ruff.toml

target-version = 'py312'
line-length = 120
extend-exclude = ['docs', 'htmlcov', '*.egg-info', 'indico.conf']

[lint]
preview = true
dummy-variable-rgx = '^(_{2,}|(_+[a-zA-Z0-9_]*[a-zA-Z0-9]+?))$' # a single `_` is already used for i18n

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
    'RUF052',  # we often use underscore-prefixed variables in tests and to cache regexps in function definitions
    'UP038',   # it looks kind of weird and it slower than a tuple
    '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
    'S704',    # somewhat noisy and many FPs: see https://github.com/astral-sh/ruff/issues/14523
    'B011',    # we don't run python with `-O` (also see S101)
    'B904',    # possibly useful but too noisy
    'COM812',  # trailing commas on multiline lists are nice, but we have 2.5k violations
    'PIE807',  # `lambda: []` is much clearer for `load_default` in schemas
    '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+
    'FURB189', # we pass those objects to code that may expect the real thing
]

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.pep8-naming]
ignore-names = [
    '_process_GET',
    '_process_POST',
    '_process_PATCH',
    '_process_PUT',
    '_process_DELETE',
]
classmethod-decorators = [
    'classmethod',
    'declared_attr',
    'strict_classproperty',
    'expression',
    'comparator',
]

[lint.pydocstyle]
convention = 'pep257'

[lint.pylint]
allow-dunder-method-names = [
    '__table_args__',
    '__tablename__',
    '__clause_element__',
]

[lint.ruff]
parenthesize-tuple-in-subscript = true

[lint.per-file-ignores]
# signals use wildcard imports to expose everything in `indico.core.signals`
'indico/core/signals/event/__init__.py' = ['E402', 'F403', 'F401']
'indico/core/signals/__init__.py' = ['F403', 'F401']
# allow long lines in migrations (only do that for raw SQL please)
'indico/migrations/versions/*.py' = ['E501', 'D400']
# allow stuff that's useful in tests
'indico/*_test.py' = ['E221', 'E241', 'E272', 'N802', 'S105', 'S106', 'PLC1901']
# sphinx config has many commented-out variables with no space after `#`
'docs/source/conf.py' = ['E265']
# legacy code with camelCase
'indico/web/http_api/*.py' = ['N802', 'N803', 'N806']
'indico/modules/events/api.py' = ['N802', 'N803', 'N806']
'indico/modules/events/agreements/api.py' = ['N802']
'indico/modules/events/notes/api.py' = ['N802']
'indico/modules/rb/api.py' = ['N802']
'indico/modules/attachments/api/hooks.py' = ['N802']
'indico/modules/users/api.py' = ['N802']
# the config class has some properties that are basically constants
'indico/core/config.py' = ['N802']
# custom sqlalchemy constructs commonly use lowercase class names
'indico/core/db/sqlalchemy/custom/*.py' = ['N801']
'indico/util/decorators.py' = ['N801']
# latex generation is much more readable without curly braces used for formatting
'indico/util/mdx_latex.py' = ['UP031']
# vendored code - don't care about most stuff in there
'indico/vendor/*.py' = ['N', 'F841', 'UP', 'D', 'PIE', 'TID252', 'TRY', 'RUF059']