Table of contents
Style and static checks
Pre-commit
PY006 Scientific Python projects often use pre-commit to check code style. It can be installed through brew (macOS) or pip (anywhere). There are two modes to use it locally; you can check manually with pre-commit run (changes only) or pre-commit run --all-files (all). You can also run pre-commit install to add checks as a git pre-commit hook (which is where it gets its name). It’s worth trying, even if you’ve tried and failed to set up a custom pre-commit hook before; it’s quite elegant and does not add or commit the changes, it just makes the changes and allows you to check and add them. You can always override the hook with -n.
PC100 Here is a minimal .pre-commit-config.yaml file with some handy options:
repos:
- repo: https://github.com/pre-commit/pre-commit-hooks
rev: "v5.0.0"
hooks:
- id: check-added-large-files
- id: check-case-conflict
- id: check-merge-conflict
- id: check-symlinks
- id: check-yaml
- id: debug-statements
- id: end-of-file-fixer
- id: mixed-line-ending
- id: name-tests-test
args: ["--pytest-test-first"]
- id: requirements-txt-fixer
- id: trailing-whitespace
Helpful tip: Pre-commit runs top-to-bottom, so put checks that modify content (like the several of the pre-commit-hooks above, or Black) above checks that might be more likely to pass after the modification (like flake8).
Keeping pinned versions fresh: You can use pre-commit autoupdate to move your tagged versions forward to the latest tags! Due to the design of pre-commit’s caching system, these must point at fixed tags, never put a branch here.
Checking in CI: You can have this checked and often automatically corrected for you using pre-commit.ci. It will even update your rev: versions every week or so if your checks update!
To use, just go to pre-commit.ci, click “Log in with GitHub”, click “Add an Installation” if adding for the first time for an org or user, or “Manage repos on GitHub” for an existing installation, then add your repository from the list in GitHub’s interface.
Now there will be a new check, and pre-commit.ci will commit changes if the pre-commit check made any changes. Note that there are a couple of missing features: Docker based checks will not work (pre-commit.ci already runs in docker), you cannot enable a --manual flag, so extra checks will not run, and jobs should not download packages (use additional-dependencies: to add what you need).
PC901 You can customize the pre-commit message with:
ci:
autoupdate_commit_msg: "chore: update pre-commit hooks"
Format
PC110 Black is a popular auto-formatter from the Python Software Foundation. One of the main features of Black is that it is “opinionated”; that is, it is almost completely unconfigurable. Instead of allowing you to come up with your own format, it enforces one on you. While I am quite sure you can come up with a better format, having a single standard makes it possible to learn to read code very fast - you can immediately see nested lists, matching brackets, etc. There also is a faction of developers that dislikes all auto-formatting tools, but inside a system like pre-commit, auto-formatters are ideal. They also speed up the writing of code because you can ignore formatting your code when you write it. By imposing a standard, all scientific Python developers can quickly read any package’s code.
Also, properly formatted code has other benefits, such as if two developers make the same change, they get the same formatting, and merge requests are easier. The style choices in Black were explicitly made to optimize git diffs!
There are a few options, mostly to enable/disable certain files, remove string normalization, and to change the line length, and those go in your pyproject.toml file.
Ruff, the powerful Rust-based linter, has a formatter that is designed with the help of some of the Black authors to look 99.9% like Black, but run 30x faster. Here is the snippet to add the formatter to your .pre-commit-config.yml (combine with the Ruff linter below):
- repo: https://github.com/astral-sh/ruff-pre-commit
rev: "v0.7.1"
hooks:
# id: ruff would go here if using both
- id: ruff-format
As you likely will be using Ruff if you follow this guide, the formatter is recommended as well.
You can add a Ruff badge to your repo as well
[](https://github.com/astral-sh/ruff)
.. image:: https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/main/assets/badge/format.json
:target: https://github.com/astral-sh/ruff
In very specific situations, like when making a 2D array, you may want to retain special formatting. After carefully deciding that it is a special use case, you can use # fmt: on and # fmt: off around a code block to have it keep custom formatting. Always consider refactoring before you try this option! Most of the time, you can find a way to make the Blacked code look better by rewriting your code; factor out long unreadable portions into a variable, avoid writing matrices as 1D lists, etc.
