docx

Unpack the .docx file, then use the Python Document library to wrap edits in <w:ins> and <w:del> tags with author, date, and RSID attributes. The library auto-injects these attributes and validates the result before packing back to .docx.

Use the docx npm package to build a Document from Paragraph, TextRun, Table, and ImageRun components, then export with Packer.toBuffer(). Always use proper numbering configs for lists and separate Paragraph elements instead of newline characters.

Convert the document 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.

Common causes include standalone PageBreak elements outside a Paragraph, missing ImageRun type parameters, unicode bullets instead of numbering configs, and invalid tracked-change nesting. Follow the schema ordering rules and run the built-in validation on save.

Yes. Use revert_insertion() to wrap an insertion in a deletion, and revert_deletion() to restore deleted content as a new insertion. Never modify text inside another author's <w:ins> or <w:del> tags directly.