The current version implements its own file and http/s readers. It would be nice to use DuckDB's native reader to support HTTPFS, S3, etc.