Browser-Based Word-to-Markdown Converter Runs Pandoc via WebAssembly Without Uploads

A developer built a Word-to-Markdown converter that runs the Pandoc engine compiled to WebAssembly entirely inside a browser tab, so documents never leave the user's device. The author argues that converting through an intermediate HTML step loses document semantics like footnotes and citations, whereas Pandoc parses formats into a typed syntax tree that preserves them. Because the roughly 58 MB module is too heavy to load on page load, it is fetched lazily on user intent and run in a Web Worker to keep the interface responsive. Initially assuming download speed was the bottleneck, the developer found compilation was slower, so compiled modules are cached in IndexedDB, with fallbacks to the HTTP cache or streaming. Safari does not allow persisting compiled modules, so there the caching step degrades to re-downloading and recompiling.