Handling httpx exceptions in Application.builder based app

[b-ferd]

I've toyed around a bit with the error handler example code errorhandlerbot.py, for Application.builder based PTB applications. The code comments seem to indicate that the handler is only called on errors that happen during the registered message handler callback functions, and the Exception Handling Docu indicates this as well. On the Handling Network Errors Page, there is some hinting towards catching exceptions, but no specific suggestion how to catch exceptions raised by httpx.

With the raw bot, this was easy: a try-catch-block around bot=Bot() and around await bot.get_updates() .
But how do I catch this with the Application.builder based bot? The Network Error info page says:
"When a network error occurs, be prepared to catch the raised exception and handle it according to your policy ... or use PTBs built-in mechanism for exception handling." ... well, yes... but how do I catch httpx exceptions when I'm not the caller?

Background: aim is to detect network outage errors asap (i.e. I don't want to tweak connection timeout settings to cover things up), log a brief warning message, recover automatically (PTB already does this)...

Is there already example code for catching httpx exceptions in Application.builder based applications that I missed?
Any help or suggestions, as usual, are very much appreciated.

[b-ferd]

Solved itself!
I found this posting 3430 by Bibo-Joshi.
This is the working code snippet, for reference:

def error_callback(error, application):
	application.create_task(application.process_error(update=None, error=error))

async def botloop() -> NoReturn:
	application = Application.builder().token(bot_token).build()
	application.add_handler(CommandHandler("start", start))
	application.add_error_handler(error_handler)
	await application.initialize()
	await application.updater.start_polling(allowed_updates=Update.ALL_TYPES, \
		error_callback=functools.partial(error_callback, application=application))
	await application.start()
	
	while True:
		await asyncio.sleep(1)
	
	# dropped from the loop -> exit. shutdown bot.
	await application.Updater.stop()
	await application.stop()
	await application.shutdown()

This will pass network errors from httpx into the error_handler() function, there they can be detected using if type(context.error) == telegram.error.NetworkError.

[Bibo-Joshi]

Glas you found your answer. Just one hint: isinstance is usually considered more pythonic than exactly matching the type and also catches subclasses :)

[b-ferd]

btw... suggest to add this info to the Networking Error page...

[Bibo-Joshi]

You are welcome to edit the wiki pages where you a seed :)