Docling failing word doc with no comments

[joe-pierce]

Bug

Converting a word doc fails with AttributeError

Steps to reproduce

Convert a word doc that has no comments.

Get the following error traceback:

Traceback (most recent call last):
   File ".venv\Lib\site-packages\docling\pipeline\base_pipeline.py", line 75, in execute
     conv_res = self._build_document(conv_res)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   File ".venv\Lib\site-packages\docling\pipeline\simple_pipeline.py", line 40, in _build_document
     conv_res.document = conv_res.input._backend.convert()
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   File ".venv\Lib\site-packages\docling\backend\msword_backend.py", line 169, in convert
     self._add_comments(self.docx_obj, doc)
   File ".venv\Lib\site-packages\docling\backend\msword_backend.py", line 1772, in _add_comments
     target.comments.append(group_ref)
     ^^^^^^^^^^^^^^^
   File ".venv\Lib\site-packages\pydantic\main.py", line 1026, in __getattr__
     raise AttributeError(f'{type(self).__name__!r} object has no attribute {item!r}')
 AttributeError: 'GroupItem' object has no attribute 'comments'

Docling version

2.72.0

Python version

3.12

[dosubot]

The AttributeError happens because GroupItem objects in Docling don't have a comments attribute—only DocItem objects do. The msword_backend.py code was trying to append to comments on all targets, including GroupItems, which caused the crash.

This has been fixed in a recent update: PR #2834 refactored comment handling to use a safer API that only attaches comments to DocItem instances and avoids this error. If you upgrade to a Docling release after 2026-01-26 (which includes this PR and requires docling-core >=2.58.0), the issue should be resolved. The new implementation also safely handles cases where comments can't be linked, placing them in a COMMENT_SECTION group instead of crashing. You can see more details in the PR #2834 discussion.

If you can't upgrade right now, a minimal workaround is to add a type check before appending to comments in your local msword_backend.py: only append to target.comments if hasattr(target, "comments").

This design—comments only on DocItem, not GroupItem—is intentional in Docling's data model to keep group containers clean and avoid attribute errors.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[joe-pierce]

That PR is what has broken it - I've had to drop to v2.70.0 to resolve as a workaround. To any admins of this repo, please consider disabling 'dosubot', I'm finding it such a hindrance when trying to navigate issues in this repo...