As @Chuck Connell​ said can you share more of your source json as that example is not json. Additionally flatten is usually to change something like {"status": {"A": 1,"B": 2}} to {"status.A": 1, "status.B": 2} which can be done easily with spark dataframes