3.8 KiB
Python Server Development
Overview
Ideally, Python code should be:
- Broken up into small, clear, testable functions.
- Documented with docstrings (For conventions see below).
- Type hinted
- Covered by unit tests (using pytest)
The server code is going through significant flux during the transition from v2 to v3, so currently our test coverage is being rebuilt and type checking does not yet pass.
Python environment, build, and dependencies
As of v3 we specify dependencies in pyproject.toml. These are not currently frozen due to difficulties matching versions on different platforms. On Raspberry Pi, it's best to specify --only-binary=:all: when installing packages, to ensure libraries like numpy and scipy are not compiled from source (which takes many hours, and/or fails). Pinning dependencies with requirements.txt and/or requirements.in may happen in the future.
Python: Formatting, linting, and tests
Ruff is used both for linting and formatting the codebase. Our custom linter rules are in our pyproject.toml.
All of the commands below assume that you are running in the OFM virtual environment, i.e. you have run
ofm activateon an OpenFlexure SD card, orsource .venv/bin/activateon Linux, or.venv/Scripts/activateon Windows Powershell, or.venv\Scripts\activate.baton Windows CMD.
Before committing you should lint and auto-format your code:
- To lint, run
ruff checkand manually fix any flagged issues - To auto-format the Python code run
ruff format
Before pushing please auto-format your code and also run the quality checks (linting, static analysis, and unit tests)
- All commands above
pytest
We use several code analysis and formatting libraries in this project. Our CI will check each of these automatically, so ensuring they pass locally will save you time. Currently ruff is used for linting/formatting and pytest for unit tests. mypy will be enabled once the codebase is ready.
To check your code before each commit we recommend putting the following script in tour .git/hooks directory with the filename pre-commit.
#!/bin/bash
set -e
function angrymessage()
{
RED='\033[1;31m'
NC='\033[0m' # No Color
printf "\n${RED}Not committing as ruff failed${NC}\n\n"
}
trap angrymessage ERR
ruff format --check
ruff check
Docstrings conventions.
We write docstrings in reStructuredText, and process them with pydoctor.
For function arguments, return values, and exceptions we use the following style recognised by reStructuredText:
def a_function(arg1: str, arg2: int) -> tuple[float, int]:
"""A one sentence summary, in imperative mood, ending with punctuation.
More detail after a blank line. This could be a whole paragraph.
When referring to ``code`` in reStructuredText used double backticks
:param arg1: This is the first argument, we can say everything we need to about it.
If the text wraps onto a new line, this line is indented.
:param arg2: This is the second argument.
:returns: Say what it returns without anything within the colons. If returning
multiple values, a bulleted list is a good idea - just make sure:
* It is indented to 1 level in
* There is a blank line before the bullets
* The bullet points are ``*`` not ``-``
"""
+The linter rules we use for docstrings are defined by pydocstyle. We do not use rules D203, D213, D400 for reasons explained in our pyproject.toml file.
The documentation generated by pydoctor from docstrings can be viewed in any merge request (if the CI pipeline has passed) using the "View App" button.