Converts .docx to clean semantic HTML entirely in the Worker with the in-house ZIP reader and XML parser (no mammoth.js): paragraph styles become headings (Heading 1–6, outline levels, Title/Subtitle), block quotes and code blocks; numbering.xml drives nested bulleted, numbered, lettered and roman lists; tables keep header rows, column spans and vertical merges; runs keep bold, italic, underline, strike, super/subscript, highlights and inline code; hyperlinks, bookmarks, footnotes/endnotes and images (embedded as data URIs or omitted) are converted; tracked insertions are accepted and deletions dropped; content controls are unwrapped. The report lists document properties, counts and anything that was not converted. Everything runs locally.