Documentation / README snippets support
PC111 If you want Black used in your documentation, you can use blacken-docs. This can even catch syntax errors in code snippets! It supports markdown and restructured text. Note that because black is in additional_dependencies, you’ll have to keep it up to date manually.
- repo: https://github.com/adamchainz/blacken-docs
rev: "1.19.1"
hooks:
- id: blacken-docs
additional_dependencies: [black==24.*]
Ruff
PC190 Ruff (docs) is a Python code linter and autofixer that replaces many other tools in the ecosystem with a ultra-fast (written in Rust), single zero-dependency package. All plugins are compiled in, so you can’t get new failures from plugins updating without updating your pre-commit hook.
- repo: https://github.com/astral-sh/ruff-pre-commit
rev: "v0.7.1"
hooks:
- id: ruff
args: ["--fix", "--show-fixes"]
PC191 The --fix argument is optional, but recommended, since you can inspect and undo changes in git.
RF001 Ruff is configured in your pyproject.toml. Here’s an example:
[tool.ruff.lint]
extend-select = [
"ARG", # flake8-unused-arguments
"B", # flake8-bugbear
"C4", # flake8-comprehensions
"EM", # flake8-errmsg
"EXE", # flake8-executable
"FURB", # refurb
"G", # flake8-logging-format
"I", # isort
"ICN", # flake8-import-conventions
"NPY", # NumPy specific rules
"PD", # pandas-vet
"PGH", # pygrep-hooks
"PIE", # flake8-pie
"PL", # pylint
"PT", # flake8-pytest-style
"PTH", # flake8-use-pathlib
"PYI", # flake8-pyi
"RET", # flake8-return
"RUF", # Ruff-specific
"SIM", # flake8-simplify
"T20", # flake8-print
"UP", # pyupgrade
"YTT", # flake8-2020
]
ignore = [
"ISC001", # Conflicts with formatter
"PLR09", # Too many <...>
"PLR2004", # Magic value used in comparison
]
typing-modules = ["mypackage._compat.typing"]
isort.required-imports = ["from __future__ import annotations"]
[tool.ruff.lint.per-file-ignores]
"tests/**" = ["T20"]
Ruff provides dozens of rule sets; you can select what you want from these. Like Flake8, plugins match by whole letter sequences (with the special exception of pylint’s “PL” shortcut), then you can also include leading or whole error codes. Codes starting with 9 must be selected explicitly, with at least the letters followed by a 9. You can also ignore certain error codes via ignore. You can also set codes per paths to ignore in per-file-ignores. If you don’t like certain auto-fixes, you can disable auto-fixing for specific error codes via unfixable.
There are other configuration options, such as typing-modules, which helps apply typing-specific rules to a re-exported typing module (a common practice for unifying typing and typing_extensions based on Python version). There’s also a file exclude set, which you can override if you are running this entirely from pre-commit (default excludes include “build”, so if you have a build module or file named build.py, it would get skipped by default without this).
