使用python-telegram-bot开发Telegram Bot遇await表达式字典错误求助

使用python-telegram-bot框架结合asyncio开发Telegram Bot，需求是用户点击"AI"按钮后可向ChatGPT提问并获取回复。处理AI会话消息时，用户发送消息后先检查是否处于AI会话，若是则将消息发送给ChatGPT获取回复，但始终触发错误"object dict can't be used in 'await' expression"，已卡在此问题5小时。

相关代码 #

main.py #

from typing import Final
from telegram import Update, KeyboardButton, ReplyKeyboardMarkup
from telegram.ext import Application, CommandHandler, MessageHandler, filters, ContextTypes
from gpt_integration import get_gpt_response

TOKEN: Final = 'xxxx'
BOT_USERNAME: Final = '@xxxx'

users_in_ai_session = {}


async def start_command(update: Update, context: ContextTypes.DEFAULT_TYPE):
    button_list = [
        [KeyboardButton("/AI"),
         KeyboardButton("/price")]
    ]
    reply_markup = ReplyKeyboardMarkup(
        button_list, resize_keyboard=True, one_time_keyboard=True)
    await update.message.reply_text("Welcome message...", reply_markup=reply_markup)


async def search_price_command(update: Update, context: ContextTypes.DEFAULT_TYPE):
    await update.message.reply_text("Please enter the contract address:")


async def AI_command(update: Update, context: ContextTypes.DEFAULT_TYPE):
    user_id = update.effective_user.id
    users_in_ai_session[user_id] = True
    await update.message.reply_text(
        "You're now chatting with AI. Go ahead, ask me anything!")


async def handle_message(update: Update, context: ContextTypes.DEFAULT_TYPE):
    user_id = update.effective_user.id
    if users_in_ai_session.get(user_id, False):
        user_input = update.message.text
        response = await get_gpt_response(user_input)
        await update.message.reply_text(response)
    else:
        print("Error")


async def error(update: Update, context: ContextTypes.DEFAULT_TYPE):
    print(f'Update {update} caused error {context.error}')

if __name__ == '__main__':
    print('Starting bot')
    app = Application.builder().token(TOKEN).build()
    app.add_handler(CommandHandler('start', start_command))
    app.add_handler(CommandHandler('AI', AI_command))
    app.add_handler(CommandHandler('price', search_price_command))
    app.add_handler(MessageHandler(filters.TEXT, handle_message))
    app.add_error_handler(error)
    print('Fetching updates...')
    app.run_polling()

gpt_integration.py #

import httpx


async def get_gpt_response(message_text):
    async with httpx.AsyncClient() as client:
        response = await client.post(
            'https://api.openai.com/v1/completions',
            headers={
                'Authorization': 'xxxx'
            },
            json={
                'model': 'gpt-3.5-turbo',
                'prompt': message_text,
                'max_tokens': 50
            },
        )
    data = await response.json()
    return data['choices'][0]['text'].strip()

已完成的排查步骤 #

• 确认所有异步函数均正确使用await关键字
• 检查get_gpt_response函数是否返回可await对象
• 查阅Python asyncio文档排查await关键字误用场景

错误原因及解决方法 #

核心原因 #

1. 直接对字典用await：gpt_integration.py中data = await response.json()是错误的，httpx的response.json()方法是同步方法，直接返回字典对象，不需要加await，对字典使用await会触发报错。
2. 接口与模型不匹配：使用gpt-3.5-turbo模型却调用了/v1/completions接口，该接口仅适用于text-davinci-003等旧模型，gpt-3.5-turbo需要调用/v1/chat/completions接口，且参数格式不同，会导致接口调用失败。

修复步骤 #

1. 修正response.json()调用：去掉await关键字，直接获取字典结果：

   data = response.json()
   

2. 适配gpt-3.5-turbo的接口与参数：

   • 将接口路径改为https://api.openai.com/v1/chat/completions
   • 修改请求参数为messages格式，替换原prompt参数
   • 调整返回结果的解析路径，适配chat接口的返回结构
   • 补充Bearer前缀到Authorization头

修复后的gpt_integration.py完整代码 #

import httpx


async def get_gpt_response(message_text):
    async with httpx.AsyncClient() as client:
        response = await client.post(
            'https://api.openai.com/v1/chat/completions',
            headers={
                'Authorization': 'Bearer xxxx'
            },
            json={
                'model': 'gpt-3.5-turbo',
                'messages': [{"role": "user", "content": message_text}],
                'max_tokens': 50
            },
        )
    # 主动抛出HTTP错误，便于排查接口调用问题
    response.raise_for_status()
    data = response.json()
    return data['choices'][0]['message']['content'].strip()

内容的提问来源于stack exchange，提问作者Yanyi