How to retry when a parsing error occurs
While in some cases it is possible to fix any parsing mistakes by only looking at the output, in other cases it isn't. An example of this is when the output is not just in the incorrect format, but is partially complete. Consider the below example.
from langchain.output_parsersimport OutputFixingParser
from langchain_core.exceptionsimport OutputParserException
from langchain_core.output_parsersimport PydanticOutputParser
from langchain_core.promptsimport PromptTemplate
from langchain_openaiimport ChatOpenAI, OpenAI
from pydanticimport BaseModel, Field
template="""Based on the user question, provide an Action and Action Input for what step should be taken.
{format_instructions}
Question: {query}
Response:"""
classAction(BaseModel):
action:str= Field(description="action to take")
action_input:str= Field(description="input to the action")
parser= PydanticOutputParser(pydantic_object=Action)
prompt= PromptTemplate(
template="Answer the user query.\n{format_instructions}\n{query}\n",
input_variables=["query"],
partial_variables={"format_instructions": parser.get_format_instructions()},
)
prompt_value= prompt.format_prompt(query="who is leo di caprios gf?")
bad_response='{"action": "search"}'
If we try to parse this response as is, we will get an error:
try:
parser.parse(bad_response)
except OutputParserExceptionas e:
print(e)
Failed to parse Action from completion {"action": "search"}. Got: 1 validation error for Action
action_input
Field required [type=missing, input_value={'action': 'search'}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.9/v/missing
For troubleshooting, visit: https://python.langchain.com/docs/troubleshooting/errors/OUTPUT_PARSING_FAILURE
If we try to use theOutputFixingParser to fix this error, it will be confused - namely, it doesn't know what to actually put for action input.
fix_parser= OutputFixingParser.from_llm(parser=parser, llm=ChatOpenAI())
fix_parser.parse(bad_response)
Action(action='search', action_input='input')
Instead, we can use the RetryOutputParser, which passes in the prompt (as well as the original output) to try again to get a better response.
from langchain.output_parsersimport RetryOutputParser
retry_parser= RetryOutputParser.from_llm(parser=parser, llm=OpenAI(temperature=0))
retry_parser.parse_with_prompt(bad_response, prompt_value)
Action(action='search', action_input='leo di caprio girlfriend')
We can also add the RetryOutputParser easily with a custom chain which transform the raw LLM/ChatModel output into a more workable format.
from langchain_core.runnablesimport RunnableLambda, RunnableParallel
completion_chain= prompt| OpenAI(temperature=0)
main_chain= RunnableParallel(
completion=completion_chain, prompt_value=prompt
)| RunnableLambda(lambda x: retry_parser.parse_with_prompt(**x))
main_chain.invoke({"query":"who is leo di caprios gf?"})
Action(action='search', action_input='leo di caprio girlfriend')
Find out api documentation forRetryOutputParser.