docx

Use the docx npm library to build documents with Document, Paragraph, and TextRun components, then export with Packer.toBuffer(). Define custom styles, tables, lists, and images, and always use separate Paragraph elements instead of newline characters.

Unpack the document with the pack/unpack scripts, then use the Document library's replace_node, suggest_deletion, revert_insertion, and revert_deletion methods. The library auto-injects RSIDs, author, and date attributes, and validates the result on save.

Convert the .docx to markdown using pandoc with the --track-changes=all flag to preserve insertions and deletions. For comments, embedded media, or complex formatting, unpack the file and read the raw XML in word/document.xml directly.

Yes, the Python Document class provides add_comment to anchor comments to elements and reply_to_comment for threaded replies. It automatically manages comments.xml, commentsExtended.xml, commentsIds.xml, and people.xml infrastructure.

Common causes include standalone PageBreak elements outside a Paragraph, unicode bullets instead of numbering config, missing ImageRun type parameters, or invalid tracked change nesting. Follow the documented critical rules and run validation on save.