使用Pytest测试Discord Bot时触发AttributeError问题求助

问题背景 #

我开发了一个Discord Bot，逻辑是先通过bot.get_guild()从缓存获取服务器（guild），如果拿不到就调用fetch_guild接口去获取。对应的工具函数代码如下：

async def get_or_fetch_guild(obj_id, bot, guild):
    guild = bot.get_guild(obj_id)
    status = 'SUCCESS'
    status_code = 200
    if guild is None:
        try:
            guild = await bot.fetch_guild(obj_id)
        except discord.errors.NotFound:
            guild = None
            status = 'INVALID_GUILD_ID'
            status_code = 400
        except discord.errors.Forbidden:
            guild = None
            status = 'FORBIDDEN'
            status_code = 403
        except discord.errors.HTTPException:
            guild = None
            status = 'HTTP_EXCEPTION'
            status_code = 400
    return guild, status, status_code

这段代码正常运行没问题，但给调用它的API写Pytest测试用例时，出现了如下错误：

if not self._global_over.is_set():  
E       AttributeError: '_MissingSentinel' object has no attribute 'is_set'
venv/lib/python3.11/site-packages/discord/http.py:605: AttributeError

对应的API接口代码：

@router.get("/guild/{guild_id}/roles", response_model=role_schema.AllRoleData)
async def get_all_roles(
    guild_id: int,
):
    print("get all roles api called")
    guild, status, status_code = await get_or_fetch(guild_id, "guild", bot)

    # try to get the guild through api call if not found in cache
    if guild is None:
        print(f"Guild not found with id: {guild_id}")
        raise HTTPException(status_code=status_code, detail=status)

触发错误的测试用例示例：

@pytest.mark.asyncio
@patch("discord.Client.get_user")
@patch("discord.utils.get")
async def test_assign_role_invalid_guild(
    mock_get_user, mock_get, bot: Tuple[commands.Bot, str], test_client
):
    user = dpytest.get_config().members[0]
    role = dpytest.get_config().guilds[0].roles[0]
    # Create mock user, guild, member, and role objects
    mock_user = MagicMock(spec=discord.User, id=user.id, name="user")
    mock_member = MagicMock(spec=discord.Member, id=user.id, name="member")
    mock_role = MagicMock(spec=discord.Role, id=role.id, name="role")

    # Make member.add_roles an AsyncMock
    mock_member.add_roles = AsyncMock()
    # Set up the return values of get_user, get
    mock_get_user.return_value = mock_user
    mock_get.return_value = mock_role

    # Call the assign_role endpoint
    response = test_client.post(
        "/api/v1/dashboard/assign/roles",
        json={"discord_user_id": user.id, "discord_guild_id": "1234", "role_ids": [role.id]},
        headers={"X-Sizzle-External-Client-Key": settings.SIZZLE_BOT_API_SERVER_KEY},
    )

    assert response.status_code == 400
    assert "detail" in response.json()
    assert response.json()["detail"] == "INVALID_GUILD_ID"

解决方案 #

这个错误的根源是测试用例里的bot实例没有正确初始化Discord HTTP客户端的内部属性，导致调用fetch_guild时访问了未初始化的_global_over。下面是几种可行的解决方法：

1. Mock fetch_guild方法，跳过真实API调用 #

直接在测试用例中mockbot.fetch_guild，让它返回预期结果或抛出指定异常，不用实际发起HTTP请求。示例修改：

@pytest.mark.asyncio
@patch("discord.Client.get_user")
@patch("discord.utils.get")
@patch("discord.Bot.fetch_guild")  # 添加这个mock
async def test_assign_role_invalid_guild(
    mock_fetch_guild, mock_get_user, mock_get, bot: Tuple[commands.Bot, str], test_client
):
    # 模拟找不到guild的场景，抛出NotFound异常
    mock_fetch_guild.side_effect = discord.errors.NotFound(404, "Not Found")
    
    # 原测试代码...

2. 初始化Bot的HTTP客户端属性 #

如果必须让fetch_guild在测试中执行，可以手动初始化Bot的HTTP相关属性：

import asyncio
from discord.http import HTTPClient

# 在测试用例中添加这段初始化代码
bot.http = HTTPClient(bot)
bot.http._global_over = asyncio.Event()
bot.http._global_over.set()

3. 给工具函数添加测试环境分支 #

在工具函数中判断是否处于测试环境，直接返回预设结果，跳过真实的fetch_guild调用：

import os

async def get_or_fetch_guild(obj_id, bot, guild):
    guild = bot.get_guild(obj_id)
    status = 'SUCCESS'
    status_code = 200
    if guild is None:
        # 测试环境下直接返回预设的无效guild结果
        if os.getenv("TEST_ENV") == "true":
            return None, 'INVALID_GUILD_ID', 400
        try:
            guild = await bot.fetch_guild(obj_id)
        # 原异常处理代码...

内容的提问来源于stack exchange，提问作者Aarti Joshi