Add example to spark-expr crate

[andygrove]

What is the problem the feature request solves?

It would be nice to have an example in datafusion-comet-spark-expr that demonstrates how to register Spark-compatible expressions with a DataFusion context and then invoke them via SQL (and/or DataFrame API).

Describe the potential solution

No response

Additional context

No response

[viczsaurav]

I will give it a try

[viczsaurav]

Can we use the following approach? Where would this example reside?

1. Initialize DataFusion Context

2. Define Custom Expression

• Implement a Scalar Function using make_scalar_function.
• Simple string conversion = StringArray => uppercase strings

3. Register the Expression

• Wrap the function in a ScalarUDF.
• Specify input/output data types and function volatility.
• Add it to SessionContext using register_udf().

4. Use in SQL Query

• Read data into a data frame / create it from in-memory Vec
• Execute an SQL query that invokes the registered function.

5. Display Results - Show the transformed output using .show().await?.

[andygrove]

Thanks for looking at this @viczsaurav.

Here is some sample code that may be useful while exploring this.

    #[tokio::test]
    async fn test() {
        let mut session_state = SessionStateBuilder::new().build();
        session_state
            .register_udf(
                create_comet_physical_fun("xxhash64", DataType::Int64, &session_state).unwrap(),
            )
            .unwrap();
        let ctx = SessionContext::new_with_state(session_state);
        let df = ctx.sql("select xxhash64('hello', 42)").await.unwrap();
        df.show().await.unwrap();
    }