Best practices¶
These checks ensures that you follow the best practices.
The source for these best practices is hidden inside countless hours we have spent debugging software or reviewing it.
How do we find an inspiration for new rules? We find some ugly code during code reviews and audits. Then we forbid to use it forever. So, this error will never return to our codebase.
Summary¶
WrongMagicCommentViolation |
Restricts to use several control (or magic) comments. |
WrongDocCommentViolation |
Forbids to use empty doc comments (#: ). |
WrongModuleMetadataViolation |
Forbids to have some module level variables. |
EmptyModuleViolation |
Forbids to have empty modules. |
InitModuleHasLogicViolation |
Forbids to have logic inside __init__ module. |
WrongKeywordViolation |
Forbids to use some keywords from python . |
WrongFunctionCallViolation |
Forbids to call some built-in functions. |
FutureImportViolation |
Forbids to use __future__ imports. |
RaiseNotImplementedViolation |
Forbids to use NotImplemented error. |
BaseExceptionViolation |
Forbids to use BaseException exception. |
BooleanPositionalArgumentViolation |
Forbids to pass booleans as non-keyword parameters. |
NestedFunctionViolation |
Forbids to have nested functions. |
NestedClassViolation |
Forbids to use nested classes. |
MagicNumberViolation |
Forbids to use magic numbers in your code. |
StaticMethodViolation |
Forbids to use @staticmethod decorator. |
BadMagicMethodViolation |
Forbids to use some magic methods. |
NestedImportViolation |
Forbids to have nested imports in functions. |
RedundantLoopElseViolation |
Forbids to use else without break in a loop. |
RedundantFinallyViolation |
Forbids to use finally in try block without except block. |
ReassigningVariableToItselfViolation |
Forbids to assign variable to itself. |
YieldInsideInitViolation |
Forbids to use yield inside of __init__ method. |
ProtectedModuleViolation |
Forbids to import protected modules. |
ProtectedAttributeViolation |
Forbids to use protected attributes and methods. |
LambdaInsideLoopViolation |
Forbids to use lambda inside loops. |
UnreachableCodeViolation |
Forbids to have unreachable code. |
StatementHasNoEffectViolation |
Forbids to have statements that do nothing. |
MultipleAssignmentsViolation |
Forbids to have statements that do nothing. |
IncorrectUnpackingViolation |
Forbids to have statements that do nothing. |
DuplicateExceptionViolation |
Forbids to have the same exception class in multiple except blocks. |
Comments¶
-
class
WrongMagicCommentViolation
(node=None, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.SimpleViolation
Restricts to use several control (or magic) comments.
We do not allow to use:
# noqa
comment without specified violations# type: some_type
comments to specify a type fortyped_ast
- Reasoning:
- We cover several different use-cases in a single rule.
# noqa
comment is restricted because it can hide other violations.# type: some_type
comment is restricted because we can already use type annotations instead. - Solution:
- Use
# noqa
comments with specified error types. Use type annotations to specify types.
We still allow to use
# type: ignore
comment. Since sometimes it is totally required.Example:
# Correct: type = MyClass.get_type() # noqa: A001 coordinate: int = 10 some.int_field = 'text' # type: ignore # Wrong: type = MyClass.get_type() # noqa coordinate = 10 # type: int
New in version 0.1.0.
-
code
= 400¶
-
error_template
= 'Found wrong magic comment: {0}'¶
-
class
WrongDocCommentViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.TokenizeViolation
Forbids to use empty doc comments (
#:
).- Reasoning:
- Doc comments are used to provide a documentation. But supplying empty doc comments breaks this use-case. It is unclear why they can be used with no contents.
- Solution:
- Add some documentation to this comment. Or remove it.
Empty doc comments are not caught by the default
pycodestyle
checks.Example:
# Correct: #: List of allowed names: NAMES_WHITELIST = ['feature', 'bug', 'research'] # Wrong: #: NAMES_WHITELIST = ['feature', 'bug', 'research']
New in version 0.1.0.
