary, true));
+        let result = crc32_func.return_field_from_args(ReturnFieldArgs {
+            arg_fields: &[field_nullable],
+            scalar_arguments: &[None],
+        })?;
+        assert!(result.is_nullable());
+        assert_eq!(result.data_type(), &DataType::Int64);
+
+        // null scalar value - user input literal NULL
+        let scalar_null = ScalarValue::Binary(None);
+        let result = crc32_func.return_field_from_args(ReturnFieldArgs {
+            arg_fields: &[field_not_null],
+            scalar_arguments: &[Some(&scalar_null)],
+        })?;
+        assert!(result.is_nullable());
+        assert_eq!(result.data_type(), &DataType::Int64);
+
+        Ok(())
+    }
+}