PyMuPDF won't load a page from a PDF that doesn't seem to have a problem: pymupdf.mupdf.FzErrorArgument: code=4: key is not a name (dictionary)

[jdjebi]

Description

Hello, I'm not very familiar with PDF manipulation, but I'm using PyMuPDF to load PDF pages with the aim of converting them to images.

Example:

import pymupdf

def to_image(doc_bytes):

     doc_repr = pymupdf.open(stream=doc_bytes)

     results = []

     for pnum in range(doc_repr.page_count):
          page = doc_repr.load_page(pnum) # <= Raised pymupdf.mupdf.FzErrorArgument: code=4: key is not a name (dictionary)
          # ...
          # page to image logics
          # ...

     return results

So far, I've never had any problems processing 1500 PDFs, and I came across a PDF that produces this exception: pymupdf.mupdf.FzErrorArgument: code=4: key is not a name (dictionary)

I haven't found a solution by searching the web. The PDF displays correctly in my file explorer, but with pymupdf, I get the above exception.

For privacy reasons, I cannot share the PDF, which could contain errors in its structure.

Do you have a solution or just an explanation of the potential causes of this exception or other suggestions?

Description

PyMuPDF version
1.26.1

Operating system
MacOS

Python version
3.12.4

[JorjMcKie]

You did not provide the PDF! We cannot do anything without a reproducer. You can use my e-mail if you have confidentiality concerns.

[jdjebi]

Got it, I've forwarded it to you.

[JorjMcKie]

Looking now.

[JorjMcKie]

This is an upstream problem: MuPDF cannot process the file. I need to involve the MuPDF team here.

[jdjebi]

Well noted. Thanks for your review 👌.

[JorjMcKie]

Here is the MuPDF bug link: https://bugs.ghostscript.com/show_bug.cgi?id=708605

[JorjMcKie]

Fixed in version 1.26.3. Here is a short confirmation snippet:

[gardner]

If updating your pymupdf dependency doesn't work, you can rewrite the pdf using: ocrmypdf --skip-text input.pdf output.pdf and that file seems to load OK 👍