The src list which tells it where to look for top level packages is no longer needed if just set to ["src"] in Ruff 0.6+. RF003
If you don’t use a
[project]table (older setuptools or Poetry), then you should also set:target-version = "py39"This selects the minimum version you want to target (primarily for
"UP"and"I") RF002
Here are some good error codes to enable on most (but not all!) projects:
E,F,W: These are the standard flake8 checks, classic checks that have stood the test of time. Not required if you useextend-select(Wnot needed if you use a formatter)B: This finds patterns that are very bug-prone. RF101I: This sorts your includes. There are multiple benefits, such as smaller diffs, fewer conflicts, a way to auto-inject__future__imports, and easier for readers to tell what’s built-in, third-party, and local. It has a lot of configuration options, but defaults to a Black-compatible style. RF102ARG: This looks for unused arguments. You might need to# noqa: ARG001occasionally, but it’s overall pretty useful.C4: This looks for places that could use comprehensions, and can autofix them.EM: Very opinionated trick for error messages: it stops you from putting the error string directly in the exception you are throwing, producing a cleaner traceback without duplicating the error string.ISC: Checks for implicit string concatenation, which can help catch mistakes with missing commas. (May collide with formatter)PGH: Checks for patterns, such as type ignores or noqa’s without a specific error code.PL: A set of four code groups that cover some (200 or so out of 600 rules) of PyLint.PT: Helps tests follow best pytest practices. A few codes are not ideal, but many are helpful.PTH: Want to move to using modern pathlib? This will help. There are some cases where performance matters, but otherwise, pathlib is easier to read and use.RUF: Codes specific to Ruff, including removing noqa’s that aren’t used.T20: Disallowprintin your code (built on the assumption that it’s a common debugging tool).UP: Upgrade old Python syntax to yourtarget-version. RF103FURB: From the refurb tool, a collection of helpful cleanups.PYI: Typing related checks
A few others small ones are included above, and there are even more available in Ruff. You can use ALL to get them all, then ignore the ones you want to ignore. New checks go into --preview before being activated in a minor release.
You can add a Ruff badge to your repo as well
)[)](https://github.com/astral-sh/ruff)
.. image:: https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/main/assets/badge/v2.json
:target: https://github.com/astral-sh/ruff
Separate tools that Ruff replaces
PyCln
PyCln will clean up your imports if you have any that are not needed. There is a Flake8 check for this, but it’s usually nicer to automatically do the cleanup instead of forcing a user to manually delete unneeded imports. If you use the manual stage, it’s opt-in instead of automatic.
- repo: https://github.com/hadialqattan/pycln
rev: "v2.4.0"
hooks:
- id: pycln
args: [--all]
stages: [manual]
Flake8
Flake8 can check a collection of good practices for you, ranging from simple style to things that might confuse or detract users, such as unused imports, named values that are never used, mutable default arguments, and more. Unlike black and some other tools, flake8 does not correct problems, it just reports them. Some of the checks could have had automated fixes, sadly (which is why Black is nice). Here is a suggested .flake8 or setup.cfg to enable compatibility with Black (flake8 does not support pyproject.toml configuration, sadly):
[flake8]
extend-ignore = E203, E501
One recommended plugin for flake8 is flake8-bugbear, which catches many common bugs. It is highly opinionated and can be made more so with the B9 setting. You can also set a max complexity, which bugs you when you have complex functions that should be broken up. Here is an opinionated config:
[flake8]
max-complexity = 12
extend-select = B9
extend-ignore = E203, E501, E722, B950
(Error E722 is important, but it is identical to the activated B001.) Here is the flake8 addition for pre-commit, with the bugbear plugin:
- repo: https://github.com/pycqa/flake8
rev: "7.1.1"
hooks:
- id: flake8
additional_dependencies: [flake8-bugbear]
This will be too much at first, so you can disable or enable any test by it’s label. You can also disable a check or a list of checks inline with # noqa: X### (where you list the check label(s)). Over time, you can fix and enable more checks. A few interesting plugins:
flake8-bugbear: Fantastic checker that catches common situations that tend to create bugs. Codes:B,B9flake8-docstrings: Docstring checker.--docstring-convention=pep257is default,numpyandgooglealso allowed.flake8-spellcheck: Spelling checker. Code:SCflake8-import-order: Enforces PEP8 grouped imports (you may prefer isort). Code:Ipep8-naming: Enforces PEP8 naming rules. Code:Nflake8-print: Makes sure you don’t have print statements that sneak in. Code:T
Flake8-print details
Having something verify you don’t add a print statement by mistake is very useful. A common need for the print checker would be to add it to a single directory (src if you are following the convention recommended). You can do the next best thing by removing directories and file just for this check (T) in your flake8 config:
[flake8]
per-file-ignores =
tests/*: T
examples/*: T
YesQA
Over time, you can end up with extra “noqa” comments that are no longer needed. This is a flake8 helper that removes those comments when they are no longer required.
