Architecture

Archivis is built as a Rust + Svelte 5 application shipped as a single binary:

• Backend: Rust with Axum web framework and Tokio async runtime
• Frontend: Svelte 5 with SvelteKit, Tailwind CSS, and shadcn-svelte components
• Database: SQLite with WAL mode for concurrent access — no external database needed
• Background jobs: Tokio-based async task system

Design Principles

1. Separated concerns — API server, background workers, and frontend have well-defined interfaces
2. Database lives separately from book files — no .db files stored alongside your ebooks
3. All heavy work is async — imports, metadata, OCR, and conversions run as background tasks
4. Pluggable architecture — metadata sources, storage backends, and auth adapters are extensible
5. API-first — the web UI is just another API consumer; everything it does, you can do via the API

Workspace Layout

The backend is a Cargo workspace with 9 crates:

Crate	Purpose
archivis-server	Binary entrypoint, config, startup
archivis-api	Axum HTTP handlers and REST endpoints
archivis-core	Domain models and shared types
archivis-db	SQLite persistence via sqlx
archivis-formats	Ebook format detection and parsing
archivis-metadata	Metadata source plugins
archivis-tasks	Background job system
archivis-storage	File storage abstraction
archivis-auth	Authentication and authorization