Configure Codespaces for Python projects

One of the great advantage of Codespaces is the ability to preconfigure the environment so you do not need to waste time installing and configuring your toolchain. Python is a perfect example of this scenario, I’ve a small project to generate Git Graph Representation and since I’m not a full time Python developer, I’ve not it installed and perfectly configured in all of my environment. Also I primarily work on Windows, so Codespaces allows me to test everything on Linux with a single click.

Configuring a codespaces for Python is really simple because you can already start for a prebuild Python enabled container that you only need to configure. In devcontainer.json lots of stuff are already preconfigured for Python, I only needed to add my personal preferences, like editor.fontsize and adding some other extension for Visual Studio Code.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

{	
    // ...
	// Set *default* container specific settings.json values on container create.
	"settings": { 
		"python.pythonPath": "/usr/local/bin/python",
		"python.languageServer": "Pylance",
		"python.linting.enabled": true,
		"python.linting.pylintEnabled": true,
		"python.formatting.autopep8Path": "/usr/local/py-utils/bin/autopep8",
		"python.formatting.blackPath": "/usr/local/py-utils/bin/black",
		"python.formatting.yapfPath": "/usr/local/py-utils/bin/yapf",
		"python.linting.banditPath": "/usr/local/py-utils/bin/bandit",
		"python.linting.flake8Path": "/usr/local/py-utils/bin/flake8",
		"python.linting.mypyPath": "/usr/local/py-utils/bin/mypy",
		"python.linting.pycodestylePath": "/usr/local/py-utils/bin/pycodestyle",
		"python.linting.pydocstylePath": "/usr/local/py-utils/bin/pydocstyle",
		"python.linting.pylintPath": "/usr/local/py-utils/bin/pylint",
		"editor.fontSize": 18
	},

	// Add the IDs of extensions you want installed when the container is created.
	"extensions": [
		"streetsidesoftware.code-spell-checker",
		"ms-python.python",
		"ms-python.vscode-pylance"
	],
    // ...
}

If you scroll down a little bit, you find the section that uses requirements.txt standard Python file to automatically install pip packages. Just uncomment it and you will have all needed pip packages preinstalled in your codespaces. In my example I have pytest and other stuff.

1
2

	// Use 'postCreateCommand' to run commands after the container is created.
	"postCreateCommand": "pip3 install --user -r requirements.txt",

As usual you can go into dockerfile to add more customization. One thing I miss in the basic codespaces environment is git-flow, so I use apt-get section to install it and have it ready. In this project I tried to use pantomjs so I also use the npm section to install npm packages.

1
2
3
4
5
6

# [Optional] Uncomment this section to install additional OS packages.
RUN apt-get update && export DEBIAN_FRONTEND=noninteractive \
    && apt-get -y install --no-install-recommends git-flow

# [Optional] Uncomment this line to install global node packages.
RUN su vscode -c "source /usr/local/share/nvm/nvm.sh && npm install -g phantomjs" 2>&1

Thanks to this configuration, if I need to work on this project I’m ready to go with a simple click. The only drawback of this approach is that, in this project, I used selenium and playwrigth to automate a browser, my first tentative was selenium, but phantom.js refuses to work (it is deprecated) and I was not able to install chrome into the container (probably my fault). I’ve then tried playwright but also it refuses to work inside a container.

This is the very first real limitation of Codespaces I’ve found, if you need to work with selenium/playwright to do browser automation, it does not work out of the box inside the container.

Actually if you want Playwright to work in docker you need to have a different approach described in this GitHub issue. For this project I’ve not tried to fix this problem, I can still live with this limitation (I’m testing the full program in Windows or Linux full environment).

Another simple problem I’ve found is test runner was not able to identify tests, this is caused by a stupid error of mine, I’ve mistakenly added the Windows path of Python environment inside the settings.json.

1
2
3
4
5
6
7
8
9

{
    "python.testing.unittestEnabled": false,
    "python.testing.nosetestsEnabled": false,
    "python.testing.pytestEnabled": true,
    "python.analysis.extraPaths": [
        "./src/git-graph-automation"
    ],
    "python.pythonPath": "C:\\programfiles\\python3\\python.exe"
}

When you work in different environments you should avoid putting environment dependant settings inside your .vscode file. Once I’ve removed the python.pythonPath settings everything works perfectly.

And, as usual, this post was written with Codespaces,

Gian Maria.