- repo: https://github.com/asottile/yesqa
rev: "v1.5.0"
hooks:
- id: yesqa
You need to have the same extra dependencies as flake8. In YAML, you can save the list given to yesqa and repeat it in flake8 using &flake8-dependencies and *flake8-dependencies after the colon.
isort
You can have your imports sorted automatically by isort. This will sort your imports, and is black compatible. One reason to have sorted imports is to reduce merge conflicts. Another is to clarify where imports come from - standard library imports are in a group above third party imports, which are above local imports. All this is configurable, as well. To use isort, the following pre-commit config will work:
- repo: https://github.com/PyCQA/isort
rev: "5.13.2"
hooks:
- id: isort
In order to use it, you need to add some configuration. You can add it to pyproject.toml or classic config files:
[tool.isort]
profile = "black"
PyUpgrade
Another useful tool is PyUpgrade, which monitors your codebase for “old” style syntax. Most useful to keep Python 2 outdated constructs out, it can even do some code updates for different versions of Python 3, like adding f-strings when clearly better (please always use them, they are faster) if you set --py36-plus (for example). This is a recommended addition for any project.
- repo: https://github.com/asottile/pyupgrade
rev: "v3.19.0"
hooks:
- id: pyupgrade
args: ["--py39-plus"]
If you set this to at least
--py37-plus, you can add the annotations import by adding the following line to your isort pre-commit hook configuration:args: ["-a", "from __future__ import annotations"]Also make sure isort comes before pyupgrade. Now when you run pre-commit, it will clean up your annotations to 3.7+ style, too!
Type checking
PC140 One of the most exciting advancements in Python in the last 10 years has been static type hints. Scientific Python projects vary in the degree to which they are type-hint ready. One of the challenges for providing static type hints is that it was developed in the Python 3 era and it really shines in a Python 3.7+ codebase (due to from __future__ import annotations, which turns annotations into strings and allows you to use future Python features in Python 3.7+ annotations as long as your type checker supports them). For now, it is recommended that you make an attempt to support type checking through your public API in the best way that you can (based on your supported Python versions). Stub files can be used instead for out-of-line typing. MyPy is suggested for type checking, though there are several other good options to try, as well. If you have built-in support for type checking, you need to add empty py.typed files to all packages/subpackages to indicate that you support it.
Read more about type checking on the dedicated page.
The MyPy addition for pre-commit:
- repo: https://github.com/pre-commit/mirrors-mypy
rev: "v1.13.0"
hooks:
- id: mypy
files: src
args: []
You should always specify args, as the hook’s default hides issues - it’s designed to avoid configuration, but you should add configuration. You can also add items to the virtual environment setup for MyPy by pre-commit, for example:
additional_dependencies: [attrs==23.1.0]
MY100 MyPy has a config section in pyproject.toml that looks like this:
[tool.mypy]
files = "src"
python_version = "3.9"
strict = true
enable_error_code = ["ignore-without-code", "redundant-expr", "truthy-bool"]
warn_unreachable = true
# You can disable imports or control per-module/file settings here
[[tool.mypy.overrides]]
module = [ "numpy.*", ]
ignore_missing_imports = true
There are a lot of options, and you can start with only typing global code and functions with at least one type annotation (the default) and enable more checks as you go (possibly by slowly uncommenting items in the list above). You can ignore missing imports on libraries as shown above, one section each. And you can disable MyPy on a line with # type: ignore. One strategy would be to enable check_untyped_defs first, followed by disallow_untyped_defs then disallow_incomplete_defs. You can add these per file by adding a # mypy: <option> at the top of a file. You can also pass --strict on the command line. strict = true is now allowed in config files, too MY101.
The extra strict options shown above, like warn_unreachable MY103, and ignore-without-code MY104, redundant-expr MY105, and truthy-bool MY106 can trigger too often (like on sys.platform checks) and have to be ignored occasionally, but can find some significant logic errors in your typing.
