Mutool – all purpose tool for dealing with PDF files | Hacker News Reader

Mutool – all purpose tool for dealing with PDF files

mankier.com

43 pointsrahimnathwani16 comments

Open articleView on HN

This is not a very good link to post - it's a manpage.

A better link (from the developer's git repository):

https://mupdf.readthedocs.io/en/latest/mupdf-command-line.ht...

Source for mutool:

https://github.com/ArtifexSoftware/mupdf/blob/master/source/...

While surfing around their org I found this which has an impressive looking sample: https://github.com/ArtifexSoftware/pdf2docx#sample

Also, strictly speaking their GitHub is labeled as a mirror, although they didn't say "mirror of what" but I believe it's this https://cgit.ghostscript.com/cgi-bin/cgit.cgi/mupdf.git/

"Ingesting Millions of PDFs and why Gemini 2.0 Changes Everything" https://news.ycombinator.com/item?id=42952605

I really wish there was a way to view and edit pdf from the terminal vim style.

You can decompress the streams with mutool and edit the Postscript-like code in any editor, if you want. (That doesn't help a lot with editing text, of course...)

The PDF format isn’t really designed for editing. For viewing, there’s pdftotext [0] or ebook-convert [1].

[0] https://manpages.debian.org/poppler-utils/pdftotext.1.html

[1] https://manpages.debian.org/calibre/ebook-convert.1.html

You’re better off converting to markdown with docling or markupsafe. But, if you’re careful, Qpdf offers an intermediate document format, like pandoc, but pandoc doesn’t read PDF.

Monkt.com offers an online version (no setup or configuration) for docling. Additionally, it has a few extra extraction methods. It could be tested for conversion to markdown.

It's ok, but some less used features have undocumented regressions.

It would be cool if it could also do “LLM style” operations, like summarizing or extracting key points of the documents

not really, very out of scope IMO

If you really have to "mutool draw -F txt somefile.pdf" will extract the text contents, that you can then pipe to a LLM.

how will it handle things like tables let me try

found this pymupdf4llm w a fast google search https://medium.com/@danushidk507/using-pymupdf4llm-a-practic...

If this is using MuPDF, beware that they are very rude and assume ill will! See https://news.ycombinator.com/item?id=40096211

That comment you linked to is yours, and flagged.

It reads to me like you are still carrying a grudge about one rude interaction. I read that whole thread and I don't want to see it repeated here.

Reply on news.ycombinator.com