-
code
= 401¶
-
error_template
= 'Found wrong doc comment'¶
Modules¶
-
class
WrongModuleMetadataViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have some module level variables.
- Reasoning:
- We discourage using module variables like
__author__
, because code should not contain any metadata. - Solution:
- Place all the metadata in
setup.py
,setup.cfg
, orpyproject.toml
. Use proper docstrings and packaging classifiers. Usepkg_resources
if you need to import this data into your app.
See
MODULE_METADATA_VARIABLES_BLACKLIST
for full list of bad names.Example:
# Wrong: __author__ = 'Nikita Sobolev' __version__ = 0.1.2
New in version 0.1.0.
-
error_template
= 'Found wrong metadata variable: {0}'¶
-
code
= 410¶
-
class
EmptyModuleViolation
(node=None, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.SimpleViolation
Forbids to have empty modules.
- Reasoning:
- Why is it even there? Do not polute your project with empty files.
- Solution:
If you have an empty module there are two ways to handle that:
- delete it
- drop some documentation in it, so you will explain why it is there
New in version 0.1.0.
-
error_template
= 'Found empty module'¶
-
code
= 411¶
-
class
InitModuleHasLogicViolation
(node=None, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.SimpleViolation
Forbids to have logic inside
__init__
module.- Reasoning:
If you have logic inside the
__init__
module it means several things:- you are keeping some outdated stuff there, you need to refactor
- you are placing this logic into the wrong file, just create another one
- you are doing some dark magic, and you should not do that
- Solution:
- Put your code in other modules.
However, we allow to have some contents inside the
__init__
module:- comments, since they are dropped before AST comes in play
- docs string, because sometimes it is required to state something
It is also fine when you have different users that use your code. And you do not want to break everything for them. In this case this rule can be configured.
- Configuration:
- This rule is configurable with
--i-control-code
. Default: True
New in version 0.1.0.
-
error_template
= 'Found `__init__.py` module with logic'¶
-
code
= 412¶
Builtins¶
-
class
WrongKeywordViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use some keywords from
python
.- Reasoning:
- We believe that some keywords are anti-patterns.
They promote bad-practices like
global
andpass
, or just not user-friendly likedel
. - Solution:
- Solutions differ from keyword to keyword.
pass
should be replaced with docstring orcontextlib.suppress
.del
should be replaced with specialized methods like.pop()
.global
andnonlocal
usages should be refactored.
Example:
# Wrong: pass del nonlocal global
New in version 0.1.0.
-
error_template
= 'Found wrong keyword: {0}'¶
-
code
= 420¶
-
class
WrongFunctionCallViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to call some built-in functions.
- Reasoning:
- Some functions are only suitable for very specific use cases, we forbid to use them in a free manner.
See
FUNCTIONS_BLACKLIST
for the full list of blacklisted functions.New in version 0.1.0.
-
error_template
= 'Found wrong function call: {0}'¶
-
code
= 421¶
-
class
FutureImportViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
__future__
imports.- Reasoning:
- Almost all
__future__
imports are legacypython2
compatibility tools that are no longer required. - Solution:
- Remove them. Drop
python2
support.
Except, there are some new ones for
python4
support. SeeFUTURE_IMPORTS_WHITELIST
for the full list of allowed future imports.Example:
# Correct: from __future__ import annotations # Wrong: from __future__ import print_function
New in version 0.1.0.
-
error_template
= 'Found future import: {0}'¶
-
code
= 422¶
-
class
RaiseNotImplementedViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
NotImplemented
error.- Reasoning:
- These two violations look so similar.
But, these violations have different use cases.
Use cases of
NotImplemented
is too limited to be generally available. - Solution:
- Use
NotImplementedError
.
Example:
# Correct: raise NotImplementedError('To be done') # Wrong: raise NotImplemented
New in version 0.1.0.
