py-turbo-api

TurboAPI is a small, fast ASGI web framework built from scratch (stdlib-only) with a focus on predictable performance.

Features

• Trie router with {param} path params
• APIRouter composition with prefixes/tags via include_router()
• Compiled route plans (signature inspection happens at route registration, not per request)
• Dependency injection with Depends() including nested and yield-based dependencies
• Parameter source markers: Query, Header, Cookie, Form, File
• Built-in validation with Model + field() (supports Annotated, Literal, Enum, datetime, UUID, Decimal)
• Optional Pydantic v2 compatibility (request/response models + OpenAPI schema integration when installed)
• Multipart/form-data parsing with UploadFile
• Multipart hardening: per-file/part/field limits + upload spooling to disk
• StreamingResponse, FileResponse, BackgroundTask, and static file mounting
• Response primitives: RedirectResponse, NegotiatedResponse, cache header helpers, and EventSourceResponse for SSE
• File hardening: ETag/If-None-Match + byte range (206/416) support
• WebSocket route support via @app.websocket(...)
• Sub-app mounting via app.mount("/prefix", sub_app) and host routing via app.mount_host("*.example.com", app)
• WebSocket auth helpers (websocket_token_auth, websocket_jwt_auth) and OpenAPI WS extension docs (x-turbo-websockets)
• Security primitives: api_key_auth(), bearer_auth(), jwt_auth(), OAuth2 password/auth-code/client-credentials helpers with OpenAPI security schemas
• Session + CSRF protection primitives (SessionMiddleware, CSRFMiddleware, CSRF helpers)
• Testing utilities: sync TestClient and async AsyncTestClient with WebSocket test sessions
• OpenAPI at /openapi.json
• Swagger UI at /docs and ReDoc at /redoc
• Lifespan startup/shutdown handlers
• Lifespan state resources with cleanup guarantees (app.add_state_resource(...))
• App-state dependency helpers (app.state_dependency(...), app_state_dependency(...), get_app_state(...))
• Custom exception handlers with @app.exception_handler(...)
• Reliability defaults: request timeout, max body size, max concurrency
• Runtime settings model (TurboSettings) + Turbo.from_env() for env-driven deploy config
• Graceful shutdown request draining (shutdown_drain_timeout)
• Background job primitives: InMemoryJobQueue with retries, delay/schedule, and idempotency keys
• Queue adapters: CeleryQueueAdapter, RQQueueAdapter, RedisQueueAdapter
• CI release gates: Python matrix tests/lint + package build and twine check, with trusted publishing workflow
• Benchmark suite + perf regression gates in CI (benchmarks/bench_runtime.py)

Stability

• Compatibility policy: API_COMPATIBILITY.md
• Current status: pre-1.0 (public API is stabilizing toward v1.0.0)

Install

From PyPI

pip install py-turbo-api

Optional: Pydantic v2 compatibility

pip install "py-turbo-api[pydantic]"

PyPI: https://pypi.org/project/py-turbo-api/

Run example

uvicorn app:app --reload

Open:

• http://127.0.0.1:8000/docs
• http://127.0.0.1:8000/openapi.json

Documentation Tracks

• Docs home (GitHub Pages entry): docs/index.md
• Complete API reference: docs/api-reference.md
• Tutorial: docs/tutorial.md
• Advanced: docs/advanced.md
• Deployment: docs/deployment.md
• Security recipes: docs/security-recipes.md
• Why TurboAPI + benchmark method: docs/why-turboapi.md

Benchmarks

python benchmarks/bench_runtime.py --baseline benchmarks/baseline.json --tolerance 1.20 --gate

License

Apache-2.0 (see LICENSE).