🚀 Performance, type safety, and elegance. A next-generation async Python framework for APIs, microservices, and web applications. 🚀

Documentation: https://ravyn.dev 📚

Source Code: https://github.com/dymmond/ravyn

The official supported version is always the latest released.

If you came looking for Esmerald, you are in the right place. Esmerald was rebranded to Ravyn.

Ravyn is a modern, powerful, flexible, high performant, web framework designed to build not only APIs but also full scalable applications from the smallest to enterprise level.

Ravyn is designed to build with Python 3.10+ and based on standard python type hints. Initially built on the top of Starlette and later on moved to Lilya and Pydantic/msgspec.

Check out the Ravyn documentation 📚

Why is this happening? Is Esmerald going away? No, absolutely not. Esmerald remains and will remain as is and will keep growing as it has its own use cases.

The reason for the rebranding its because the ecosystem has grown a lot and Esmerald was the first tool being created. Since then it was released as version 3+.

This happened because of internal dependencies, and we already removed all of them but goes a bit off compared to the rest of the ecosystem.

This is the reason for Ravyn to come into existence and to keep everything aligned with the future projects coming out.

There are great frameworks out there like FastAPI, Flama, Flask, Django... All of them solving majority of the current day-to-day problems of 99% of the applications but leaving the 1% that is usually around structure and design/business without to much to do.

Ravyn got the inspiration from those great frameworks out there and was built with all the known amazing features but with business in mind as well. Starlite, for example, at the very beginning, gave the inspiration for the transformers and for the Signature models, something very useful that helped Ravyn integerating with pydantic. FastAPI gave the inspiration for API designing, Django for the permissions, Flask for the simplicity, NestJS for the controllers and the list goes on.

For a job to be done properly, usually it is never done alone and there is always a driver and inspiration to it.

• Python 3.10+

Ravyn wouldn't be possible without two pillars:

• Lilya
• Pydantic

$ pip install ravyn

If you want the ravyn client and all the niceties

$ pip install ravyn[standard]

An ASGI server is also needed to run in production, we recommend Uvicorn but it is entirely up to you.

$ pip install uvicorn

If you want install ravyn with specifics:

Support for the internal scheduler:

$ pip install ravyn[schedulers]

Support for the jwt used internally by Ravyn:

$ pip install ravyn[jwt]

If you want to use the ravyn testing client:

$ pip install ravyn[test]

If you want to use the ravyn shell:

More details about this topic in the docs

$ pip install ravyn[ipython] # default shell
$ pip install ravyn[ptpython] # ptpython shell

!!! Warning This is for more advanced users that are already comfortable with Ravyn (or Python in general) or feel like it is not a problem using these directives. If you do not feel comfortable yet to use this, please continue reading the documentation and learning more about Ravyn.

If you wish to start an Ravyn project with a simple suggested structure.

ravyn createproject <YOUR-PROJECT-NAME> --simple

This will generate a scaffold for your project with some pre-defined files in a simple fashion with a simple ready to go Ravyn application.

This will also generate a file for the tests using the ravynTestClient, so make sure you run:

$ pip install ravyn[test]

Or you can skip this step if you don't want to use the ravynTestClient.

You can find more information about this directive and how to use it.

• Fluid and Fast: Thanks to Starlette and Pydantic/msgspec.
• Fast to develop: Thanks to the simplicity of design, the development times can be reduced exponentially.
• Intuitive: If you are used to the other frameworks, Ravyn is a no brainer to develop.
• Easy: Developed with design in mind and easy learning.
• Short: With the OOP available natively there is no need for code duplication. SOLID.
• Ready: Get your application up and running with production-ready code.
• OOP and Functional: Design APIs in any desired way. OOP or Functional is available.
• Async and Sync: Do you prefer sync or async? You can have both.
• Middleware: Apply middlewares on the application level or API level.
• Exception Handlers: Apply exception handlers on any desired level.
• Permissions: Apply specific rules and permissions on each API.
• Interceptors: Intercept requests and add logic before reaching the endpoint.
• Observables - Support for observables allowing to create reactive programming within your application
• Extensions: Create plugins for Ravyn and hook them into any application and/or distribute them.
• DAO and AsyncDAO: Avoid database calls directly from the APIs. Use business objects instead.
• ORM Support: Native support for Edgy.
• ODM Support: Native support for Mongoz.
• Controller: Class Based endpoints for your beloved OOP design.
• JSON serialization/deserialization: Both UJSON and ORJON support.
• Lifespan: Support for the newly lifespan and on_start/on_shutdown events.
• Scheduler: Yes, that's right, it comes with a scheduler for those automated tasks.
• Dependency Injection: Like any other great framework out there.
• Simplicity from settings: Yes, we have a way to make the code even cleaner by introducing settings based systems.
• Encoders - Support for custom encoders allowing compatibility with any favourity validation library: msgspec, attrs....

