Summary

Fixes annotation projection methods to handle HTML fragments with multiple root elements and HTML named entities, which are common in sanitized HTML output from libraries like DOMPurify.

Key Changes

• HTML Fragment Parsing: Added ParseHtmlString() method that:

  • Replaces HTML named entities ( , –, etc.) with numeric XML equivalents for XML compatibility
  • Wraps multi-root HTML in a synthetic <docxodus-fragment-root> element to satisfy XElement.Parse() requirements
  • Falls back to wrapping only when single-root parsing fails

• HTML Serialization: Added SerializeHtmlString() method that:

  • Removes the synthetic wrapper when serializing back to string
  • Preserves original HTML structure for single-root documents

• Updated Methods: Modified three public methods to use the new parsing/serialization:

  • ProjectAnnotationsOntoHtml()
  • AddAnnotationToHtml()
  • RemoveAnnotationFromHtml()

• Comprehensive Tests: Added 6 new test cases covering:

  • Multiple root elements handling
  • HTML named entities (generic and   specifically)
  • Backward compatibility with single-root HTML
  • All three affected methods

Implementation Details

• HTML entity mapping includes 25 common entities (nbsp, ndash, mdash, quotes, currency symbols, etc.)
• Synthetic wrapper is only added when necessary (multi-root detection via exception handling)
• No changes to annotation logic itself—purely input/output handling improvements
• Maintains backward compatibility with well-formed single-root HTML documents

https://claude.ai/code/session_01B3Yy9hY9o3iUBf5KD8wXUP