Starlite is a powerful, flexible and highly performant ASGI API framework built on top of Starlette and pydantic.
Check out the Starlite documentation ๐
pip install starlite
from starlite import Starlite, get
@get("/")
def hello_world() -> dict[str, str]:
"""Keeping the tradition alive with hello world."""
return {"hello": "world"}
app = Starlite(route_handlers=[hello_world])
- Both functional and OOP python support
- Class based controllers
- Extended testing support
- Builtin Validation and Parsing using Pydantic
- Dataclass Support
- Dependency Injection
- Layered Middleware
- Layered Parameter declaration
- Route Guards based Authorization
- Life Cycle Hooks
- Plugin System
- SQLAlchemy Support (via plugin)
- Tortoise-ORM Support (via plugin)
- Trio support (built-in, via AnyIO)
- Automatic OpenAPI 3.1 schema generation
- Support for Redoc
- Support for Swagger-UI
- Support for Stoplight Elements
- Ultra-fast json serialization and deserialization using orjson
- starlite-pg-redis-docker: In addition to Starlite, this demonstrates a pattern of application modularity, SQLAlchemy 2.0 ORM, Redis cache connectivity, and more. Like all Starlite projects, this application is open to contributions, big and small.
- starlite-hello-world: A bare-minimum application setup. Great for testing and POC work.
Although Starlite uses the Starlette ASGI toolkit, it does not simply extend Starlette, as FastAPI does. Starlite uses selective pieces of Starlette while implementing its own routing and parsing logic, the primary reason for this is to enforce a set of best practices and discourage misuse. This is done to promote simplicity and scalability - Starlite is simple to use, easy to learn, and unlike both Starlette and FastAPI - it keeps complexity low when scaling.
Additionally, Starlite is very fast in comparison to other ASGI frameworks. In fact, the only framework that is faster in our benchmarks is BlackSheep, which is almost completely written in cython and does not work with pydantic out of the box as such:
Legend:
- a-: async, s-: sync
- np: no params, pp: path param, qp: query param, mp: mixed params
You can see and run the benchmarks here.
While supporting function based route handlers, Starlite also supports and promotes python OOP using class based controllers:
from typing import List, Optional
from pydantic import UUID4
from starlite import Controller, Partial, get, post, put, patch, delete
from datetime import datetime
from my_app.models import User
class UserController(Controller):
path = "/users"
@post()
async def create_user(self, data: User) -> User:
...
@get()
async def list_users(self) -> List[User]:
...
@get(path="/{date:int}")
async def list_new_users(self, date: datetime) -> List[User]:
...
@patch(path="/{user_id:uuid}")
async def partial_update_user(self, user_id: UUID4, data: Partial[User]) -> User:
...
@put(path="/{user_id:uuid}")
async def update_user(self, user_id: UUID4, data: User) -> User:
...
@get(path="/{user_name:str}")
async def get_user_by_name(self, user_name: str) -> Optional[User]:
...
@get(path="/{user_id:uuid}")
async def get_user(self, user_id: UUID4) -> User:
...
@delete(path="/{user_id:uuid}")
async def delete_user(self, user_id: UUID4) -> None:
...
While running Starlite, you can view the generated OpenAPI documentation using a ReDoc site, a Swagger-UI as well as a Stoplight Elements site.
One key difference between Starlite and Starlette/FastAPI is in parsing of form data and query parameters- Starlite supports mixed form data and has faster and better query parameter parsing.
Starlite is rigorously typed, and it enforces typing. For example, if you forget to type a return value for a route handler, an exception will be raised. The reason for this is that Starlite uses typing data to generate OpenAPI specs, as well as to validate and parse data. Thus typing is absolutely essential to the framework.
Furthermore, Starlite allows extending its support using plugins.
Starlite has a plugin system that allows the user to extend serialization/deserialization, OpenAPI generation and other features. It ships with a builtin plugin for SQL Alchemy, which allows the user to use SQLAlchemy declarative classes "natively", i.e. as type parameters that will be serialized/deserialized and to return them as values from route handlers.
Starlite also supports the programmatic creation of DTOs with a DTOFactory
class, which also supports the use of
plugins.
Starlite has custom logic to generate OpenAPI 3.1.0 schema, the latest version. The schema generated by Starlite is
significantly more complete and more correct than those generated by FastAPI, and they include optional generation of
examples using the pydantic-factories
library.
Starlite has a simple but powerful DI system inspired by pytest. You can define named dependencies - sync or async - at different levels of the application, and then selective use or overwrite them.
Starlite supports the Starlette Middleware system while simplifying it and offering builtin configuration of CORS and some other middlewares.
Starlite has an authorization mechanism called guards
, which allows the user to define guard functions at different
level of the application (app, router, controller etc.) and validate the request before hitting the route handler
function.
Starlite supports request life cycle hooks, similarly to Flask - i.e. before_request
and after_request
Starlite is open to contributions big and small. You can always join our discord server or join our Matrix space to discuss contributions and project maintenance. For guidelines on how to contribute, please see the contribution guide.
Thanks goes to these wonderful people (emoji key):
This project follows the all-contributors specification. Contributions of any kind welcome!