Setuptools specific checks
If you use setuptools, these checks are useful:
Setuptools-only checks
Check-Manifest (setuptools only)
Check-manifest is a fantastic, highly recommended tool that verifies you have working SDists. You can install it from PyPI. Run it on your repository and see what it says. If you want to ignore files (like test folders, example folders, docs, etc) you can add these into your pyproject.toml file:
[tool.check-manifest]
ignore = [
".travis.yml",
]
Add the following to your pre-commit config:
- repo: https://github.com/mgedmin/check-manifest
rev: "0.50"
hooks:
- id: check-manifest
If you use setuptools_scm, you might want to add:
additional_dependencies: ["setuptools_scm[toml]"]
If this is too slow:
Warning: For a complex package, this may be slow. You can optionally set stages: [manual] just below the id, and then only run this explicitly (probably in CI only). In GHA, you should enable the manual stage, which will run all checks:
- uses: pre-commit/action@v3.0.1
with:
extra_args: --show-diff-on-failure --all-files --hook-stage manual
Setup.cfg format (setuptools only)
There is a tool that keeps your setup.cfg organized, and makes sure that important parts (like Python classifiers) are in sync. This tool, setup-cfg-fmt, has native support for pre-commit:
- repo: https://github.com/asottile/setup-cfg-fmt
rev: "v2.7.0"
hooks:
- id: setup-cfg-fmt
args: [--include-version-classifiers, --max-py-version=3.12]
Make sure you list the highest version of Python you are testing with here.
Spelling
PC160 You can and should check for spelling errors in your code too. If you want to add this, you can use codespell for common spelling mistakes. Unlike most spell checkers, this has a list of mistakes it looks for, rather than a list of “valid” words. To use:
- repo: https://github.com/codespell-project/codespell
rev: "v2.3.0"
hooks:
- id: codespell
args: ["-L", "sur,nd"]
You can list allowed spellings in a comma separated string passed to -L (or --ignore-words-list - usually it is better to use long options when you are not typing things live). The example above will allow “Big Sur” and “ND”. You can instead use a comma separated list in setup.cfg or .codespellrc:
[codespell]
ignore-words-list = sur,nd
If you add the toml extra (or use Python 3.11+), you can instead put a tool.codespell section in your pyproject.toml.
You can also use a local pygrep check to eliminate common capitalization errors, such as the one below:
- repo: local
hooks:
- id: disallow-caps
name: Disallow improper capitalization
language: pygrep
entry: PyBind|Numpy|Cmake|CCache|Github|PyTest
exclude: .pre-commit-config.yaml
You can also add the -w flag to have it automatically correct errors - this is very helpful to quickly make corrections if you have a lot of them when first adding the check.
PyGrep hooks
PC170 This is a repository with a collection of pre-commit extra hooks that protect against some common, easy to detect, mistakes. You can pick and choose the hooks you want from the repo; here are some common ones for Restructured Text:
- repo: https://github.com/pre-commit/pygrep-hooks
rev: "v1.10.0"
hooks:
- id: rst-backticks
- id: rst-directive-colons
- id: rst-inline-touching-normal
If you want to add specific type ignores, see mypy_clean_slate for a tool that will add the specific ignores for you. You’ll need to remove the existing type ignores (git ls-files '*.py' | xargs sed -i '' 's/ # type: ignore//g'), copy the pre-commit output (with --show-error-codes in mypy’s args) to a file called mypy_error_report.txt, then run pipx run mypy_clean_slate -a.
Note that if you are not using Ruff’s “PGH” code, there are
python-*hooks also:- id: python-check-blanket-noqa - id: python-check-blanket-type-ignore - id: python-no-log-warn - id: python-no-eval - id: python-use-type-annotations
Clang-format (C++ only)
If you have C++ code, you should have a .clang-format file and use the following pre-commit config:
- repo: https://github.com/pre-commit/mirrors-clang-format
rev: "v19.1.2"
hooks:
- id: clang-format
types_or: [c++, c, cuda]
This will use 1-2 MB binary wheels from PyPI on all common platforms. You can generated such a file using pipx run clang-format -style=llvm -dump-config > .clang-format.
