Additionally installed OCR language is rejected by web UI backend

[lehnerpat]

Description
After installing an additional OCR language (for example, Japanese) as described in the docs, the additional language can be used in OCR by setting it as the default, but it cannot be used from the web UI because the backend rejects it as an invalid value.

Expected
Additionally installed languages should be usable from web UI, just like the default languages.

Actual
The additional language shows up in the language selection dropdown for running OCR:

But when you click "Start", the backend responds with a 422 error saying the additional language is not an allowed value for the enum.

Additionally, the UI completely ignores this error and doesn't show any error message :(

Full error payload:

{
    "detail": [
        {
            "type": "enum",
            "loc": [
                "body",
                "lang"
            ],
            "msg": "Input should be 'deu','fra','eng','ita','spa','por' or 'ron'",
            "input": "jpn",
            "ctx": {
                "expected": "'deu','fra','eng','ita','spa','por' or 'ron'"
            }
        }
    ]
}

Browser console screenshot:

Info:

• OS: macOS Sonoma 14.1.2 (23B92), Architecture: Intel (x86_64)
• Browser: Safari 17.1.2 (19616.2.9.11.12)
• Database: SQLite
• Papermerge Version: 3.0

More info about setup:

• Using custom docker image with Japanese language package for tesseract installed, following instructions: https://docs.papermerge.io/3.0/setup/add-ocr-langs/

  • Dockerfile:

    FROM papermerge/papermerge:3.0
    
    # add Japanese OCR language
    RUN apt install tesseract-ocr-jpn

  • Built with: docker build -t mypaper:3.0 -f Dockerfile .

• Using Docker Compose, following instructions: https://docs.papermerge.io/3.0/setup/docker-compose/

  • Changed image to use my custom one (mypaper:3.0)
  • Changed username and password
  • Set additional env var PAPERMERGE__OCR__DEFAULT_LANGUAGE: jpn

[ciur]

Thank you for well structured bug report!

The issue happens because currently the language codes are hardcoded:

1. in backaned
2. in UI
3. and here

The fix would be to, well, just extend current set of hardcoded values with another batch of languages (incl. Japanese).

[ciur]

PR#300 to include extra language codes (incl. Japanese)

Pull request was merged and it will available as part of Papermerge 3.0.1 release.

[ciur]

Fixed in 3.0.2