MangaRead

A self-hosted manga library for you and the people you share it with. It fetches chapters from supported sites, merges the same title from several sources into one series, and serves a fast reader with accounts, synced progress, and bookmarks.

One Docker container, one SQLite file, sized for 1–20 readers. Runs on a Raspberry Pi 4.

Content notice: chapters come from third-party sites. Run MangaRead for yourself and people you know. Hosting it as a public site redistributes other people's work and invites takedowns.

Install

curl -O https://raw.githubusercontent.com/DevYatsu/mangareader/master/compose.yaml
docker compose up -d
docker compose logs mangaread | grep "setup token"

1. Open http://localhost:3000/setup.
2. Paste the setup token from the log and create the admin account. Until you do, every page redirects to /setup, so nobody else can claim the instance.
3. Set up email at /admin/email (setup takes you there). Readers sign up with an emailed link, so sign-up stays closed until SMTP works. "Send test email" checks the settings.
4. Optional: set the title, colors, and homepage at /admin/branding.

The first sync starts on boot and runs every 6 hours. A large source takes a while on first run.

Before exposing it beyond your LAN, put it behind HTTPS and set MANGAREAD_BASE_URL and MANGAREAD_COOKIE_SECURE=1 in compose.yaml.

Screenshots

Home	Series List

Series Details	Chapter Reader

Accounts and roles

Level	How you get it	Can
Visitor	Not logged in	Read everything. Progress stays in the browser.
User	Signs up by email link, or Google/GitHub/Discord	Progress, bookmarks, and notes synced across devices.
Moderator	Promoted by an admin	Will moderate chapter comments once they ship. No access to site settings or the catalog.
Admin	/setup, or promoted by an admin	Everything under /admin: branding, email, users (roles, bans, reset links).

• Sign-up never reveals whether an address already has an account.
• Sign-up links last 24 hours and reset links 1 hour. Both are single-use and stored only as hashes.
• Each address gets at most 3 emails per hour, and each IP at most 10 requests per hour.
• The last admin can never be demoted or banned. Every role change and ban is logged.
• Without SMTP, an admin can allow sign-up without verification (/admin/email, off by default). Anyone can then create accounts with any address.

Configuration

Branding and email are set in /admin and take effect without a restart. Environment variables are boot defaults; see .env.example.

Variable	Default	Purpose
MANGAREAD_DB	mangaread.db	SQLite path (/app/data/mangaread.db in Docker)
MANGAREAD_CACHE_DIR	cache/ beside the database	Reader image cache (/app/data/cache in Docker); size cap set in Admin → Sources
MANGAREAD_BASE_URL	http://localhost:3000	Public URL for emailed links and OAuth callbacks
MANGAREAD_COOKIE_SECURE	unset	1 behind HTTPS
MANGAREAD_TRUST_PROXY	unset	1 only behind a proxy that sets X-Forwarded-For
SMTP_URL, SMTP_FROM	unset	Email defaults; /admin/email overrides
MANGAREAD_OAUTH_{GOOGLE,GITHUB,DISCORD}_{ID,SECRET}	unset	Enables that login button
SITE_TITLE, SITE_COLOR, SITE_BG	MangaRead, #f59e0b, #131210	Branding defaults; /admin/branding overrides
CHROMIUM_FLAGS	--no-sandbox --disable-dev-shm-usage --disable-gpu --headless	Chromium args for scraping
RUST_LOG	info	Log level

Resources: each sync runs up to 3 headless Chromium sessions, so give the container about 2 GB of RAM. Chromium's sandbox needs SYS_ADMIN; a Chrome seccomp profile is the stricter alternative.

Backups: everything lives in the mangaread_data volume. Stop the container, then copy mangaread.db (with its -wal and -shm files if present).

Development

Prerequisites: Rust 1.88+ and chromedriver on PATH.

cargo run -p website          # http://localhost:3000; prints a setup token on a fresh DB
cargo test --workspace
docker compose -f compose.build.yaml up --build   # build the image from source

Three crates, one workspace:

Crate	What it does
mangaread-core	Domain types, SQLite repository (db.rs), series merging, title normalization
scraper	SiteAdapter trait, per-site adapters, and a debugging CLI
website	Axum server: routes in routes/, Maud templates in templates/, Bulma + vanilla JS in static/

Route tests drive the real router in-process over an in-memory database (website/src/test_support.rs). Outbound mail goes through the Mailer trait, which tests replace with an in-memory recorder.

Design specs and implementation plans live in docs/superpowers/. See docs/architecture/README.md for the data model, scraping pipeline, and sync loop.

Sources

Adapter	Site
DemonicscansAdapter	demonicscans.org
ManhuaplusAdapter	manhuaplus.top

To add one, implement SiteAdapter in scraper/src/adapters/ and register it in website/src/scraper_helper.rs and scraper/src/main.rs.

Scraper CLI

The server syncs on its own; these commands are for debugging.

cargo run -p manga-scraper scan                 # discover manga on every source
cargo run -p manga-scraper sync-manga <id>      # fetch one manga's chapter list
cargo run -p manga-scraper sync-chapter <id>    # fetch one chapter's page URLs

License

Training project — use freely.