GET /gateway/bot timeout can lead to NoneType crash in client startup

[Alkapuce]

Summary

In qq-botpy 1.2.1, if GET https://api.sgroup.qq.com/gateway/bot times out during startup, the HTTP layer logs the timeout and returns None. Then client.py assumes the return value is a dict and immediately indexes self._ws_ap["session_start_limit"], causing:

TypeError: 'NoneType' object is not subscriptable

Observed Flow

Startup path:

• client.start()
• _bot_login()
• self._ws_ap = await self.api.get_ws_url()
• timeout happens in BotHttp.request()
• timeout is logged, but no exception is raised to the caller
• _bot_login() indexes self._ws_ap["session_start_limit"]

Relevant Behavior

In local source for qq-botpy 1.2.1:

botpy/http.py

• catches asyncio.TimeoutError
• logs 请求超时，请求连接: https://api.sgroup.qq.com/gateway/bot
• returns None

botpy/client.py

• immediately does:

self._ws_ap = await self.api.get_ws_url()
max_async = self._ws_ap["session_start_limit"]["max_concurrency"]

Actual Failure

请求超时，请求连接: https://api.sgroup.qq.com/gateway/bot
TypeError: 'NoneType' object is not subscriptable

Expected Behavior

One of these should happen instead:

• retry transient timeout automatically
• raise a clear exception from BotHttp.request() after retry budget is exhausted
• validate /gateway/bot response before indexing and raise a structured startup error

Suggested Fix

At minimum:

1. do not silently return None on timeout in BotHttp.request()
2. either retry or raise an exception
3. in _bot_login(), validate self._ws_ap before indexing session_start_limit

This would turn a transient network issue into a recoverable or at least diagnosable error, instead of an opaque NoneType crash.

Notes

This was observed in production usage via AstrBot 4.22.0 bundling qq-botpy 1.2.1, but the root cause appears to be in qq-botpy itself.