Datastar Python SDK

The datastar-py package provides a Python SDK for working with Datastar.

Datastar sends responses back to the browser using SSE. This allows the backend to send any number of events, from zero to infinity in response to a single request.

datastar-py has helpers for creating those responses, formatting the events, reading signals from the frontend, and generating the data-* HTML attributes.

The event generator can be used with any framework. There are also custom helpers included for the following frameworks:

Framework-specific helpers are kept in their own packages. e.g. datastar_py.quart Make sure to use the helpers from the package of the framework you are using.

Here is a full example using the quart framework showing many of the features available in this package.

import asyncio
from datetime import datetime

from datastar_py import ServerSentEventGenerator as SSE, attribute_generator as data
from datastar_py.quart import datastar_response, read_signals
from quart import Quart

app = Quart(__name__)

# Import frontend library via Content Distribution Network, create targets for Server Sent Events
@app.route("/")
def index():
    return f"""
        <html>
            <head>
                <script type="module" src="https://cdn.jsdelivr.net/gh/starfederation/datastar@v1.0.0-RC.7/bundles/datastar.js"></script>
            </head>
            <body {data.init("@get('/updates')")}>
                <span id="currentTime"></span><br>
                <span data-text="$currentTime"></span>
            </body>
        </html>
    """


@app.route("/updates")
@datastar_response
async def updates():
    # Retrieve a dictionary with the current state of the signals from the frontend
    signals = await read_signals()
    # Alternate updating an element from the backend, and updating a signal from the backend
    while True:
        yield SSE.patch_elements(
            f"""<span id="currentTime">{datetime.now().isoformat()}"""
        )
        await asyncio.sleep(1)
        yield SSE.patch_signals({"currentTime": f"{datetime.now().isoformat()}"})
        await asyncio.sleep(1)


app.run()

Starting examples for each framework can be found in the examples directory.

Event Generation Helpers

This helper is used to generate the actual events that are sent over SSE. They are just text blobs that can be sent using any framework. These can even be used by frameworks not directly supported in this library if you set up the headers of the SSE response yourself.

Response Helpers

A datastar response consists of 0..N datastar events. There are response classes included to make this easy in all of the supported frameworks. Each framework also exposes a @datastar_response decorator that will wrap return values (including generators) into the right response class while preserving sync handlers as sync so frameworks can keep them in their threadpools.

The following examples will work across all supported frameworks when the response class is imported from the appropriate framework package. e.g. from datastar_py.quart import DatastarResponse The containing functions are not shown here, as they will differ per framework.

# per framework Response import. (Replace 'fastapi' with your framework.) e.g.:
# from datastar_py.fastapi import DatastarResponse
from datastar_py import ServerSentEventGenerator as SSE

# 0 events, a 204
@app.get("zero")
def zero_event():
    return DatastarResponse()
# 1 event
@app.get("one")
def one_event():
    return DatastarResponse(SSE.patch_elements("<div id='mydiv'></div>"))
# 2 events
@app.get("two")
def two_event():
    return DatastarResponse([
        SSE.patch_elements("<div id='mydiv'></div>"),
        SSE.patch_signals({"mysignal": "myval"}),
    ])

# N events, a long lived stream (for all frameworks but sanic)
@app.get("/updates")
async def updates():
    async def _():
        while True:
            yield SSE.patch_elements("<div id='mydiv'></div>")
            await asyncio.sleep(1)
    return DatastarResponse(_())

# A long lived stream for sanic
@app.get("/updates")
async def updates(request):
    response = await datastar_respond(request)
    # which is just a helper for the following
    # response = await request.respond(DatastarResponse())
    while True:
        await response.send(SSE.patch_elements("<div id='mydiv'></div>"))
        await asyncio.sleep(1)

Response Decorator

To make returning a DatastarResponse simpler, there is a decorator datastar_response available that automatically wraps a function result in DatastarResponse. It works on async and regular functions and generator functions. The main use case is when using a generator function, as you can avoid a second generator function inside your response function. The decorator works the same for any of the supported frameworks, and should be used under any routing decorator from the framework.

# Import the decorator from the package specific to your framework
from datastar_py.sanic import datastar_response, ServerSentEventGenerator as SSE

@app.get('/my_route')
@datastar_response
async def my_route(request):
    while True:
        yield SSE.patch_elements("<div id='mydiv'></div>")
        await asyncio.sleep(1)

Signal Helpers

The current state of the datastar signals is included by default in every datastar request. A helper is included to load those signals for each framework. read_signals. The usage varies per framework so check the signature for your framework. You usually need to pass the request in.

from datastar_py.quart import read_signals

@app.route("/updates")
async def updates():
    signals = await read_signals()

Attribute Generation Helper

Datastar allows HTML generation to be done on the backend. datastar-py includes a helper to generate data-* attributes in your HTML with IDE completion and type checking. It can be used with many different HTML generation libraries.

from datastar_py import attribute_generator as data

# htpy
button(data.on("click", "console.log('clicked')").debounce(1000).stop)["My Button"]
# FastHTML
Button("My Button", data.on("click", "console.log('clicked')").debounce(1000).stop)
Button(data.on("click", "console.log('clicked')").debounce(1000).stop)("My Button")
# f-strings
f"<button {data.on("click", "console.log('clicked')").debounce(1000).stop}>My Button</button>"
# Jinja, but no editor completion :(
<button {{data.on("click", "console.log('clicked')").debounce(1000).stop}}>My Button</button>

When using datastar with a different alias, you can instantiate the class yourself.

from datastar_py.attributes import AttributeGenerator

data = AttributeGenerator(alias="data-star-")

# htmy (htmy will transform _ into - unless the attribute starts with _, which will be stripped)
data = AttributeGenerator(alias="_data-")
html.button("My Button", **data.on("click", "console.log('clicked')").debounce("1s").stop)

Release files for datastar-py 1.0.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for datastar-py 1.0.2
File	Size	Uploaded
datastar_py-1.0.2.tar.gz	129.5 kB	Jun 2, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for datastar-py 1.0.2
File	Interpreter	ABI	Platform	Reset
datastar_py-1.0.2-py3-none-any.whl 19.7 kB Jun 2, 2026	Python 3	none	any	Details

Total release size: 149.2 kB

Release files / datastar_py-1.0.2.tar.gz

Download URL	datastar_py-1.0.2.tar.gz
Size	129.5 kB
Tags	Source
SHA-256 checksum How to use checksums	`c6f86e2b48c3a5ea08570a4e958267307a5ffc0253c259f054dc3bb6de8a2c0e`
BLAKE2b-256 checksum How to use checksums	`d956697b0c2300985f5143384696fd4854cfdcadf6839577fde49d5b9e0d9000`
Upload date	Jun 2, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`uv/0.11.17 {"installer":{"name":"uv","version":"0.11.17","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"CachyOS Linux","version":null,"id":null,"libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}`

Release files / datastar_py-1.0.2-py3-none-any.whl

Download URL	datastar_py-1.0.2-py3-none-any.whl
Size	19.7 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`d84735e5cc55ed85e02f80bbab01c2877e7ed7a6185cc144f802a9821616c9d3`
BLAKE2b-256 checksum How to use checksums	`0f142cffb0ef9a69d113349f483786c435abf06ea5091e275328cc163b2a2f23`
Upload date	Jun 2, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`uv/0.11.17 {"installer":{"name":"uv","version":"0.11.17","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"CachyOS Linux","version":null,"id":null,"libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}`