output_parser_pydantic_multiple_examples.py

import os
from dotenv import load_dotenv

load_dotenv()

"""
Here is a sample code for Pydantic class to process multiple outputs. It requests the model to 
suggest a list of words and present the reasoning behind each proposition.
Asks the model to present its reasoning, and the suggestion class declares a new output named reasons. 
Also, the validator function manipulates the output to ensure every reasoning ends with a dot. Another 
use case of the validator function could be output manipulation.
"""

from langchain.output_parsers import PydanticOutputParser
from pydantic import BaseModel, Field, FieldValidationInfo, field_validator
from typing import List


# Define your desired data structure.
class Suggestions(BaseModel):
    words: List[str] = Field(description="list of substitue words based on context")
    reasons: List[str] = Field(
        description="the reasoning of why this word fits the context"
    )

    @field_validator("words")
    def not_start_with_number(cls, info: FieldValidationInfo):
        for item in info:
            if item[0].isnumeric():
                raise ValueError("The word can not start with numbers!")
        return info

    @field_validator("reasons")
    def end_with_dot(cls, info: FieldValidationInfo):
        for idx, item in enumerate(info):
            if item[-1] != ".":
                info[idx] += "."
        return info


parser = PydanticOutputParser(pydantic_object=Suggestions)

from langchain.prompts import PromptTemplate

template = """
Offer a list of suggestions to substitute the specified target_word based on the presented context and the reasoning for each word.
{format_instructions}
target_word={target_word}
context={context}
"""

prompt = PromptTemplate(
    template=template,
    input_variables=["target_word", "context"],
    partial_variables={"format_instructions": parser.get_format_instructions()},
)

model_input = prompt.format_prompt(
    target_word="behaviour",
    context="The behaviour of the students in the classroom was disruptive and made it difficult for the teacher to conduct the lesson.",
)

from langchain.llms import OpenAI

model = OpenAI(model_name="text-davinci-003", temperature=0.0)

output = model(model_input.to_string())

print(parser.parse(output))