how to programmatically determine text from one page is garbled?

[animebing]

I am processing thousands of pdf files, in some of them, i find the text from some pages are garbled, i hope to detect it, then use OCR to get texts from these pages.

what i have done: based on https://github.com/pymupdf/PyMuPDF/issues/530 and https://github.com/pymupdf/PyMuPDF/issues/365, i know i can get fonts, then check whether it has /tounicode, i find it works sometimes, but for some fonts, it has no /tounicode, but the text is still normal. Is there something i miss to make it work?