Using LangChain Output Parsers to get what you want out of LLMs