I think you have misunderstood what that does. See this thread.
I suggest you ask a new question, saying: I am repeatedly fetching rows from a database. I insert them into elasticsearch using the unique key as the document_id. For any fields not on the current document I want to add any missing columns to the exiting documents. For fields that do exist on the current document (including hashes) I want to overwrite them. How can I do this?