Then just OCR the rendered output. It's cheap and easy these days.

i heard it works best on iphones, best of luck

Apple’s vision framework is actually very good, so effective on-device extraction is really just a few lines of code on iPhone(all Apple device actually).

Good lick OCR-ing images and videos, though. Twitter isn't just text.

Why do I need to OCR images and videos? Why not display those as-is?

Which one do you show for which tweet, if all you have is the OCR (that is, you can't read the markup)?

Apple's vision framework is not limited to text, can detect shapes and other things.

Twitter WebUI loads images and videos when you scroll. Which means you have a good signal what image and video is associated with what tweet.

Doing this requires a fair amount of development time, especially when you can break something else with this and you previously removed all the guardrails.

That didn't stop Facebook doing it (and breaking all kinds of stuff in the process).

Did Facebook lay off half of the company before they did it?

Facebook had far more engineers available to do this sort of work.

They don't do this when they can just disable the app without being logged in.