DOCX Parsing

A .docx file is fundamentally a zip archive containing XML files. Rendering it in the browser involves unzipping the blob in-memory and semantically mapping Word's proprietary XML schema to standard HTML.

Semantic Conversion Strategy

Unlike PDFs which mandate pixel-perfect coordinate layouts, Word documents flow. Our approach prioritizes semantic fidelity over visual perfection. We use a tailored implementation of mammoth.js.

Mammoth ignores extraneous formatting (like exact font sizes or obscure margin adjustments) and maps Word styles directly to HTML equivalents:

Word Style	HTML Output
Heading 1	<h1>
Normal	<p>
List Paragraph	<ul><li>
Emphasis	<em>

Handling Embedded Media

Images embedded within the DOCX archive (found in the word/media/ directory of the zip) are extracted as raw binary strings and converted to data:image/*;base64 URIs. This allows them to render immediately in the DOM without requiring a temporary server host.