pythonintermediate

LangChain Pydantic Output Parser

Use LangChain's PydanticOutputParser to reliably parse structured data from LLM text responses.

pythonPress ⌘/Ctrl + Shift + C to copy

from langchain_openai import ChatOpenAI
from langchain.output_parsers import PydanticOutputParser
from langchain_core.prompts import ChatPromptTemplate
from pydantic import BaseModel, Field
from typing import Optional

class JobListing(BaseModel):
    title:       str            = Field(description='Job title')
    company:     str            = Field(description='Company name')
    location:    str            = Field(description='City or Remote')
    salary_min:  Optional[int]  = Field(description='Minimum salary USD')
    salary_max:  Optional[int]  = Field(description='Maximum salary USD')
    skills:      list[str]      = Field(description='Required skills')

parser = PydanticOutputParser(pydantic_object=JobListing)
prompt = ChatPromptTemplate.from_template(
    'Extract job details from:\n{text}\n\n{format_instructions}'
).partial(format_instructions=parser.get_format_instructions())

chain = prompt | ChatOpenAI(model='gpt-4o-mini') | parser

job_text = 'Senior Data Engineer at Acme Corp in New York. $130k-$160k. Requires Python, Spark, dbt.'
result = chain.invoke({'text': job_text})
print(result.model_dump_json(indent=2))