AniVault Scraper API — Cloudflare Workers port

Ported from the original Express/Railway app to run on Cloudflare Workers (Hono + native fetch). Type-checks clean and bundles successfully with wrangler deploy --dry-run (~1MB / 238KB gzipped).

Deploy via GitHub + Cloudflare Workers Builds (no CLI needed)

1. Push this folder as a new GitHub repo.
2. In the Cloudflare dashboard: Workers & Pages → Create → Import a repository, pick this repo. Cloudflare will detect wrangler.toml and handle npm install + build/deploy automatically on every push.
3. Set environment variables/secrets under Settings → Variables on the deployed Worker (see list below). Don't put secrets in wrangler.toml.
4. Every future git push auto-deploys.

Environment variables to set in the dashboard

Name	Used by	Notes
BOT_SECRET	discord-relay	Secret — must match your PHP config + Vercel bot
VERCEL_BOT_URL	discord-relay /relay	e.g. https://anivault-bot.vercel.app
DISCORD_APP_ID	discord-relay	Discord application ID
SITE_URL	discord-relay	Defaults to https://www.anivault.co
FLARESOLVERR_URL	fetch.ts, discord-relay	Only needed if you still run FlareSolverr somewhere reachable from the public internet — Workers can't run it itself
RATE_LIMIT_WINDOW_MS / RATE_LIMIT_MAX	index.ts fallback limiter	Optional, see caveat below
CACHE_TTL_MAPPING / CACHE_TTL_EPISODES / CACHE_TTL_STREAM	cache.ts	Optional, seconds

Sources

Only AnimeHeaven and Anikoto are wired in. Senshi and Miruro were dropped (both had stopped working) along with everything only they needed: scrapers/senshi.ts, scrapers/miruro.ts, and resolvers/megacloud.ts (Senshi's separate embed resolver — not to be confused with the self-contained megacloud decryption logic inside anikoto.ts, which stays). crypto-js was removed from dependencies since it was only used by that deleted resolver.

What changed from the Express version

• Removed entirely: image-migrator.ts (used basic-ftp/raw TCP sockets and archiver — neither can run on Workers, no workaround exists).
• Removed: the two /api/debug/miruro* routes (marked "remove before production" in the original) and process.uptime() from /health (no persistent process on Workers).
• Dropped as dead code: anidao.ts, animepahe.ts, aniwaves.ts, reanime.ts — none of these were actually imported by routes.ts.
• Express → Hono, axios → native fetch via a small compatibility shim (src/utils/http.ts) so the scraper files barely changed.
• node-cache → in-memory Map (src/utils/cache.ts). Caveat: this only survives as long as the current Worker isolate does — Cloudflare reuses isolates for a while under steady traffic, but a cold isolate starts with an empty cache. For persistence across isolates/deploys, swap to Workers KV.
• express-rate-limit → best-effort in-memory limiter, same isolate caveat as above. The reliable replacement is Cloudflare's native Rate Limiting Rules, configured in the dashboard against this Worker's route — no code needed, and it actually works across the whole edge, not just one isolate. Recommended over the in-code fallback.
• node:zlib, node:buffer (used by miruro.ts/anikoto.ts) work natively under the nodejs_compat flag with the compatibility date set in wrangler.toml — no changes needed beyond the node: import prefix.
• express.static → Workers Static Assets ([assets] binding in wrangler.toml, serves public/index.html).

Known unresolved gap

discord-relay.ts's original InfinityFree call used new https.Agent({ rejectUnauthorized: false }) to skip TLS certificate verification. Workers' fetch has no way to disable TLS verification — this isn't a missing polyfill, it's not exposed at all. If InfinityFree's cert is genuinely invalid, this specific call will fail on Workers with no in-Worker fix; it'd need a valid cert at the source, or that one call proxied through something that isn't Workers.

Local dev

npm install
npx wrangler dev