-
error_template
= 'Found raise NotImplemented'¶
-
code
= 423¶
-
class
BaseExceptionViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
BaseException
exception.- Reasoning:
- We can silence system exit and keyboard interrupt with this exception
handler. It is almost the same as raw
except:
block. - Solution:
- Handle
Exception
,KeyboardInterrupt
,GeneratorExit
, andSystemExit
separately. Do not use the plainexcept:
keyword.
Example:
# Correct: except Exception as ex: ... # Wrong: except BaseException as ex: ...
New in version 0.3.0.
See also
https://docs.python.org/3/library/exceptions.html#exception-hierarchy https://help.semmle.com/wiki/pages/viewpage.action?pageId=1608527
-
error_template
= 'Found except `BaseException`'¶
-
code
= 424¶
-
class
BooleanPositionalArgumentViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to pass booleans as non-keyword parameters.
- Reasoning:
- Passing boolean as regular positional parameters is very non-descriptive. It is almost impossible to tell, what does this parameter means. And you almost always have to look up the implementation to tell what is going on.
- Solution:
- Pass booleans as keywords only. This will help you to save extra context on what’s going on.
Example:
# Correct: UsersRepository.update(cache=True) # Wrong: UsersRepository.update(True)
New in version 0.6.0.
-
error_template
= 'Found boolean non-keyword argument'¶
-
code
= 425¶
Design¶
-
class
NestedFunctionViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have nested functions.
- Reasoning:
- Nesting functions is a bad practice. It is hard to test them, it is hard then to separate them. People tend to overuse closures, so it’s hard to manage the dataflow.
- Solution:
- Just write flat functions, there’s no need to nest them. Pass parameters as normal arguments, do not use closures. Until you need them for decorators or factories.
We also disallow to nest
lambda
andasync
functions.See
NESTED_FUNCTIONS_WHITELIST
for the whole list of whitelisted names.Example:
# Correct: def do_some(): ... def other(): ... # Wrong: def do_some(): def inner(): ...
New in version 0.1.0.
-
error_template
= 'Found nested function: {0}'¶
-
code
= 430¶
-
class
NestedClassViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use nested classes.
- Reasoning:
- Nested classes are really hard to manage. You can not even create an instance of this class in many cases. Testing them is also really hard.
- Solution:
- Just write flat classes, there’s no need nest them. If you are nesting classes inside a function for parametrization, then you will probably need to use different design (or metaclasses).
See
NESTED_CLASSES_WHITELIST
for the full list of whitelisted names.Example:
# Correct: class Some(object): ... class Other(object): ... # Wrong: class Some(object): class Inner(object): ...
New in version 0.1.0.
-
error_template
= 'Found nested class: {0}'¶
-
code
= 431¶
-
class
MagicNumberViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use magic numbers in your code.
What we call a “magic number”? Well, it is actually any number that appears in your code out of nowhere. Like
42
. Or0.32
.- Reasoning:
- It is very hard to remember what these numbers actually mean.
Why were they used? Should they ever be changed?
Or are they eternal like
3.14
? - Solution:
- Give these numbers a name! Move them to a separate variable, giving more context to the reader. And by moving things into new variables you will trigger other complexity checks.
Example:
# Correct: price_in_euro = 3.33 # could be changed later total = get_items_from_cart() * price_in_euro # Wrong: total = get_items_from_cart() * 3.33
What are numbers that we exclude from this check? Any numbers that are assigned to a variable, array, dictionary, or keyword arguments inside a function.
int
numbers that are in range[-10, 10]
and some other common numbers, that are defined inMAGIC_NUMBERS_WHITELIST
New in version 0.1.0.
-
code
= 432¶
-
error_template
= 'Found magic number: {0}'¶
-
class
StaticMethodViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
@staticmethod
decorator.- Reasoning:
- Static methods are not required to be inside the class. Because they even do not have access to the current instance.
- Solution:
- Use instance methods,
@classmethod
, or functions instead.
New in version 0.1.0.
