Processing of TOC objects in Word Documents DOCX fails

[w0o]

Bug

We are seeing an odd behavior where the processing of a TOC table in a word document fails without any errors with the resulting document missing the content that was originally in the TOC.
What we have tried:

• Using OCR fails with the TOC content being omitted.
• Exporting to PDF (using Word) and then using docling to convert to markdown works as expected with no content omissions.

Steps to reproduce

Use attached minimal example docx file and run:

docling sample.docx

resulting in the attached Markdown file which has the TOC content missing.

Docling version

Docling version: 2.13.0
Docling Core version: 2.12.1
Docling IBM Models version: 3.1.0
Docling Parse version: 3.0.0

Python version

Python 3.11.11

Note: all shared samples and publicly available documents.
sample.md
sample.docx

[w0o]

any updates on this team?

[andretost]

I am seeing the same with one of my documents. The TOC is generated by Word. I have other documents where the TOC is manually added, i.e. with plain text, and that works, of course.

I have experimented with the DocItemLabel labels in the save_to_markdown method. Supposedly, the label DocItemLabel.DOCUMENT_INDEX should cause the ToC to be included, but it doesn't.

[w0o]

any news on this team? Did we manage to reproduce this and have an idea of what is happening?

[maxmnemonic]

I'm looking into this issue @w0o !

[maxmnemonic]

@w0o @andretost
support for Table Of Contents in docx backend is implemented in this PR: #853