utils/openai.py

from dotenv import load_dotenv

load_dotenv()  # take environment variables from .env

import os
from openai import OpenAI
import vcr

client = OpenAI(
    base_url=os.environ.get("OPENAI_API_BASE"),
    api_key=os.environ.get("OPENAI_API_KEY"),
)


def completion(prompt, max_tokens=None, temperature=0):
    _completion = client.completions.create(
        model="gpt-3.5-turbo-instruct",
        prompt=prompt,
        max_tokens=max_tokens,
        temperature=temperature,
    )
    return _completion.choices[0].text.strip()


def chat_completion(message, model="gpt-3.5-turbo", prompt=None, temperature=0):
    # Initialize the messages list
    messages = []

    # Add the prompt to the messages list
    if prompt is not None:
        messages += [{"role": "system", "content": prompt}]

    if message is not None:
        # Add the user's message to the messages list
        messages += [{"role": "user", "content": message}]

    # Make an API call to the OpenAI ChatCompletion endpoint with the model and messages
    _completion = client.chat.completions.create(
        model=model, messages=messages, temperature=temperature
    )

    # Extract and return the AI's response from the API response
    return _completion.choices[0].message.content.strip()


def __vcr():
    return vcr.VCR(
        serializer="yaml",
        cassette_library_dir="tests/fixtures/cassettes",
        record_mode="new_episodes",
        match_on=["uri", "method", "path", "query", "body"],
        record_on_exception=False,
    )


def cassette_for(name):
    filename = name + ".yaml"

    return __vcr().use_cassette(filename, filter_headers=[("authorization", None)])