feat(docx): extract VML images with v:imagedata elements

[ceberam]

Summary

This PR adds support for extracting VML images with v:imagedata elements from Word documents. While VML shapes (vector graphics like rectangles, ovals, lines) were already supported for positioning, embedded images referenced via VML v:imagedata elements were not being extracted.

Problem

When converting Word documents containing VML images (commonly used for embedded Visio drawings, EMF/WMF files, and OLE objects), these images were missing from the output. The backend only detected DrawingML images (using a:blip elements) but missed VML images (using v:imagedata elements).

Solution

1. VML Image Detection - Added XPath expression to detect v:imagedata elements
2. VML Image Handler - Implemented _handle_vml_pictures() method that:
   • Extracts images from VML v:imagedata relationship IDs
   • Attempts direct PIL loading for common formats (PNG, JPEG, etc.)
   • Falls back to LibreOffice conversion for EMF/WMF formats
3. Code Consolidation - Refactored duplicate code across all image handlers:
   • _get_image_from_relationship() - Unified image retrieval from relationship IDs
   • _add_picture_to_doc() - Unified picture element addition
   • _convert_elements_via_docx() - Unified DOCX→PDF→PNG conversion (supports both single and multiple elements)

Key Distinction

• Before: VML namespace detection was only used for positioning vector shapes
• After: VML v:imagedata elements are now detected and their embedded images are extracted

Testing

• ✅ All 25 existing tests pass
• ✅ New test document docx_vml_images.docx with 2 VML images and ground truth
• ✅ Test document programmatically created with proper VML structure and v:imagedata elements
• ✅ Both VML images successfully extracted with embedded data

Files Changed

• docling/backend/msword_backend.py - VML image extraction + code consolidation (~150 lines added, ~100 lines removed via consolidation)
• tests/data/docx/docx_vml_images.docx - Test document with VML images
• tests/data/groundtruth/docling_v2/docx_vml_images.docx.json - Ground truth

Resolves #3298

Checklist:

• Documentation has been updated, if necessary.
• Examples have been added, if necessary.
• Tests have been added, if necessary.

[mergify]

Merge Protections

Your pull request matches the following merge protections and will not be merged until they are valid.

🔴 Require two reviewer for test updates

Waiting for:

• #approved-reviews-by >= 2

This rule is failing.

When test data is updated, we require two reviewers

• #approved-reviews-by >= 2

🟢 Enforce conventional commit

Wonderful, this rule succeeded.

Make sure that we follow https://www.conventionalcommits.org/en/v1.0.0/

• title ~= ^(fix|feat|docs|style|refactor|perf|test|build|ci|chore|revert)(?:\(.+\))?(!)?:

[github-actions]

✅ DCO Check Passed

Thanks @ceberam, all your commits are properly signed off. 🎉

[codecov]

Codecov Report

❌ Patch coverage is 70.40816% with 29 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
docling/backend/msword_backend.py	70.40%	29 Missing ⚠️

📢 Thoughts on this report? Let us know!