Is there support for specifying coordinates instead of having to manually select with the mouse?

[FM1337]

Title is self-explanatory but since there isn't an API for this package and the help output for there CLI doesn't indicate that there is a way to specify coordinates, I'm hoping there might just be something hidden that would allow it?

[dynobo]

Hi @FM1337 , thanks for your question.

Unfortunately no, that's not possible at the moment.

May I ask, what your use-case is, to better understand your requirements?

[FM1337]

So it turns out I might not actually need your library/app for this (I went with pytesseract and PIL's ImageGrab and now I just need a way to automatically find a window's coordinates and then mathematically determine each letter's position because tesseract seemingly can't figure out the grid by itself).

Originally I was looking for a way to read text from a screen and while this could do it, it required manual interaction whereas I needed to automate it for a script to try and figure out if the developers of Wordament had either messed up a level accidently or intentionally by running through a gigantic list of all possible words including ones that weren't in the official scrabble wordlist automatically rather than manually having to do it by hand. (All because I'm at 248/250 words in this one level and the last two according to every solver I've tried (including the one I wrote) do not work)

[dynobo]

Wow, interesting use case, thanks for sharing!

I'm glad you found a solution for the OCR part. Glueing some screenshot tool to plain tesseract (or pytesseract) whould have been my recommendation here, too.

Regarding grid-detection, I've planned to add something similar to NormCap (table detection, see #17), but it turned out to be quite complex to do this in a robust way.
In your case, the grid is probably with fixed dimension, then manually measure top-left and cell width/height and iterating over that might be a quick win.

Good luck! 👍