Skip to main content
In this guide, you’ll learn how to convert large language model (LLM) responses into native Python structures using LangChain’s output parsers. We’ll demonstrate:
  • Turning a numbered list into a comma-separated string
  • Parsing that string into a Python list
This approach makes LLM outputs easy to manipulate in your codebase.

​
Table of Contents

  1. Prerequisites
  2. Initializing LangChain and OpenAI
  3. Generating an Unstructured List
  4. Formatting as CSV with CommaSeparatedListOutputParser
  5. Parsing the CSV into a Python List
  6. Next Steps
  7. References

​
Prerequisites

  • Python 3.7+
  • An OpenAI API key
  • langchain installed
pip install langchain langchain-openai

​
Initializing LangChain and OpenAI

Import the core components: 
from langchain_openai import OpenAI
from langchain.prompts import PromptTemplate
from langchain.output_parsers import CommaSeparatedListOutputParser
from langchain.output_parsers.list import ListOutputParser
Create the OpenAI LLM client: 
llm = OpenAI()

​
Generating an Unstructured List

Define a simple prompt template without format constraints: 
prompt = PromptTemplate(
    template="List 3 {things}.",
    input_variables=["things"]
)
Invoke the model to list three World Cup cricket teams: 
response = llm.invoke(input=prompt.format(things="countries that play cricket in the World Cup"))
print(response)
Output might look like: 
1. India
2. Australia
3. England
This human-readable list is great for display but difficult to process in scripts. Next, we’ll add format instructions for a comma-separated response.

​
Formatting as CSV with CommaSeparatedListOutputParser

  1. Instantiate the CSV parser and retrieve its instructions: 
    output_parser = CommaSeparatedListOutputParser()
format_instructions = output_parser.get_format_instructions()
print(format_instructions)
    Expected instruction: 
    Your response should be a list of comma separated values, eg: `foo, bar, baz`
  2. Embed these instructions into a new prompt template: 
    prompt_with_format = PromptTemplate(
    template="List 3 {things}.\n{format_instructions}",
    input_variables=["things"],
    partial_variables={"format_instructions": format_instructions}
)
  3. Invoke the LLM with the CSV constraint: 
    final_prompt = prompt_with_format.format(things="countries that play cricket in the World Cup")
output = llm.invoke(input=final_prompt).strip()
print(output)
    Now the model returns: 
    India, Australia, England

​
Parsing the CSV into a Python List

Convert the raw string into a Python list: 
# Before parsing
print(type(output))  # <class 'str'>

# Parse into list
things = output_parser.parse(output)
print(things)        # ['India', 'Australia', 'England']
print(type(things))  # <class 'list'>
Now you can work with things directly in your application.

​
Next Steps

Explore more advanced parsers for structured outputs:
Parser TypeUse CaseExample
CommaSeparatedListOutputParserSimple lists in CSVfoo, bar, baz
ListOutputParserNumbered or bullet lists1. foo\n2. bar\n
JSONOutputParserComplex nested data structures{ "name": "Alice", "age": 30 }
You can также try the JSONOutputParser to extract richer data types from LLM responses.

​
References

Watch Video