Extract text from screen

[john--]

I was playing around with a client-side OCR last night and ended up with this:
https://github.com/john--/kvm/tree/extract_text

It adds an "Extract text" button to the Action bar

which allows you to select an area, which gets passes through to the tesseract.js OCR.
I was surprised by how simple this was to implement, but the caveat is that tesseract.js seems to introduce a fairly large dependency--so I really doubt the team would want to include this as part of the official code base. Not to mention on discord I found old conversations where OCR was described as "a hack".

But in any case, just wanted to share! Demo here:
https://www.youtube.com/watch?v=BIUonWaRAfc