OCR search for PDF files no longer works The OCR search for PDF files no longer seems to work in bot

Features
1

The OCR search for PDF files no longer seems to work in both the current productive Joplin release win11 and Android 16.**

  • Joplin 3.7.18 (prod, win32)

  • I completely reinstalled Joplin on both assets, deleted all Onedrive data and the data via JEX is imported and synchronized.

  • Excerpt LOG file:

    • PNG files are recognized:

    • 2026-09-18 17:06:11: OcrService: Found 1 resources to process... 2026-09-18 17:06:12: OcrService: Processing resource 1/1: 2ed9d05de75445eb8665ea0064b018c4 (type image/png)...

    • PDF files are not recognized and processed:

      • 2026-09-18 17:12:57: OcrService: Found 0 resources to process...
      • 2026-09-18 17:12:58: OcrService: 0 resources have been processed. 2026-09-18 17:12:58: OcrService:

Does anyone of you have the same problem?

The OCR search worked perfectly in older releases.

I am grateful for every tip!

2

Hi,

I will run a test and see what results I come up with. I have not tried searching for the contents of a PDF lately.

Post results later.

3

Results,

I don't have a lot of PDFs in my Joplin database but I do have a couple that were pre-version 7 and I cannot search for text inside those documents. If I add a new PDF I can search it perfectly.

4

Hi,

Just came up with an idea that works. In the search box I searched for this.

resource:application/pdf

This brings up a list of all of your notes that contain a PDF file attachment. I exported one of the notes that had PDF attachments then deleted the note and then re imported and now it is searchable.

Each of my pdf documents that I failed to scan Do have OCR text associated with them when you right click

5

Thank you very much for your support!

  • I have noticed an almost identical behavior as you have described.

    • It is sufficient to reinsert the PDF file with cut & copy in the note.
    • Then the OCR process starts immediately.

  • However, this is of course not a convenient solution.

@Laurent: Maybe you can help here.

  • Does the behavior correspond to the motto "software works as designed" or is there a bug here.
  • It would be great if the OCR function would work like in previous Joplin releases.
  • But for the rest, I'm an absolute Joplin fan.