Paper Reader

Paste the URL of a hard-to-read PDF (e.g. a dense, two-column academic paper) and get back a clean, black-on-white article in a normal reading font, with the paper's figures extracted as click-to-zoom images.

Live: https://theodo-group.github.io/paper-reader/

bun install
bun dev        # http://localhost:3001  (hot reload)
# or
bun start

The default URL is the Deci/Olafsen/Ryan SDT paper — just press Make readable.

How it works

• server.ts — a Bun fullstack server for local use. Serves the React app and exposes /api/proxy?url=…, which downloads the PDF server-side so the browser never hits CORS.
• src/fetchPdf.ts — picks how the PDF bytes are fetched: the local /api/proxy when running on the Bun server, or public CORS proxies when running as a static site (GitHub Pages).
• src/pdf/ — the conversion engine, running in the browser with pdf.js:
  • extracts the real text layer, merges fragments into lines, and detects 1- vs 2-column layouts to recover reading order;
  • groups lines into paragraphs/headings (de-hyphenating line breaks);
  • walks the page operator list while tracking the transform matrix to find and rasterise each embedded figure to a PNG;
  • ignores full-page scan backdrops (a page-sized image sitting behind an invisible text layer) so the real text reads through instead of coming back as one giant image — while recovering the real figures/tables baked into that scan by their captions ("FIG. 1", "TABLE 2") and cropping them as images;
  • suppresses garbled OCR debris that scanned diagrams paint over their artwork (and crops full-page scanned tables as images instead of leaking their cells as text), stitches sentences that continue across a page break back into one paragraph, repairs letter-spaced OCR words ("o f j o b" → "of job"), and avoids mistaking a large-font body paragraph for a heading on size-jittery scans;
  • falls back to OCR (tesseract.js, in-browser) for pages that have no text layer at all.
• src/App.tsx / src/Lightbox.tsx — re-typesets the result and provides a full-screen zoomable image viewer (click a figure, click again for actual size, Esc to close).

Deployment

Pushing to main triggers .github/workflows/deploy.yml, which runs bun run build (bundling index.html into dist/) and publishes it to GitHub Pages.

Note: GitHub Pages serves static files only, so the server-side /api/proxy isn't available there. The static build routes PDF downloads through public CORS proxies instead. For a fully self-hosted, private path, run bun start (which uses the built-in proxy).

Notes / limits

• Works best on PDFs with a real text layer (most digital papers), including scanned papers that carry an invisible OCR text layer behind a full-page image (the backdrop is dropped and the text reads through).
• Pages with no text layer at all are OCR'd in the browser with tesseract.js. This is slow (a few seconds per page) and downloads the language model from a CDN the first time; accuracy depends on scan quality.
• Column detection is heuristic; unusual layouts may interleave oddly.
• The pdf.js worker (and the tesseract.js core/model) are loaded from a CDN, so the page needs internet access the first time.