docx

Use the docx npm package (docx-js) to build a Document object with paragraphs, tables, images, and styles, then write it with Packer.toBuffer. Always set the page size explicitly since docx-js defaults to A4, and validate the output file afterward.

Unpack the .docx into XML, then wrap inserted text in w:ins elements and deleted text in w:del elements containing w:delText, each with a unique ID, author, and date. Replace entire w:r run elements rather than injecting tags inside runs, and preserve the original w:rPr formatting.

Tables need dual widths: set columnWidths on the table and a matching width on every cell, both in DXA units that sum exactly to the table width. Never use WidthType.PERCENTAGE because it breaks in Google Docs, and use ShadingType.CLEAR instead of SOLID for cell shading.

Yes, run LibreOffice headless with the soffice.py helper using --convert-to docx to transform legacy .doc files before editing. The same tool converts .docx to PDF, and pdftoppm from Poppler renders pages to images.

Run the accept_changes.py script with the input and output file paths. It uses a LibreOffice macro to accept all tracked changes headlessly and saves a clean document, requiring LibreOffice to be installed.

Do not use it for PDFs, spreadsheets, Google Docs, or tasks unrelated to Word document generation. Also note that auto-repair only fixes durableId overflow and missing xml:space attributes; malformed XML or schema violations must be fixed manually.