ZEanth Laboratory — Paper Archive

A minimal, self-hosted paper archive for ZEanth Laboratory's computer science research, built as a static site for GitHub Pages — no server, no database, no build tool beyond a small Node script that GitHub Actions runs for you.

Live site: https://zeanth.github.io

What this is

• A listing page (/) with search, category/status/year filters, sorting, and pagination.
• A dedicated, permanent abstract page per paper (/abs/2601.00001) — each with its own identifier, status, categories, submission history, BibTeX citation, and links to the PDF/DOI/code/dataset.
• Papers are identified by a structured ID, YYMM.NNNNN (submission year + month, plus a serial number), instead of a raw PDF filename.
• A paper status field (preprint, under-review, accepted, published, withdrawn) shown as a badge everywhere the paper appears, so visitors always know where a paper stands.
• sitemap.xml and rss.xml are generated automatically so the archive is crawlable and can be followed with a feed reader.

How it works

Everything is driven by a single source of truth: papers.json. A tiny, dependency-free Node script (scripts/build.mjs) reads that file and generates the static abstract pages, the PDF availability manifest, the sitemap and the RSS feed. A GitHub Actions workflow (.github/workflows/deploy.yml) runs that script on every push to main and deploys the result to GitHub Pages — so you never have to remember to run a build command.

├── index.html                 → the listing page (fetches papers.json)
├── papers.json                → single source of truth for all papers
├── pdf/                       → contains paper folders with versioned PDFs and LICENSE files
├── abs/                       → generated permalink abstract pages (one folder per id)
├── templates/
│   ├── abs.template.html      → HTML template used to generate abs/ pages
│   └── pdf.template.html      → HTML template used to generate pdf/ viewer pages
├── assets/
│   ├── style.css              → shared styling for every page
│   ├── common.js               → shared helpers (badges, BibTeX, dates)
│   └── listing.js              → listing page logic (search/filter/pagination)
├── scripts/
│   ├── build.mjs               → generates abs/ and pdf/ pages, manifest, sitemap, rss, with validation
│   ├── new-paper.mjs           → interactive CLI to add new papers and copy PDF/LICENSE files
│   ├── new-version.mjs         → interactive CLI to publish new versions and bump history
│   └── manage-paper.mjs        → interactive CLI to list, edit, status-update, delete, and validate papers
└── .github/workflows/deploy.yml→ CI: build + deploy to GitHub Pages

Adding and Managing Papers

All operations are automated using helper scripts in the scripts/ directory:

1. Adding a paper: Run the interactive helper to scaffold folders, configure license settings, place the PDF/LICENSE, update papers.json, and rebuild the site:

   node scripts/new-paper.mjs

2. Publishing a new version: To bump paper version (e.g. v1 -> v2), update abstract details, place the new PDF, and record history:

   node scripts/new-version.mjs

3. Managing existing papers: To list, edit metadata, update status quickly, delete papers, or validate repository integrity:

   node scripts/manage-paper.mjs

Pushing to main is enough — GitHub Actions builds and deploys the rest.

Deploying this repository yourself

1. Push this repository to GitHub (any name is fine).
2. In Settings → Pages, set Source to GitHub Actions.
3. Push to main — the included workflow builds the site and publishes it automatically. No gh-pages branch, no manual steps.
4. Update the "archive" block at the top of papers.json (name, url, github, logo, ...) to match your own lab/repo.

Running the build locally (optional)

You don't need Node installed to just browse or edit papers.json — GitHub Actions builds the site for you. But if you want to preview the generated abstract pages locally before pushing:

node scripts/build.mjs
python3 -m http.server 8000   # or any static file server
# open http://localhost:8000

(A local static server is required because the listing page fetches papers.json via fetch(), which most browsers block on file:// URLs.)

Design

The visual language is intentionally quiet: a single accent-free black-on-white palette, Inter for text and JetBrains Mono for identifiers/metadata, hairline dividers instead of cards or shadows. Status badges use one small dot of colour each so the archive stays scannable without turning into a dashboard. All of it lives in assets/style.css and is shared by the listing page and every generated abstract page, so the two never drift apart visually.

License

The site's source code is released under the AGPL-3.0 License. This does not apply to the papers, PDFs, or research content hosted here — copyright of each paper remains with its authors and/or publisher.