-
error_template
= 'Found using `@staticmethod`'¶
-
code
= 433¶
-
class
BadMagicMethodViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use some magic methods.
- Reasoning:
- We forbid to use magic methods related to the forbidden language parts.
Likewise, we forbid to use
del
keyword, so we forbid to use all magic methods related to it. - Solution:
- Refactor you code to use custom methods instead. It will give more context to your app.
See
MAGIC_METHODS_BLACKLIST
for the full blacklist of the magic methods.New in version 0.1.0.
-
error_template
= 'Found using restricted magic method: {0}'¶
-
code
= 434¶
-
class
NestedImportViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have nested imports in functions.
- Reasoning:
- Usually nested imports are used to fix the import cycle. So, nested imports show that there’s an issue with you design.
- Solution:
- You don’t need nested imports, you need to refactor your code. Introduce a new module or find another way to do what you want to do. Rethink how your layered architecture should look like.
Example:
# Correct: from my_module import some_function def some(): ... # Wrong: def some(): from my_module import some_function
New in version 0.1.0.
-
error_template
= 'Found nested import'¶
-
code
= 435¶
-
class
RedundantLoopElseViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
else
withoutbreak
in a loop.We use the same logic for
for
andwhile
loops.- Reasoning:
- When there’s no
break
keyword in loop’s body it means thatelse
will always be called. This rule will reduce complexity, improve readability, and protect from possible errors. - Solution:
- Refactor your
else
case logic to be inside the loop’s body. Or right after it.
Example:
# Correct: for letter in 'abc': if letter == 'b': break else: print('"b" is not found') for letter in 'abc': print(letter) print('always called') # Wrong: for letter in 'abc': print(letter) else: print('always called')
New in version 0.3.0.
-
error_template
= 'Found `else` in a loop without `break`'¶
-
code
= 436¶
-
class
RedundantFinallyViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
finally
intry
block withoutexcept
block.- Reasoning:
- This rule will reduce complexity and improve readability.
- Solution:
- Refactor your
try
logic. Replace thetry-finally
statement with awith
statement.
Example:
# Correct: with open("filename") as f: f.write(...) # Wrong: try: f = open("filename") f.write(...) finally: f.close()
New in version 0.3.0.
-
error_template
= 'Found `finally` in `try` block without `except`'¶
-
code
= 437¶
-
class
ReassigningVariableToItselfViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to assign variable to itself.
- Reasoning:
- There is no need to do that. Generally it is an indication of some error or just dead code.
Example:
# Correct: some = some + 1 x_coord, y_coord = y_coord, x_coord # Wrong: some = some x_coord, y_coord = x_coord, y_coord
New in version 0.3.0.
-
error_template
= 'Found reassigning variable to itself'¶
-
code
= 438¶
-
class
YieldInsideInitViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
yield
inside of__init__
method.- Reasoning:
__init__
should be used to initialize new objects. It shouldn’tyield
anything, because it should returnNone
by the convention.
Example:
# Correct: class Example(object): def __init__(self): self._public_items_count = 0 # Wrong: class Example(object): def __init__(self): yield 10
New in version 0.3.0.
-
error_template
= 'Found `yield` inside `__init__` method'¶
-
code
= 439¶
-
class
ProtectedModuleViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to import protected modules.
- Reasoning:
- When importing protected modules we break a contract that authors of this module enforce. This way we are not respecting encapsulation and it may break our code at any moment.
- Solution:
- Do not import anything from protected modules. Respect the encapsulation.
Example:
# Correct: from some.public.module import FooClass # Wrong: import _compat from some._protected.module import BarClass from some.module import _protected
New in version 0.3.0.
-
error_template
= 'Found protected module import'¶
-
code
= 440¶
-
class
ProtectedAttributeViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use protected attributes and methods.
- Reasoning:
- When using protected attributes and method we break a contract that authors of this class enforce. This way we are not respecting encapsulation and it may break our code at any moment.
- Solution:
- Do not use protected attributes and methods. Respect the encapsulation.
