uxfion/my_tool

My Tool

This is a Python package named my_tool. It is used for my own convenience.

Quick Start

Setup

pip install git+https://github.com/uxfion/my_tool.git
wget https://raw.githubusercontent.com/uxfion/my_tool/main/.env.example -O .env
# config your own api
vim .env

Use in Python

• Necessary configuration

from my_tool.openai_tool import ChatClient, AsyncChatClient
import os
from dotenv import load_dotenv


load_dotenv()
config = {
    "api_key": os.environ.get("OPENAI_API_KEY"),
    "base_url": os.environ.get("OPENAI_BASE_URL"),
    "model": "gpt-3.5-turbo-1106",
    "concurrency": 10,
    "rpm": 300
}

• Single chat

client = ChatClient(**config)

# single chat
normal_prompt = "鲁迅为什么打周树人？"
print(client.chat(normal_prompt))

# single chat with JSON response
content = "1.这是第一句话。2.这是第二句话。3.这是第三句话。"
prompt = f"{content}\n将这段话断句，不要修改原来段落中的任何字符。"
json_schema = '''{
    "sents": "sents_list",
    "count": "len_of_sents_list"
}'''
# return as a dict
result = client.chat_json(prompt, json_schema)
print(result)
count = result["count"]
sents = result["sents"]
print(count)
print(sents)

• Async chat

import asyncio


async def main():
    async_client = AsyncChatClient(**config)

    prompts = [
        "Hello, what's the weather today?",
        "Tell me a joke.",
        "How does quantum computing work?",
        "越南首都在哪？",
        "瑞士首都在哪？"
    ]

    responses = await async_client.batch_chat(prompts)

    print("\n\n\n")
    for prompt, response in zip(prompts, responses):
        print("============")
        print(f"Prompt: {prompt}")
        print(f"Response: {response}")


if __name__ == "__main__":
    asyncio.run(main())

• if you want to control asyncio by yourself, you can call async_client.chat() and async_client.chat_json directly

async def main():
    async_client = AsyncChatClient(**config)

    prompts = [
        "Hello, what's the weather today?",
        "Tell me a joke.",
        "How does quantum computing work?",
        "越南首都在哪？",
        "瑞士首都在哪？"
    ]

    tasks = []
    for prompt in prompts:
        task = async_client.chat(prompt)
        tasks.append(task)
    responses = await asyncio.gather(*tasks)

    print("\n\n\n")
    for prompt, response in zip(prompts, responses):
        print("============")
        print(f"Prompt: {prompt}")
        print(f"Response: {response}")


if __name__ == "__main__":
    asyncio.run(main())