Docx to markdown with embedded images

[Finn-Sa]

How can I convert docx to markdown with embedded images?

I found this example for pdf with embedded images: https://github.com/DS4SD/docling/blob/main/docs/examples/export_figures.py. But I couldn't make it work for docx.

Thanks for your help!

[sanjana-7890]

Yes I also have the same query bcz for DOCX I observed-

• Image placeholders are not present in markdown.
• Figure tags are not present in doctags.
• Images are not gettting downloaded using "figure_export.py" code.

Although I can see in "docling.backend.msword_backend.py", inline images and pictures are handled. Anyone with the solution?

[PeterStaar-IBM]

There are two solutions to the current problem:

1. Upgrade/fix the export_to_markdown
2. Upgrade/fix the msword backend

I am not sure where the bug is, but for sure we need to hunt it down.

[mailtoshwetha09]

@PeterStaar-IBM will it be possible to reverse back to docx. I have use case where i am converting to markdown where it is extracted markdown and image locally, i have done the translation using LLM, now i need to convert back to docx or pdf. Will it be possible

[maxmnemonic]

@Finn-Sa @sanjana-7890 @PeterStaar-IBM, yesterday we merged to main a PR: #330 that should enable us to export into markdown with embedded images.
We were missing image data export from DOCX, but now it's there.