Example:
# Correct: self._protected = 1 cls._hidden_method() some.public() # Wrong: print(some._protected) instance._hidden() self.container._internal = 10
Note, that it is possible to use protected attributes with
self
andcls
as base names. We allow this so you can create and use protected attributes and methods inside the class context. This is how protected attributes should be used.New in version 0.3.0.
-
error_template
= 'Found protected attribute usage: {0}'¶
-
code
= 441¶
-
class
LambdaInsideLoopViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to use
lambda
inside loops.- Reasoning:
- It is error-prone to use
lambda
insidefor
andwhile
loops due to the famous late-binding. - Solution:
- Use regular functions, factory functions, or
partial
functions. Save yourself from possible confusion.
Example:
# Correct: for index in range(10): some.append(partial_function(index)) # Wrong: for index in range(10): some.append(lambda index=index: index * 10)) other.append(lambda: index * 10))
New in version 0.5.0.
-
error_template
= "Found `lambda` in loop's body"¶
-
code
= 442¶
-
class
UnreachableCodeViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have unreachable code.
What is unreachable code? It is some lines of code that can not be executed by python’s interpreter.
This is probably caused by
return
orraise
statements. However, we can not cover 100% of truly unreachable code by this rule. This happens due to the dynamic nature of python. For example, detecting that1 / some_value
would sometimes raise an exception is too complicated and is out of scope of this rule.- Reasoning:
- Having dead code in your project is an indicator that you do not care about your code base at all. It dramatically reduces code quality and readability. It also demotivates team members.
- Solution:
- Delete any unreachable code your have. Or refactor it, if this happens by your mistake.
Example:
# Correct: def some_function(): print('This line is reachable, all good') return 5 # Wrong: def some_function(): return 5 print('This line is unreachable')
New in version 0.5.0.
-
error_template
= 'Found unreachable code'¶
-
code
= 443¶
-
class
StatementHasNoEffectViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have statements that do nothing.
- Reasoning:
- Statements that just access the value, or expressions used as statements indicate that your code contains dead lines. They just pollute your codebase and do nothing.
- Solution:
- Refactor your code in case it was a typo or error. Or just delete this code.
Example:
# Correct: def some_function(): price = 8 + 2 return price # Wrong: def some_function(): 8 + 2 print
New in version 0.5.0.
-
error_template
= 'Found statement that has no effect'¶
-
code
= 444¶
-
class
MultipleAssignmentsViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have statements that do nothing.
- Reasoning:
- Multiple assignments on the same line might not do what you think they do. They can also grown pretty long. And you will not notice the rising complexity of your code.
- Solution:
- Use separate lines for each assignment.
Example:
# Correct: a = 1 b = 1 # Wrong: a = b = 1
New in version 0.6.0.
-
error_template
= 'Found multiple assign targets'¶
-
code
= 445¶
-
class
IncorrectUnpackingViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have statements that do nothing.
- Reasoning:
- Having unpacking with side-effects is very dirty. You might get in serious and very hard-to-debug troubles because of this technique. So, do not use it.
- Solution:
- Use unpacking with only variables, not any other entities.
Example:
# Correct: first, second = some() # Wrong: first, some_dict['alias'] = some()
New in version 0.6.0.
-
error_template
= 'Found incorrect unpacking target'¶
-
code
= 446¶
-
class
DuplicateExceptionViolation
(node, text=None)[source]¶ Bases:
wemake_python_styleguide.violations.base.ASTViolation
Forbids to have the same exception class in multiple
except
blocks.- Reasoning:
- Having the same exception name in different blocks means that something is not right: since only one branch will work. Other one will always be ignored. So, that is clearly an error.
- Solution:
- Use unique exception handling rules.
Example:
# Correct: try: ... except ValueError: ... # Wrong: try: ... except ValueError: ... except ValueError: ...
New in version 0.6.0.
-
error_template
= 'Found duplicate exception: {0}'¶
-
code
= 447¶