Asterion — Desktop Novel Reader

A local-first Linux desktop reader for web novels, built with Tauri 2 + React. Scrapes novels from novelfire.net (Rust scraper), stores them in Supabase, and delivers an immersive reading experience ported from the Asterion iOS app.

No web3, no accounts to manage — per-device identity uses Supabase anonymous auth.

Architecture (hybrid: GCP + Supabase)

Desktop app (Tauri + React)                  scraper-service (GCP VM, Rust)
  ├─ metadata/search ─► Supabase (anon)        ├─ scrape novelfire (reqwest+scraper)
  ├─ chapter text ────► GCS (public bucket)     ├─ chapter TEXT ─► Google Cloud Storage
  ├─ progress/library/prefs ─► Supabase (RLS)   └─ metadata+index ─► Supabase (service key)
  └─ "request novel" ─► Supabase queue ─────────────────▲ (VM drains the queue)

• scraper-service/ — standalone Rust crate that runs on a GCP Compute Engine VM. Bulk chapter text (the TB) goes to Google Cloud Storage; lightweight metadata + a chapter index go to Supabase. The service-role key lives only on the VM. See scraper-service/README.md.
• Desktop app is a pure reader — it reads metadata from Supabase and chapter text from GCS, and queues scrape_requests for novels not yet scraped (no client-side scraping).
• Reader (src/pages/Reader.tsx) — port of asterion-ios ReaderView.swift: immersive auto-hiding controls, serif/sans + adjustable size (14–28) / spacing / width, dark·sepia·warm·light themes, per-paragraph progress + resume, keyboard nav (←/→ chapters, +/- font, Esc back), offline chapter cache, download-to-file.

Prerequisites (Ubuntu/Debian)

sudo apt install -y libwebkit2gtk-4.1-dev build-essential curl wget file \
  libxdo-dev libssl-dev libayatana-appindicator3-dev librsvg2-dev

Node 20+, pnpm, and the Rust toolchain are also required.

Setup

pnpm install

# 1. Supabase backend (see supabase/README.md)
supabase link --project-ref <ref>
supabase db push
SUPABASE_URL=… SUPABASE_SERVICE_ROLE_KEY=… node supabase/seed/seed-catalog.mjs

# 2. GCP scraper + storage (see scraper-service/README.md)
#    create a GCS bucket + VM, run the catalog-runner to populate content

# 3. Client env
cp .env.example .env   # VITE_SUPABASE_URL + VITE_SUPABASE_ANON_KEY + VITE_GCS_BUCKET

Develop / build

pnpm tauri dev      # run the desktop app
pnpm tauri build    # produce .deb / .rpm / AppImage in src-tauri/target/release/bundle
pnpm build          # frontend-only type-check + build (no system deps needed)

Status

• Phase 0 — Tauri + Vite/React scaffold, theme tokens
• Phase 1 — Supabase schema, RLS, ingest-novel function, catalog seed
• Phase 2 — Rust scraper (extraction validated against live novelfire)
• Phase 3 — Immersive Reader
• Phase 4 — Discover / Library / NovelDetail / Ranking / Profile
• Phase 5 — Packaging polish (icons, desktop entry, AppImage QA)