Fix async_udf batch size behaviour (apache#18819)

## 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 apache#123` indicates that this PR will close issue apache#123.
-->

- Closes apache#18822.

## 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.
-->

This PR fixes the bug outlined in the issue, we shouldn't use
`ColumnarValue::values_to_arrays` on the batches collected in
`async_scalar_function.rs`.

## 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.
-->

Added a test to cover this behaviour and fixed the issue in the async
scalar function physical expression.

## 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)?
-->

Yes, I added a new `user_defined_async_scalar_functions.rs` test file
similar to `user_defined_scalar_functions.rs` which contains a test that
covers this behaviour.

## Are there any user-facing changes?

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

Yes

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

---------

Co-authored-by: Shiv Bhatia <sbhatia@palantir.com>

Author: 2 people

datafusion/core/tests/user_defined/mod.rs

@@ -15,6 +15,9 @@
 // specific language governing permissions and limitations
 // under the License.
 
+/// Tests for user defined Async Scalar functions
+mod user_defined_async_scalar_functions;
+
 /// Tests for user defined Scalar functions
 mod user_defined_scalar_functions;
 

datafusion/core/tests/user_defined/user_defined_async_scalar_functions.rs

@@ -0,0 +1,139 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.  See the NOTICE file
+// distributed with this work for additional information
+// regarding copyright ownership.  The ASF licenses this file
+// to you under the Apache License, Version 2.0 (the
+// "License"); you may not use this file except in compliance
+// with the License.  You may obtain a copy of the License at
+//
+//   http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied.  See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+use std::sync::Arc;
+
+use arrow::array::{Int32Array, RecordBatch, StringArray};
+use arrow::datatypes::{DataType, Field, Schema};
+use async_trait::async_trait;
+use datafusion::prelude::*;
+use datafusion_common::{assert_batches_eq, Result};
+use datafusion_expr::async_udf::{AsyncScalarUDF, AsyncScalarUDFImpl};
+use datafusion_expr::{
+    ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility,
+};
+
+// This test checks the case where batch_size doesn't evenly divide
+// the number of rows.
+#[tokio::test]
+async fn test_async_udf_with_non_modular_batch_size() -> Result<()> {
+    let num_rows = 3;
+    let batch_size = 2;
+
+    let schema = Arc::new(Schema::new(vec![
+        Field::new("id", DataType::Int32, false),
+        Field::new("prompt", DataType::Utf8, false),
+    ]));
+
+    let batch = RecordBatch::try_new(
+        schema.clone(),
+        vec![
+            Arc::new(Int32Array::from((0..num_rows).collect::<Vec<i32>>())),
+            Arc::new(StringArray::from(
+                (0..num_rows)
+                    .map(|i| format!("prompt{i}"))
+                    .collect::<Vec<_>>(),
+            )),
+        ],
+    )?;
+
+    let ctx = SessionContext::new();
+    ctx.register_batch("test_table", batch)?;
+
+    ctx.register_udf(
+        AsyncScalarUDF::new(Arc::new(TestAsyncUDFImpl::new(batch_size)))
+            .into_scalar_udf(),
+    );
+
+    let df = ctx
+        .sql("SELECT id, test_async_udf(prompt) as result FROM test_table")
+        .await?;
+
+    let result = df.collect().await?;
+
+    assert_batches_eq!(
+        &[
+            "+----+---------+",
+            "| id | result  |",
+            "+----+---------+",
+            "| 0  | prompt0 |",
+            "| 1  | prompt1 |",
+            "| 2  | prompt2 |",
+            "+----+---------+"
+        ],
+        &result
+    );
+
+    Ok(())
+}
+
+#[derive(Debug, PartialEq, Eq, Hash, Clone)]
+struct TestAsyncUDFImpl {
+    batch_size: usize,
+    signature: Signature,
+}
+
+impl TestAsyncUDFImpl {
+    fn new(batch_size: usize) -> Self {
+        Self {
+            batch_size,
+            signature: Signature::exact(vec![DataType::Utf8], Volatility::Volatile),
+        }
+    }
+}
+
+impl ScalarUDFImpl for TestAsyncUDFImpl {
+    fn as_any(&self) -> &dyn std::any::Any {
+        self
+    }
+
+    fn name(&self) -> &str {
+        "test_async_udf"
+    }
+
+    fn signature(&self) -> &Signature {
+        &self.signature
+    }
+
+    fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
+        Ok(DataType::Utf8)
+    }
+
+    fn invoke_with_args(&self, _args: ScalarFunctionArgs) -> Result<ColumnarValue> {
+        panic!("Call invoke_async_with_args instead")
+    }
+}
+
+#[async_trait]
+impl AsyncScalarUDFImpl for TestAsyncUDFImpl {
+    fn ideal_batch_size(&self) -> Option<usize> {
+        Some(self.batch_size)
+    }
+    async fn invoke_async_with_args(
+        &self,
+        args: ScalarFunctionArgs,
+    ) -> Result<ColumnarValue> {
+        let arg1 = &args.args[0];
+        let results = call_external_service(arg1.clone()).await?;
+        Ok(results)
+    }
+}
+
+/// Simulates calling an async external service
+async fn call_external_service(arg1: ColumnarValue) -> Result<ColumnarValue> {
+    Ok(arg1)
+}

datafusion/physical-expr/src/async_scalar_function.rs

@@ -16,7 +16,8 @@
 // under the License.
 
 use crate::ScalarFunctionExpr;
-use arrow::array::{make_array, MutableArrayData, RecordBatch};
+use arrow::array::RecordBatch;
+use arrow::compute::concat;
 use arrow::datatypes::{DataType, Field, FieldRef, Schema};
 use datafusion_common::config::ConfigOptions;
 use datafusion_common::Result;
@@ -192,17 +193,21 @@ impl AsyncFuncExpr {
             );
         }
 
-        let datas = ColumnarValue::values_to_arrays(&result_batches)?
+        let datas = result_batches
+            .into_iter()
+            .map(|cv| match cv {
+                ColumnarValue::Array(arr) => Ok(arr),
+                ColumnarValue::Scalar(scalar) => Ok(scalar.to_array_of_size(1)?),
+            })
+            .collect::<Result<Vec<_>>>()?;
+
+        // Get references to the arrays as dyn Array to call concat
+        let dyn_arrays = datas
             .iter()
-            .map(|b| b.to_data())
+            .map(|arr| arr as &dyn arrow::array::Array)
             .collect::<Vec<_>>();
-        let total_len = datas.iter().map(|d| d.len()).sum();
-        let mut mutable = MutableArrayData::new(datas.iter().collect(), false, total_len);
-        datas.iter().enumerate().for_each(|(i, data)| {
-            mutable.extend(i, 0, data.len());
-        });
-        let array_ref = make_array(mutable.freeze());
-        Ok(ColumnarValue::Array(array_ref))
+        let result_array = concat(&dyn_arrays)?;
+        Ok(ColumnarValue::Array(result_array))
     }
 }