add unit test

Signed-off-by: Runji Wang <[email protected]>

Author: wangrunji0408

crates/duckdb/src/vscalar/arrow.rs

@@ -6,7 +6,7 @@ use arrow::{
 };
 
 use crate::{
-    core::{DataChunkHandle, LogicalTypeId},
+    core::DataChunkHandle,
     vtab::arrow::{data_chunk_to_arrow, to_duckdb_logical_type, write_arrow_array_to_vector, WritableVector},
 };
 
@@ -331,4 +331,58 @@ mod test {
 
         Ok(())
     }
+
+    #[test]
+    fn test_split_function() -> Result<(), Box<dyn Error>> {
+        struct SplitFunction {}
+
+        impl VArrowScalar for SplitFunction {
+            type State = ();
+
+            fn invoke(_: &Self::State, input: RecordBatch) -> Result<Arc<dyn Array>, Box<dyn std::error::Error>> {
+                let strings = input.column(0).as_any().downcast_ref::<StringArray>().unwrap();
+
+                let mut builder = arrow::array::ListBuilder::new(arrow::array::StringBuilder::with_capacity(
+                    strings.len(),
+                    strings.len() * 10,
+                ));
+
+                for s in strings.iter() {
+                    let s = s.unwrap();
+                    for split_value in s.split(' ').collect::<Vec<_>>() {
+                        builder.values().append_value(split_value);
+                    }
+                    builder.append(true);
+                }
+
+                Ok(Arc::new(builder.finish()))
+            }
+
+            fn signatures() -> Vec<ArrowFunctionSignature> {
+                vec![ArrowFunctionSignature::exact(
+                    vec![DataType::Utf8],
+                    DataType::List(Arc::new(arrow::datatypes::Field::new("item", DataType::Utf8, true))),
+                )]
+            }
+        }
+
+        let conn = Connection::open_in_memory()?;
+        conn.register_scalar_function::<SplitFunction>("split_string")?;
+
+        // Test with single string
+        let batches = conn
+            .prepare("select split_string('hello world') as result")?
+            .query_arrow([])?
+            .collect::<Vec<_>>();
+
+        let array = batches[0].column(0);
+        let list_array = array.as_any().downcast_ref::<arrow::array::ListArray>().unwrap();
+        let values = list_array.value(0);
+        let string_values = values.as_any().downcast_ref::<StringArray>().unwrap();
+
+        assert_eq!(string_values.value(0), "hello");
+        assert_eq!(string_values.value(1), "world");
+
+        Ok(())
+    }
 }