Skip to content

Output schema of the CopyTo logical plan is not correct. #16704

@bert-beyondloops

Description

@bert-beyondloops

Describe the bug

The output schema of the CopyTo logical plan currently directly outputs the underlying schema of it's child.
The physical plan however always returns a 1 row/column count which represents the number of affected rows.

This is inconsistent.

To Reproduce

Following test fails :

#[tokio::test]
async fn test_copy_schema() -> Result<()> {
    let tmp_dir = TempDir::new()?;

    let session_state = SessionStateBuilder::new_with_default_features().build();

    let session_ctx = SessionContext::new_with_state(session_state);

    let schema = Arc::new(Schema::new(vec![Field::new("a", DataType::Int64, true)]));

    // Create and register the source table with the provided schema and data
    let source_table = Arc::new(MemTable::try_new(schema.clone(), vec![vec![]])?);
    session_ctx.register_table("source_table", source_table.clone())?;

    let target_path = tmp_dir.path().join("target.csv");

    let query = format!(
        "COPY source_table TO '{:?}' STORED AS csv",
        target_path.to_str().unwrap()
    );

    let result = session_ctx.sql(&query).await?;
    assert_logical_expr_schema_eq_physical_expr_schema(result).await?;
    Ok(())
}

The schema of the physical output is not the same as the logical plan.

Expected behavior

The CopyTo logical plan returns the same schema as the physical plan, being a 1 column named "count"

Additional context

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions