Issue: Unable to use `hi` language code in `ocr_lang`

[sumanthaka]

Hi, I'm trying to use a new language code hi for ocr_lang, but I'm encountering an issue where it doesn't seem to be downloadable due to it being disabled.

Is there currently a way to enable or trigger the download from docling-serve?

If so, could you please guide me on where I can configure or pass this setting?

Screenshot for reference:

Thank you!

[athoik]

Hi,

This PR most probably fixes your problem #183 in case you are using tesseract.

Check the correct OCR lang here: https://tesseract-ocr.github.io/tessdoc/Data-Files-in-different-versions.html

It's hin for Hindi.

[dolfim-ibm]

@sumanthaka are you using the Container image or running the python module yourself?

I think the problem is that we only installed english support for EasyOCR in the image.

If you run it with the env DOCLING_SERVE_ARTIFACTS_PATH="" the system will download the model weights at runtime and you should be getting all the languages you need.

[sumanthaka]

The issue is resolved. Running with DOCLING_SERVE_ARTIFACTS_PATH="" allowed the models to download at runtime.

Closing this issue.