And a lot more...

Ravyn uses Starlette under the hood. The reason behind this decison comes with the fact that performance is there and no issues with routing.

Once the application is up, all the routes are mounted and therefore the url paths are defined. Ravyn encourages standard practices and design in mind which means that any application, big or small, custom or enterprise, fits within Ravyn ecosystem without scalability issues.

To quickly start with Ravyn, you can just do this. Using uvicorn as example.

#!/usr/bin/env python
import uvicorn

from ravyn import Ravyn, Gateway, JSONResponse, Request, get


@get()
def welcome() -> JSONResponse:
    return JSONResponse({"message": "Welcome to Ravyn"})


@get()
def user(user: str) -> JSONResponse:
    return JSONResponse({"message": f"Welcome to Ravyn, {user}"})


@get()
def user_in_request(request: Request) -> JSONResponse:
    user = request.path_params["user"]
    return JSONResponse({"message": f"Welcome to Ravyn, {user}"})


app = Ravyn(
    routes=[
        Gateway("/ravyn", handler=welcome),
        Gateway("/ravyn/{user}", handler=user),
        Gateway("/ravyn/in-request/{user}", handler=user_in_request),
    ]
)

if __name__ == "__main__":
    uvicorn.run(app, port=8000)

Then you can access the endpoints.

To quickly start with Ravyn you can also use it as decorator, you can just do this. Using uvicorn as example.

#!/usr/bin/env python
import uvicorn

from ravyn import Ravyn, Gateway, JSONResponse, Request, get

app = Ravyn()


@app.get("/ravyn")
def welcome() -> JSONResponse:
    return JSONResponse({"message": "Welcome to Ravyn"})


@app.get("/ravyn/{user}")
def user(user: str) -> JSONResponse:
    return JSONResponse({"message": f"Welcome to Ravyn, {user}"})


@app.get("/ravyn/in-request/{user}")
def user_in_request(request: Request) -> JSONResponse:
    user = request.path_params["user"]
    return JSONResponse({"message": f"Welcome to Ravyn, {user}"})


if __name__ == "__main__":
    uvicorn.run(app, port=8000)

Like every other framework, when starting an application, a lot of settings can/need to be passed to the main object and this can be very dauting and ugly to maintain and see.

Ravyn comes with the settings in mind. A set of defaults that can be overridden by your very own settings module but not limited to it, as you can still use the classic approach of passing everything into a Ravyn instance directly when instantiating.

Example of classic approach:

from example import ExampleObject

# ExampleObject is an instance of another application
# and it serves only for example

app = ExampleObject(setting_one=..., setting_two=..., setting_three=...)

Inspired by the great Django and using pydantic, Ravyn has a default object ready to be used out-of-the-box.

Ravyn:

from ravyn import Ravyn

app = Ravyn()

And that's it! All the default settings are loaded! This is simple of course but can you override inside the object as well? Yes!

from ravyn import Ravyn

app = Ravyn(app_name='My App', title='My title')

Same as the classics.

So how does Ravyn know about the default settings? Enters Ravyn settings module.

This is the way Ravyn defaults the values. When starting an application, the system looks for a RAVYN_SETTINGS_MODULE environment variable. If no variable is supplied then the system will default to RavynSettings settings and start.

Separation of settings by enviromment is a must have these days and starting with default of Ravyn will not be enough for any application.

The settings are pydantic standard settings and therefore compatible with Ravyn. The system brings some defaults that can be used out-of-the-box but it's not mandatory to be used. The environment defaults to production.

from ravyn import RavynSettings
from ravyn.conf.enums import EnvironmentType


class Development(RavynSettings):
    app_name: str = 'My app in dev'
    environment: str = EnvironmentType.DEVELOPMENT

Load the settings into your Ravyn application:

Assuming your Ravyn app is inside an src/app.py.

RAVYN_SETTINGS_MODULE='myapp.settings.Development' python -m src.app.py

Starlette offers the Route classes for simple path assignments but this is also very limiting if something more complex in mind. Ravyn extends that functionality and adds some flair and levels up by having the Gateway, WebSocketGateway and Include.

Those are special objects that allow all the magic of Ravyn to happen.

For a classic, direct, one file single approach.

