docx

Use the docx npm library to build documents with paragraphs, tables, images, and headings. Set page dimensions in DXA units (1440 per inch), define table column widths on both the table and each cell, and use a numbering config for bullet lists instead of literal bullet characters.

Unzip the .docx, edit word/document.xml directly, then rezip the directory. Run merge_runs.py first to coalesce fragmented text runs, since Word splits visible text across many w:r elements that break naive find-and-replace.

Word fragments text across multiple w:r run elements due to revision IDs and spell-check markers, so a visible phrase may not exist as a contiguous string in the XML. Merging adjacent identically-formatted runs before editing resolves this without changing rendering.

Wrap inserted runs in w:ins elements and deleted runs in w:del elements with w:id, w:author, and w:date attributes, using w:delText instead of w:t inside deletions. Validate with the --author flag to confirm every edit is properly tracked.

Yes, convert legacy .doc files using LibreOffice headless mode with the command soffice --headless --convert-to docx file.doc. The resulting .docx can then be unpacked, edited, and validated like any other Word document.

Both pandoc --track-changes=accept and LibreOffice-based acceptance can leave stray empty paragraphs when a fully deleted paragraph should merge into the next one. Check paragraph deletions directly in the XML rather than trusting either rendered view.