FolioCove: browser PDF workflows with PDF.js previews and reproducible sample outputs

[mehbul]

I maintain FolioCove, an MIT-licensed browser PDF workspace. I’m sharing its PDF.js integration and looking for feedback on a small synthetic sample workflow.

PDF.js provides page rendering, organizer previews and text extraction; pdf-lib handles PDF creation and page copying. The interface is our own workspace rather than the stock PDF.js viewer.

A few implementation choices:

• Local files are passed as Uint8Array data to getDocument(). The build bundles the worker and decoder assets and configures their paths for the GitHub Pages project subpath.
• Structural page operations and raster conversion are separate. Raster compression loses selectable text, forms and accessibility information, so its warning appears before processing.
• Synthetic merge and extraction inputs have visible page markers and reference outputs. A parseable download alone does not establish correct page order or rendering.

Try one synthetic task · Source · Processing details

If you try a sample, feedback on blank previews, incorrect page order or unclear output-check instructions would help. Please include your browser/device and reproduce the issue using synthetic files. Application reports belong in FolioCove’s issue tracker or our support channel, so they do not create support work for PDF.js maintainers.

This remains a public beta. Keep originals and inspect outputs. Physical-device camera/download/Safari validation is incomplete; Office conversion has fidelity limits, redaction is experimental, browser AI is conditional, and verified sanitization/PDF/A are unavailable. Site assets and optional models use the network; on-device document processing does not imply complete offline operation.

Disclosure: I am FolioCove’s maintainer. Development and this post used OpenAI Codex assistance.