In a nutshell:

from ravyn import Ravyn, get, status, Request, ORJSONResponse, Gateway, WebSocketGateway, Websocket


@get(status_code=status.HTTP_200_OK)
async def home() -> ORJSONResponse:
    return ORJSONResponse({
        "detail": "Hello world"
    })


@get()
async def another(request: Request) -> dict:
    return {
        "detail": "Another world!"
    }


@websocket(path="/{path_param:str}")
async def world_socket(socket: Websocket) -> None:
    await socket.accept()
    msg = await socket.receive_json()
    assert msg
    assert socket
    await socket.close()


app = Ravyn(routes=[
    Gateway(handler=home),
    Gateway(handler=another),
    WebSocketGateway(handler=world_socket),
])

Good design is always encouraged and Ravyn allows complex routing on any level.

from ravyn import get, post, put, status, websocket, Controller, Request, JSONResponse, Response, WebSocket
from pydantic import BaseModel


class Product(BaseModel):
    name: str
    sku: str
    price: float


@put('/product/{product_id}')
def update_product(product_id: int, data: Product) -> dict:
    return {"product_id": product_id, "product_name": product.name}


@get(status_code=status.HTTP_200_OK)
async def home() -> JSONResponse:
    return JSONResponse({
        "detail": "Hello world"
    })


@get()
async def another(request: Request) -> dict:
    return {
        "detail": "Another world!"
    }


@websocket(path="/{path_param:str}")
async def world_socket(socket: Websocket) -> None:
    await socket.accept()
    msg = await socket.receive_json()
    assert msg
    assert socket
    await socket.close()


class World(Controller):

    @get(path='/{url}')
    async def home(self, request: Request, url: str) -> Response:
        return Response(f"URL: {url}")

    @post(path='/{url}', status_code=status.HTTP_201_CREATED)
    async def mars(self, request: Request, url: str) -> JSONResponse:
        ...

    @websocket(path="/{path_param:str}")
    async def pluto(self, socket: Websocket) -> None:
        await socket.accept()
        msg = await socket.receive_json()
        assert msg
        assert socket
        await socket.close()

If a path is not provided, defaults to /.

from ravyn import Gateway, WebSocketGateway
from .controllers import home, another, world_socket, World

route_patterns = [
    Gateway(handler=update_product),
    Gateway(handler=home),
    Gateway(handler=another),
    Gateway(handler=World),
    WebSocketGateway(handler=world_socket),
]

If a path is not provided, defaults to /.

This is a very special object that allows the import of any route from anywhere in the application.

Include accepts the import via namespace or via routes list but not both.

When using a namespace, the Include will look for the default route_patterns object list in the imported namespace unless a different pattern is specified.

The pattern only works if the imports are done via namespace and not via routes.

Importing using namespace:

from ravyn import Include

route_patterns = [
    Include(namespace='myapp.accounts.urls')
]

Importing using routes:

from ravyn import Include
from myapp.accounts import urls

route_patterns = [
    Include(routes=urls.route_patterns)
]

If a path is not provided, defaults to /.

from ravyn import Gateway, WebSocketGateway
from .controllers import home, another, world_socket, World

my_urls = [
    Gateway(handler=update_product),
    Gateway(handler=home),
    Gateway(handler=another),
    Gateway(handler=World),
    WebSocketGateway(handler=world_socket),
]

Importing using namespace:

from ravyn import Include

route_patterns = [
    Include(namespace='myapp.accounts.urls', pattern='my_urls')
]

The Include can be very helpful mostly when the goal is to avoid a lot of imports and massive list of objects to be passed into one single object. This can be particulary useful to make a Ravyn instance.

Example:

from ravyn import Include

route_patterns = [
    Include(namespace='myapp.accounts.urls', pattern='my_urls')
]

from ravyn import Ravyn, Include

app = Ravyn(routes=[Include('src.urls')])

As mentioned before, we recommend uvicorn for production but it's not mandatory.

Using uvicorn:

uvicorn src:app --reload

INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)
INFO:     Started reloader process [28720]
INFO:     Started server process [28722]
INFO:     Waiting for application startup.
INFO:     Application startup complete.

Using uvicorn:

RAVYN_SETTINGS_MODULE=myapp.AppSettings uvicorn src:app --reload

INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)
INFO:     Started reloader process [28720]
INFO:     Started server process [28722]
INFO:     Waiting for application startup.
INFO:     Application startup complete.

Ravyn also comes with OpenAPI docs integrated. For those used to that, this is roughly the same and to make it happen, there were inspirations that helped Ravyn getting there fast.

Ravyn starts automatically the OpenAPI documentation by injecting the OpenAPIConfig default from the settings and makes Swagger, ReDoc an Stoplight elements available to you out of the box.

To access the OpenAPI, simply start your local development and access:

• Swagger - /docs/swagger.
• Redoc - /docs/redoc.
• Stoplight Elements - /docs/elements.

There are more details about how to configure the OpenAPIConfig within the documentation.

There is also a good explanation on how to use the OpenAPIResponse as well.

This is just a very high-level demonstration of how to start quickly and what Ravyn can do. There are plenty more things you can do with Ravyn. Enjoy! 😊

Currently there are no sponsors of Ravyn but you can financially help and support the author though GitHub sponsors and become a Special one or a Legend.

Worth mentioning who is helping us.

JetBrains