Post #2363293

2026-05-08 11:34 UTC

Hey, #AskFedi... I&#39;m trying to covert PDFs to ODTs (on Debian) with some semblance of formatting retained. So far, I&#39;ve tried LibreOffice Draw, pdf2odt, pdftotext, and Calibre. All of them either give me images or garbled text. Has anyone got this working? Edit: the PDFs in question are scans.

Replies (5)

@Steve@social.coop 2026-05-08 11:45

It doesn't help matters that I'm working on scans. I can OCR them, but the structure of the resulting text is divided up in truly bizarre ways.

Open ##3067955
@robryk@social.wuatek.is 2026-05-08 11:58

@Steve@social.coop do you want to be able to edit them later? If no, would an odt with one full page image per page be satisfactory?

Open ##3067958
@RogerBW@discordian.social 2026-05-08 12:09

@Steve@social.coop IME there is no clean path for this. Only the most recent iterations of the PDF standard even support the concept of "this text in order" rather than "paint a string here".

Open ##3067960
@oblomov@sociale.network 2026-05-08 13:36

@Steve@social.coop are the original PDF actual text PDFs or just scanned documents? In the latter case you'll need some OCR tool

Open ##3067962
@tedel@writing.exchange 2026-05-08 20:14

@Steve@social.coop Pandoc should work most of the times, but for scans you probably want to extract the text, so you may want to take a look at Tesseract OCR too.

Open ##3067964