fix(telegram): add explicit read_timeout for get_updates polling (#173)

On macOS, when the machine sleeps and wakes up (e.g. closing and
reopening a MacBook), the existing long-polling connection can become
stale. PTB may then reuse this dead connection from its pool, and
without an explicit client-side read_timeout it can hang indefinitely,
leaving the bot completely unresponsive.

Set an explicit HTTPXRequest(read_timeout=30) for get_updates so that
hanging connections are capped at 30s before being discarded and
retried.

Author: oilbeater

src/bub/channels/telegram.py

@@ -12,6 +12,7 @@
 from telegram import Bot, Message, Update
 from telegram.ext import Application, CommandHandler, ContextTypes, filters
 from telegram.ext import MessageHandler as TelegramMessageHandler
+from telegram.request import HTTPXRequest
 
 from bub.channels.base import Channel
 from bub.channels.message import ChannelMessage, MediaItem, MediaType
@@ -169,7 +170,8 @@ async def start(self, stop_event: asyncio.Event) -> None:
             len(self._allow_chats),
             bool(proxy),
         )
-        builder = Application.builder().token(self._settings.token)
+        get_updates_request = HTTPXRequest(read_timeout=30)
+        builder = Application.builder().token(self._settings.token).get_updates_request(get_updates_request)
         if proxy:
             builder = builder.proxy(proxy).get_updates_proxy(proxy)
         self._app = builder.build()