Source code: https://github.com/volfpeter/fasthx
Documentation and examples: https://volfpeter.github.io/fasthx
FastAPI server-side rendering with built-in HTMX support.
- Decorator syntax that works with FastAPI as one would expect, no need for unused or magic dependencies in routes.
- Built for HTMX, but can be used without it.
- Works with any templating engine or server-side rendering library, e.g.
htmy,jinja2, ordominate. - Gives the rendering engine access to all dependencies of the decorated route.
- HTMX routes work as expected if they receive non-HTMX requests, so the same route can serve data and render HTML at the same time.
- Response headers you set in your routes are kept after rendering, as you would expect in FastAPI.
- Correct typing makes it possible to apply other (typed) decorators to your routes.
- Works with both sync and async routes.
Consider supporting the development and maintenance of the project through sponsoring, or reach out for consulting so you can get the most out of the library.
The package is available on PyPI and can be installed with:
$ pip install fasthxThe package has optional dependencies for the following official integrations:
The core concept of FastHX is to let FastAPI routes do their usual job of handling the business logic and returning the result, while the FastHX decorators take care of the entire rendering / presentation layer using a declarative, decorator-based approach.
Interally, FastHX decorators always have access to the decorated route's result, all of its arguments (sometimes called the request context), and the current request. Integrations convert these values into data that can be consumed by the used rendering engine (for example htmy or jinja), run the rendering engine with the selected component (more on this below) and the created data, and return the result to the client. For more details on how data conversion works and how it can be customized, please see the API documentation of the rendering engine integration of your choice.
The ComponentSelector abstraction makes it possible to declaratively specify and dynamically select the component that should be used to render the response to a given request. It is also possible to define an "error" ComponentSelector that is used if the decorated route raises an exception -- a typical use-case being error rendering for incorrect user input.
For complete, but simple examples that showcase the basic use of FastHX, please see the examples folder.
Requires: pip install fasthx[htmy].
Serving HTML and HTMX requests with htmy is as easy as creating a fasthx.htmy.HTMY instance and using its hx() and page() decorator methods on your routes.
The example below assumes the existence of an IndexPage and a UserList htmy component. The full working example with the htmy components can be found here.
from datetime import date
from fastapi import FastAPI
from pydantic import BaseModel
from fasthx.htmy import HTMY
# Pydantic model for the application
class User(BaseModel):
name: str
birthday: date
# Create the FastAPI application.
app = FastAPI()
# Create the FastHX HTMY instance that renders all route results.
htmy = HTMY()
@app.get("/users")
@htmy.hx(UserList) # Render the result using the UserList component.
def get_users(rerenders: int = 0) -> list[User]:
return [
User(name="John", birthday=date(1940, 10, 9)),
User(name="Paul", birthday=date(1942, 6, 18)),
User(name="George", birthday=date(1943, 2, 25)),
User(name="Ringo", birthday=date(1940, 7, 7)),
]
@app.get("/")
@htmy.page(IndexPage) # Render the index page.
def index() -> None: ...Requires: pip install fasthx[jinja].
To start serving HTML and HTMX requests, all you need to do is create an instance of fasthx.Jinja and use its hx() or page() methods as decorators on your routes. hx() only triggers HTML rendering for HTMX requests, while page() unconditionally renders HTML. See the example code below:
from fastapi import FastAPI
from fastapi.templating import Jinja2Templates
from fasthx import Jinja
from pydantic import BaseModel
# Pydantic model of the data the example API is using.
class User(BaseModel):
first_name: str
last_name: str
# Create the app.
app = FastAPI()
# Create a FastAPI Jinja2Templates instance and use it to create a
# FastHX Jinja instance that will serve as your decorator.
jinja = Jinja(Jinja2Templates("templates"))
@app.get("/")
@jinja.page("index.html")
def index() -> None:
...
@app.get("/user-list")
@jinja.hx("user-list.html")
async def htmx_or_data() -> list[User]:
return [
User(first_name="John", last_name="Lennon"),
User(first_name="Paul", last_name="McCartney"),
User(first_name="George", last_name="Harrison"),
User(first_name="Ringo", last_name="Starr"),
]
@app.get("/admin-list")
@jinja.hx("user-list.html", no_data=True)
def htmx_only() -> list[User]:
return [User(first_name="Billy", last_name="Shears")]See the full working example here.
Requires: pip install fasthx.
If you would like to use a rendering engine without FastHX integration, you can easily build on the hx() and page() decorators which give you all the functionality you will need. All you need to do is implement the HTMLRenderer protocol.
Similarly to the Jinja case, hx() only triggers HTML rendering for HTMX requests, while page() unconditionally renders HTML. See the example code below:
from typing import Annotated, Any
from fastapi import Depends, FastAPI, Request
from fasthx import hx, page
# Create the app.
app = FastAPI()
# Create a dependecy to see that its return value is available in the render function.
def get_random_number() -> int:
return 4 # Chosen by fair dice roll.
DependsRandomNumber = Annotated[int, Depends(get_random_number)]
# Create the render methods: they must always have these three arguments.
# If you're using static type checkers, the type hint of `result` must match
# the return type annotation of the route on which this render method is used.
def render_index(result: list[dict[str, str]], *, context: dict[str, Any], request: Request) -> str:
return "<h1>Hello FastHX</h1>"
def render_user_list(result: list[dict[str, str]], *, context: dict[str, Any], request: Request) -> str:
# The value of the `DependsRandomNumber` dependency is accessible with the same name as in the route.
random_number = context["random_number"]
lucky_number = f"<h1>{random_number}</h1>"
users = "".join(("<ul>", *(f"<li>{u['name']}</li>" for u in result), "</ul>"))
return f"{lucky_number}\n{users}"
@app.get("/")
@page(render_index)
def index() -> None:
...
@app.get("/htmx-or-data")
@hx(render_user_list)
def htmx_or_data(random_number: DependsRandomNumber) -> list[dict[str, str]]:
return [{"name": "Joe"}]
@app.get("/htmx-only")
@hx(render_user_list, no_data=True)
async def htmx_only(random_number: DependsRandomNumber) -> list[dict[str, str]]:
return [{"name": "Joe"}]See the full working example here.
- FastAPI-HTMX-Tailwind example: A complex
Jinja2example with features like active search, lazy-loading, server-sent events, custom server-side HTMX triggers, dialogs, and TailwindCSS and DaisyUI integration.
The only dependency of this package is fastapi.
Use ruff for linting and formatting, mypy for static code analysis, and pytest for testing.
The documentation is built with mkdocs-material and mkdocstrings.
We welcome contributions from the community to help improve the project! Whether you're an experienced developer or just starting out, there are many ways you can contribute:
- Discuss: Join our Discussion Board to ask questions, share ideas, provide feedback, and engage with the community.
- Document: Help improve the documentation by fixing typos, adding examples, and updating guides to make it easier for others to use the project.
- Develop: Prototype requested features or pick up issues from the issue tracker.
- Share: Share your own project by adding it to the external examples list, helping others discover and benefit from your work.
- Test: Write tests to improve coverage and enhance reliability.
The package is open-sourced under the conditions of the MIT license.
Thank you to Smart-Now for supporting the project.
