Tutorial: Serving video#

In this tutorial we will build a very basic video server. It will serve a video directly.

This tutorial is meant to serve as an introduction to serving large files with conditional responses in Quart. If you want to skip to the end the code is on Github.

1: Creating the project#

We need to create a project for our video server, I like to use Poetry to do this. Poetry is installed via pip (or via Brew):

pip install poetry

We can then use Poetry to create a new video project:

poetry new --src video

Our project can now be developed in the video directory, and all subsequent commands should be in run the video directory.

2: Adding the dependencies#

We only need Quart to build this simple video server, which we can install as a dependency of the project by running the following:

poetry add quart

Poetry will ensure that this dependency is present and the paths are correct by running:

poetry install

3: Creating the app#

We need a Quart app to be our web server, which is created by the following addition to src/video/__init__.py:

from quart import Quart

app = Quart(__name__)

def run() -> None:

To make the app easy to run we can call the run method from a poetry script, by adding the following to pyproject.toml:

start = "video:run"

Which allows the following command to start the app:

poetry run start

4: Serving the UI#

When users visit our website we will show them the same video served directly, and via chunks. The following HTML template should be added to src/video/templates/index.html:

<video controls width="100%">
  <source src="/video.mp4" type="video/mp4">

This is a very basic UI in terms of the styling.

We can now serve this template for the root path i.e. / by adding the following to src/video/__init__.py:

from quart import render_template

async def index():
    return await render_template("index.html")

5: Implementing the routes#

As we are serving a large file we should allow for conditional responses. This is where the data returned in the response is conditional on what the request asked for. This is done via the Range header field which can be inspected via the request.range attribute.

Quart has in-built methods to make a response conditional on the request range. The first is to use the conditional argument when sending a file, the second is to use the response make_conditional method. The former is shown below, which should be added to src/video/__init__.py:

async def auto_video():
    return await send_file(app.static_folder / "video.mp4", conditional=True)

6: Testing#

To test our app we need to check that the full video is returned unless a conditional range request is made. This is done by adding the following to tests/test_video.py:

from video import app

async def test_auto_video() -> None:
    test_client = app.test_client()
    response = await test_client.get("/video.mp4")
    data = await response.get_data()
    assert len(data) == 255_849

    response = await test_client.get("/video.mp4", headers={"Range": "bytes=200-1000"})
    data = await response.get_data()
    assert len(data) == 801

As the test is an async function we need to install pytest-asyncio by running the following:

poetry add --dev pytest-asyncio

Once installed it needs to be configured by adding the following to pyproject.toml:

asyncio_mode = "auto"

Finally we can run the tests via this command:

poetry run pytest tests/

If you are running this in the Quart example folder you’ll need to add a -c pyproject.toml option to prevent pytest from using the Quart pytest configuration.

7: Summary#

We’ve built a server that will serve large files conditionally as requested by the client, including the ability to limit the maximum partial size.