DocRender - Fast Local DOCX Preview for Chrome

Needed to open a .docx. No Word, no Google account, slow connection. Every option was either a 4GB install or uploading my file to someone's cloud.

So I built this. Drag a .docx into Chrome, read it, done. No uploads, no accounts, no installs. Everything stays on your device.

The problem (you've been there)

Opening one Word file shouldn't mean:

• installing a 4+ GB office suite you'll use once
• uploading private/corporate stuff to cloud services
• waiting forever on a slow connection for something that's already on your disk

DocRender kills that. Open, read, close.

How it works

.docx file > background parser > semantic HTML > styled viewer

DocRender doesn't try to be Word. It does semantic-first rendering: extract structure, restore visual hierarchy (headings, fonts, colors, spacing), optimize for reading. Result looks close to Google Docs but fully local, fully private.

What renders well:

• headings, paragraphs, alignment, spacing
• text colors and font families (Google Fonts fallback)
• ordered/unordered lists, tables, images
• multi-column layouts, dark mode

What's intentionally skipped (viewer not editor):

• floating images, complex text wrapping
• page headers/footers, exact print layout
• fonts not in Google Fonts (falls back to system)

Deliberate trade-offs. Speed and privacy over pixel-perfect reproduction.

Privacy

• all processing happens locally in the browser
• no servers, no uploads, no tracking
• files never leave your device
• only external request: Google Fonts (optional)

Works for confidential docs, corporate environments, air-gapped setups.

Usage

• drag and drop a .docx, or use the file picker
• recent docs are cached for quick access
• .doc intentionally not supported (legacy binary format, different beast)

Dev section

Quick start

npm install        # applies a mammoth patch via patch-package
npm run build      #   (preserves text colors + font metadata)

chrome://extensions > Developer mode > Load unpacked > dist/

Architecture

Download/Upload > Background stores ArrayBuffer in IndexedDB
               > Viewer requests by documentId
               > DOCX > HTML (mammoth + custom post-processing) > DocumentViewer renders

Core pipeline:

const doc = await getDocument(documentId);
const parsed = await convertDocxToHtml(doc.data);
render(parsed.html);

Mammoth does DOCX to HTML but out of the box it strips colors and font metadata. Patched it (patch-package) to preserve those. Custom post-processing layer then restores visual hierarchy that raw conversion loses.

Stack

• TypeScript, Chrome Extension Manifest V3
• Mammoth.js (patched) for DOCX parsing
• custom post-render pipeline for semantic styling
• IndexedDB for document caching
• no backend, no runtime dependencies

What this project shows

• parsing complex binary formats client-side under browser constraints
• memory management in extension context (large files, no Node.js luxuries)
• UX in constrained environments (popup + viewer inside an extension)
• pragmatic trade-offs: shipping something useful vs overengineering a Word clone

Status

Maintained. Not actively adding features, but bugs get fixed and PRs are welcome.

License

MIT