Shellcheck (shell scripts only)
If you have shell scripts, you can protect against common mistakes using shellcheck.
- repo: https://github.com/shellcheck-py/shellcheck-py
rev: "v0.10.0.1"
hooks:
- id: shellcheck
Prettier
PC180 The prettier tool can format a large number of different file types. An example of usage:
- repo: https://github.com/rbubley/mirrors-prettier
rev: "v3.3.3"
hooks:
- id: prettier
types_or: [yaml, markdown, html, css, scss, javascript, json]
Since this formats a variety of very common file types (like .html, .md, .yaml, .js, and .json), you will usually want to provide a types_or setting (shown above) with the files you are interested in auto-formatting. You can try it without the types_or first to see what it can do. Special markups in html/markdown files might clash with auto-formatting - check to verify your files are supported. This check runs using node, but pre-commit handles this for you.
If you have a .editor-config file, prettier will respect the rules in it. You can also specify a config file for prettier, or pass options to args: in pre-commit. One such option is --prose-wrap, which can be set to "never" or "always" to have prettier reflow text. You can turn off prettier for blocks with comments depending on language.
Schema validation
There are two tools, both based on JSON Schema, that you can use to validate various configuration files. The first, validate-pyproject, validates your pyproject.toml file. By default, it checks the standards-defined sections (build-system and project), along with tool.setuptools. There are also plugins for some other tools, like scikit-build-core and cibuildwheel. You can even get all SchemaStore’s plugins with the [validate-pyproject-schema-store][] plugin. Using it looks like this:
- repo: https://github.com/abravalheri/validate-pyproject
rev: "v0.22"
hooks:
- id: validate-pyproject
additional_dependencies: ["validate-pyproject-schema-store[all]"]
You can also validate various other types of files with check-jsonschema. It supports a variety of common files built-in (see the docs) like various CI configuration files. You can also write/provide your own schemas and validate using those - SchemaStore provides a few hundred different common schemas, and you can load them via URL. It work on JSON, YAML, and TOML.
- repo: https://github.com/python-jsonschema/check-jsonschema
rev: "0.29.4"
hooks:
- id: check-dependabot
- id: check-github-workflows
- id: check-readthedocs
PyLint (noisy)
PyLint is very opinionated, with a high signal-to-noise ratio. However, by limiting the default checks or by starting off a new project using them, you can get some very nice linting, including catching some problematic code that otherwise is hard to catch. PyLint is generally not a good candidate for pre-commit, since it needs to have your package installed - it is less static of check than Ruff or Flake8. Here is a suggested pyproject.toml entry to get you started:
[tool.pylint]
py-version = "3.9"
jobs = "0"
reports.output-format = "colorized"
similarities.ignore-imports = "yes"
messages_control.enable = ["useless-suppression"]
messages_control.disable = [
"design",
"fixme",
"line-too-long",
"wrong-import-position",
]
And a noxfile entry:
@nox.session
def pylint(session: nox.Session) -> None:
session.install("-e.")
session.install("pylint>=3.2")
session.run("pylint", "<your package>", *session.posargs)
And you can add this to your GitHub Actions using run: pipx run nox -s pylint -- --output-format=github. You can replace <your package> with the module name.
Jupyter notebook support
Ruff
Ruff natively supports notebooks. You no longer need to enable it, it’s on by default with Ruff 0.6+. If you want to control rules based on being notebooks, you can just match with **.ipynb like any other file.
Black
For Black, just make sure you use the id: black-jupyter hook instead of id: black; that will also include notebooks.
NBQA
You can adapt other tools to notebooks using nbQA. However, check to see if the tool natively supports notebooks first, several of them do now.
Stripping output
You also might like the following hook, which cleans Jupyter outputs:
- repo: https://github.com/kynan/nbstripout
rev: "0.7.1"
hooks:
- id: nbstripout