PyMuPDF extraction

[meghanaviyyapu]

Can PyMuPDF extract PDF with this text
microsoft.pdf

[JorjMcKie]

What kept you keep away from trying this yourself:

[JorjMcKie]

OCR engines try to match image content with characters they know about. Every OCR will of course recognize what "o" means and return an o (the Latin character).
This does not mean that this is the correct, intended original character encoded by the PDF creator!
OCR engines do not semantically understand what they are extracting!!!

[JorjMcKie]

If an OCR detects some dirty spot on a page, it will try to match one of the characters it knows about. It cannot understand what it is looking at.

[meghanaviyyapu]

I haven't used ocr to extract the above text. I have used PyMuPDF page.get_text() for the PDF "ocr.pdf" and the text extraction was the same as in input PDF whereas for the PDF uploaded yesterday o was missing. I understand that it is difficult to identify which characters can be extracted and which cannot. How can it be decided when to use ocr for text based multi lingual content PDFs?

[JorjMcKie]

Trial an error!

Please also accept that while OCR-ing may produce an acceptably looking result, it may still not produce the intended original text - see above example of that "o" symbol: if an OCR identifies it as a Latin character then this may look good, but is still wrong compared to the original text's meaning.

[meghanaviyyapu]

May I know if you are referring to the below?