ruff.toml

target-version = 'py312'
line-length = 120

[lint]
preview = true

select = [
    'E',      # pycodestyle
    'F',      # pyflakes
    'N',      # pep8-naming
    'Q',      # flake8-quotes
    'RUF',    # ruff
    'UP',     # pyupgrade
    'D',      # pydocstyle
    'S',      # flake8-bandit
    'C4',     # flake8-comprehensions
    'INT',    # flake8-gettext
    'LOG',    # flake8-logging
    'G',      # flake8-logging-format
    'B',      # flake8-bugbear
    'A001',   # flake8-builtins
    'COM',    # flake8-commas
    'T10',    # flake8-debugger
    'EXE',    # flake8-executable
    'ISC',    # flake8-implicit-str-concat
    'PIE',    # flake8-pie
    'PT',     # flake8-pytest-style
    'RSE',    # flake8-raise
    'RET504', # flake8-return
    'SIM',    # flake8-simplify
    'TID',    # flake8-tidy-imports
    'PGH',    # pygrep-hooks
    'PL',     # pylint
    'TRY',    # tryceratops
    'PERF',   # perflint
    'FURB',   # refurb
    # 'ERA',  # eradicate -- too many FPs for now, see https://github.com/astral-sh/ruff/issues/6100
]
ignore = [
    'E226',    # allow omitting whitespace around arithmetic operators
    'E731',    # allow assigning lambdas (it's useful for single-line functions defined inside other functions)
    'N818',    # not all our exceptions are errors
    'RUF012',  # ultra-noisy and dicts in classvars are very common
    'RUF015',  # not always more readable, and we don't do it for huge lists
    'RUF022',  # autofix messes up our formatting instead of just sorting
    'RUF027',  # also triggers on i18n functions -> too noisy for now
    'D205',    # too many docstrings which have no summary line
    'D301',    # https://github.com/astral-sh/ruff/issues/8696
    'D1',      # we have way too many missing docstrings :(
    'D401',    # too noisy (but maybe useful to go through at some point)
    'D412',    # we do not use section, and in click docstrings those blank lines are useful
    'S101',    # we use asserts outside tests, and do not run python with `-O` (also see B011)
    'S113',    # enforcing timeouts would likely require config in some places - maybe later
    'S311',    # false positives, it does not care about the context
    'S324',    # all our md5/sha1 usages are for non-security purposes
    'S404',    # useless, triggers on *all* subprocess imports
    'S403',    # there's already a warning on using pickle, no need to have one for the import
    'S405',    # we don't use lxml in unsafe ways
    'S603',    # useless, triggers on *all* subprocess calls: https://github.com/astral-sh/ruff/issues/4045
    'S607',    # we trust the PATH to be sane
    'B011',    # we don't run python with `-O` (also see S101)
    'B904',    # possibly useful but too noisy
    'PIE807',  # `lambda: []` is much clearer for `load_default` in schemas
    'PT004',   # pretty weird + not a pytest convention: https://github.com/astral-sh/ruff/issues/8796
    'PT005',   # ^ likewise
    'PT011',   # very noisy
    'PT015',   # nice for tests but not so nice elsewhere
    'PT018',   # ^ likewise
    'SIM102',  # sometimes nested ifs are more readable
    'SIM103',  # sometimes this is more readable (especially when checking multiple conditions)
    'SIM105',  # try-except-pass is faster and people are used to it
    'SIM108',  # noisy ternary
    'SIM114',  # sometimes separate ifs are more readable (especially if they just return a bool)
    'SIM117',  # nested context managers may be more readable
    'PLC0415', # local imports are there for a reason
    'PLC2701', # some private imports are needed
    'PLR09',   # too-many-<whatever> is just noisy
    'PLR0913', # very noisy
    'PLR2004', # extremely noisy and generally annoying
    'PLR6201', # sets are faster (by a factor of 10!) but it's noisy and we're in nanoseconds territory
    'PLR6301', # extremely noisy and generally annoying
    'PLW0108', # a lambda often makes it more clear what you actually want
    'PLW1510', # we often do not care about the status code of commands
    'PLW1514', # we expect UTF8 environments everywhere
    'PLW1641', # false positives with SA comparator classes
    'PLW2901', # noisy and reassigning to the loop var is usually intentional
    'TRY002',  # super noisy, and those exceptions are pretty exceptional anyway
    'TRY003',  # super noisy and also useless w/ werkzeugs http exceptions
    'TRY300',  # kind of strange in many cases
    'TRY301',  # sometimes doing that is actually useful
    'TRY400',  # not all exceptions need exception logging
    'PERF203', # noisy, false positives, and not applicable for 3.11+
    'FURB113', # less readable
    'FURB140', # less readable and actually slower in 3.12+
]

extend-safe-fixes = [
    'RUF005',  # we typically don't deal with objects overriding `__add__` ir `__radd__`
    'C4',      # they seem pretty safe
    'UP008',   # ^ likewise
    'D200',    # ^ likewise
    'D400',    # ^ likewise
    'PT014',   # duplicate test case parametrizations are never intentional
    'RSE102',  # we do not use `raise func()` (with `func` returning the exception instance)
    'RET504',  # looks pretty safe
    'SIM110',  # ^ likewise
    'PERF102', # ^ likewise
]

[format]
quote-style = 'single'

[lint.flake8-builtins]
builtins-ignorelist = ['id', 'format', 'input', 'type', 'credits']

[lint.flake8-pytest-style]
fixture-parentheses = false
mark-parentheses = false
parametrize-names-type = 'tuple'
parametrize-values-type = 'tuple'
parametrize-values-row-type = 'tuple'

[lint.flake8-quotes]
inline-quotes = 'single'
multiline-quotes = 'single'
docstring-quotes = 'double'
avoid-escape = true

[lint.pydocstyle]
convention = 'pep257'