Skip to content

README.md

Latest commit

 

History

History
215 lines (158 loc) · 7.82 KB

README.md

File metadata and controls

215 lines (158 loc) · 7.82 KB
Raw

Example Litestar Service

A reference Python web service built on Litestar — a working template for production REST + async services with clean architecture, observability, JWT auth and RBAC.

The service implements a small "library" domain (books, users, OpenLibrary integration, async book upload via NATS) and is meant to be cloned and adapted under your project, not used as-is.

Features

  • Layered architecture (domains / adapters / presentors / application) with import boundaries enforced by import-linter
  • DI via dishka (App + Request scopes, managed lifecycle for engine / Redis / HTTP / NATS broker)
  • Two transports in one app: REST (Litestar) + async pub/sub (FastStream + NATS JetStream)
  • JWT auth: /auth/register, /auth/login, /auth/refresh, /auth/me — access + refresh token pair, TTLs configurable via env
  • RBAC: Permission enum, AuthUser entity, permission_guard / any_permission_guard for route-level protection
  • Password hashing with passlib + argon2
  • Health probes: /health/live (liveness) and /health/ready (parallel Postgres / Redis / NATS checks with timeouts) backed by a dedicated adapter
  • Observability: Prometheus on /metrics, structlog (console in dev, JSON in prod), Sentry with traces and profiles sampling
  • Request correlation: middleware sets X-Request-ID, binds it to structlog.contextvars and echoes it back in the response
  • Database: SQLAlchemy + asyncpg + alembic, UoW pattern around async sessions
  • Hybrid serialization: Pydantic at the REST boundary, msgspec inside adapters (no hard coupling to Pydantic)
  • Testing diamond: unit tests for domain services with FakeStorage, integration tests for DB / REST / NATS against real Postgres + Redis + NATS in docker-compose, ~94% coverage
  • Tooling: ruff, mypy, import-linter, pytest + coverage, pre-commit, uv

Quick start

make develop                                              # create .venv via uv, install deps, set up pre-commit
make local                                                # start Postgres + Redis + NATS via docker-compose (foreground)
# in another terminal:
set -a && source .env.dev && set +a                       # load default env
.venv/bin/python -m library.adapters.database upgrade head  # apply migrations
.venv/bin/python -m library                               # serve on http://127.0.0.1:8000

OpenAPI UIs:

Working with the repo

Install dependencies

Creates a fresh .venv via uv sync and installs pre-commit hooks:

make develop

Run dev containers

Starts Postgres, Redis and NATS from docker-compose.dev.yaml:

make local

Stop and clean up volumes:

make local-down

Run tests

DB containers must be up (make local running in another terminal):

.venv/bin/pytest -vx ./tests

For CI-style with coverage report:

make test-ci

The minimum coverage is set to 80% in pyproject.toml; current coverage is above 90%.

Run linters

make lint-ci    # ruff + mypy + import-linter

Or individually:

.venv/bin/ruff check ./library
.venv/bin/mypy --config-file ./pyproject.toml ./library
.venv/bin/lint-imports

import-linter enforces these architectural contracts:

  • library must not import tests
  • library.domains must not import library.adapters
  • library.domains must not import library.presentors
  • library.adapters must not import library.presentors

Database migrations

Apply all migrations to head:

.venv/bin/python -m library.adapters.database upgrade head

Generate a new migration after changing tables in library/adapters/database/tables.py:

.venv/bin/python -m library.adapters.database revision --autogenerate -m "your message"

Both commands require APP_DATABASE_* env vars to point at a running Postgres.

CI workflow

make develop-ci  # install deps without venv (CI runs as root in container)
make lint-ci     # ruff + mypy + import-linter
make test-ci     # pytest with junit + coverage report

Configuration

All configuration goes through environment variables, parsed by dataclass-based config objects in library/application/config.py and per-adapter config.py files. A working set of defaults for local development is in .env.dev.

Key variables:

Variable Default Notes
APP_DATABASE_HOST / _PORT / _USER / _PASSWORD / _NAME from .env.dev Postgres connection
APP_REDIS_HOST / APP_REDIS_PORT 127.0.0.1 / 6379
APP_NATS_HOST / APP_NATS_PORT 127.0.0.1 / 4222
APP_HTTP_HOST / APP_HTTP_PORT 0.0.0.0 / 8080
APP_SECRET secret JWT signing secret. Must be set in prod.
APP_AUTH_ACCESS_TOKEN_TTL_SECONDS 900 15 minutes
APP_AUTH_REFRESH_TOKEN_TTL_SECONDS 604800 7 days
APP_LOG_LEVEL DEBUG
APP_LOG_JSON true JSON logs in prod, set to false for console output
APP_SENTRY_USE False
APP_SENTRY_DSN empty
APP_SENTRY_TRACES_SAMPLE_RATE 0.0
APP_SENTRY_PROFILES_SAMPLE_RATE 0.0

Routes

OpenAPI is the source of truth — open http://127.0.0.1:8000/docs/swagger after starting the service. Reference list:

Auth

POST    /auth/register             Register a new user (username + email + password)
POST    /auth/login                Log in by username + password, returns access + refresh
POST    /auth/refresh              Refresh access + refresh token pair
GET     /auth/me                   Get current user from Bearer access token
GET     /auth/admin-zone           Demo of permission_guard ([Permission.MANAGE_BOOKS])

Books

GET     /api/v1/books/             Fetch books (paginated)
POST    /api/v1/books/             Create a book
GET     /api/v1/books/{book_id}/   Fetch book by ID
PATCH   /api/v1/books/{book_id}/   Update book by ID
DELETE  /api/v1/books/{book_id}/   Delete book by ID

Users

GET     /api/v1/users/             Fetch users (paginated)
POST    /api/v1/users/             Create user (admin CRUD; no password is set)
GET     /api/v1/users/{user_id}/   Fetch user by ID
PATCH   /api/v1/users/{user_id}/   Update user by ID
DELETE  /api/v1/users/{user_id}/   Delete user by ID

OpenLibrary

GET     /api/v1/open-library/search?query=...   Search books on OpenLibrary

Service / observability

GET     /metrics                   Prometheus metrics
GET     /health/live               Liveness probe (always 200 while process is alive)
GET     /health/ready              Readiness probe (Postgres + Redis + NATS, 200 or 503)
GET     /docs                      OpenAPI JSON
GET     /docs/swagger              Swagger UI
GET     /docs/redoc                Redoc UI

Project layout

library/
├── application/        Logging, exceptions, base config, Sentry setup
├── adapters/           Outbound: database, redis, nats, open_library, healthcheck
├── domains/            Entities, interfaces, services, use cases (no infra imports)
└── presentors/
    ├── rest/           Litestar app, controllers, middleware, route handlers
    └── faststream/     NATS subscribers
tests/
├── domains/services/   Unit tests of domain services with FakeStorage
├── adapters/           Integration tests against real Postgres / Redis / HTTP
└── presentors/         REST + FastStream integration tests

License

MIT.