# Licensed under the GPL: https://www.gnu.org/licenses/old-licenses/gpl-2.0.html # For details: https://github.com/PyCQA/pylint/blob/master/COPYING import collections import copy import itertools import tokenize from functools import reduce from typing import List import astroid from astroid import decorators from pylint import checkers, interfaces from pylint import utils as lint_utils from pylint.checkers import utils from pylint.checkers.utils import node_frame_class KNOWN_INFINITE_ITERATORS = {"itertools.count"} BUILTIN_EXIT_FUNCS = frozenset(("quit", "exit")) def _if_statement_is_always_returning(if_node, returning_node_class) -> bool: for node in if_node.body: if isinstance(node, returning_node_class): return True return False def _is_trailing_comma(tokens: List[tokenize.TokenInfo], index: int) -> bool: """Check if the given token is a trailing comma :param tokens: Sequence of modules tokens :type tokens: list[tokenize.TokenInfo] :param int index: Index of token under check in tokens :returns: True if the token is a comma which trails an expression :rtype: bool """ token = tokens[index] if token.exact_type != tokenize.COMMA: return False # Must have remaining tokens on the same line such as NEWLINE left_tokens = itertools.islice(tokens, index + 1, None) def same_start_token( other_token: tokenize.TokenInfo, _token: tokenize.TokenInfo = token ) -> bool: return other_token.start[0] == _token.start[0] same_line_remaining_tokens = list( itertools.takewhile(same_start_token, left_tokens) ) # Note: If the newline is tokenize.NEWLINE and not tokenize.NL # then the newline denotes the end of expression is_last_element = all( other_token.type in (tokenize.NEWLINE, tokenize.COMMENT) for other_token in same_line_remaining_tokens ) if not same_line_remaining_tokens or not is_last_element: return False def get_curline_index_start(): """Get the index denoting the start of the current line""" for subindex, token in enumerate(reversed(tokens[:index])): # See Lib/tokenize.py and Lib/token.py in cpython for more info if token.type == tokenize.NEWLINE: return index - subindex return 0 curline_start = get_curline_index_start() expected_tokens = {"return", "yield"} for prevtoken in tokens[curline_start:index]: if "=" in prevtoken.string or prevtoken.string in expected_tokens: return True return False class RefactoringChecker(checkers.BaseTokenChecker): """Looks for code which can be refactored This checker also mixes the astroid and the token approaches in order to create knowledge about whether an "else if" node is a true "else if" node, or an "elif" node. """ __implements__ = (interfaces.ITokenChecker, interfaces.IAstroidChecker) name = "refactoring" msgs = { "R1701": ( "Consider merging these isinstance calls to isinstance(%s, (%s))", "consider-merging-isinstance", "Used when multiple consecutive isinstance calls can be merged into one.", ), "R1706": ( "Consider using ternary (%s)", "consider-using-ternary", "Used when one of known pre-python 2.5 ternary syntax is used.", ), "R1709": ( "Boolean expression may be simplified to %s", "simplify-boolean-expression", "Emitted when redundant pre-python 2.5 ternary syntax is used.", ), "R1726": ( "Boolean condition '%s' may be simplified to '%s'", "simplifiable-condition", "Emitted when a boolean condition is able to be simplified.", ), "R1727": ( "Boolean condition '%s' will always evaluate to '%s'", "condition-evals-to-constant", "Emitted when a boolean condition can be simplified to a constant value.", ), "R1702": ( "Too many nested blocks (%s/%s)", "too-many-nested-blocks", "Used when a function or a method has too many nested " "blocks. This makes the code less understandable and " "maintainable.", {"old_names": [("R0101", "old-too-many-nested-blocks")]}, ), "R1703": ( "The if statement can be replaced with %s", "simplifiable-if-statement", "Used when an if statement can be replaced with 'bool(test)'. ", {"old_names": [("R0102", "old-simplifiable-if-statement")]}, ), "R1704": ( "Redefining argument with the local name %r", "redefined-argument-from-local", "Used when a local name is redefining an argument, which might " "suggest a potential error. This is taken in account only for " "a handful of name binding operations, such as for iteration, " "with statement assignment and exception handler assignment.", ), "R1705": ( 'Unnecessary "%s" after "return"', "no-else-return", "Used in order to highlight an unnecessary block of " "code following an if containing a return statement. " "As such, it will warn when it encounters an else " "following a chain of ifs, all of them containing a " "return statement.", ), "R1707": ( "Disallow trailing comma tuple", "trailing-comma-tuple", "In Python, a tuple is actually created by the comma symbol, " "not by the parentheses. Unfortunately, one can actually create a " "tuple by misplacing a trailing comma, which can lead to potential " "weird bugs in your code. You should always use parentheses " "explicitly for creating a tuple.", ), "R1708": ( "Do not raise StopIteration in generator, use return statement instead", "stop-iteration-return", "According to PEP479, the raise of StopIteration to end the loop of " "a generator may lead to hard to find bugs. This PEP specify that " "raise StopIteration has to be replaced by a simple return statement", ), "R1710": ( "Either all return statements in a function should return an expression, " "or none of them should.", "inconsistent-return-statements", "According to PEP8, if any return statement returns an expression, " "any return statements where no value is returned should explicitly " "state this as return None, and an explicit return statement " "should be present at the end of the function (if reachable)", ), "R1711": ( "Useless return at end of function or method", "useless-return", 'Emitted when a single "return" or "return None" statement is found ' "at the end of function or method definition. This statement can safely be " "removed because Python will implicitly return None", ), "R1712": ( "Consider using tuple unpacking for swapping variables", "consider-swap-variables", "You do not have to use a temporary variable in order to " 'swap variables. Using "tuple unpacking" to directly swap ' "variables makes the intention more clear.", ), "R1713": ( "Consider using str.join(sequence) for concatenating " "strings from an iterable", "consider-using-join", "Using str.join(sequence) is faster, uses less memory " "and increases readability compared to for-loop iteration.", ), "R1714": ( 'Consider merging these comparisons with "in" to %r', "consider-using-in", "To check if a variable is equal to one of many values," 'combine the values into a tuple and check if the variable is contained "in" it ' "instead of checking for equality against each of the values." "This is faster and less verbose.", ), "R1715": ( "Consider using dict.get for getting values from a dict " "if a key is present or a default if not", "consider-using-get", "Using the builtin dict.get for getting a value from a dictionary " "if a key is present or a default if not, is simpler and considered " "more idiomatic, although sometimes a bit slower", ), "R1716": ( "Simplify chained comparison between the operands", "chained-comparison", "This message is emitted when pylint encounters boolean operation like" '"a < b and b < c", suggesting instead to refactor it to "a < b < c"', ), "R1717": ( "Consider using a dictionary comprehension", "consider-using-dict-comprehension", "Emitted when we detect the creation of a dictionary " "using the dict() callable and a transient list. " "Although there is nothing syntactically wrong with this code, " "it is hard to read and can be simplified to a dict comprehension." "Also it is faster since you don't need to create another " "transient list", ), "R1718": ( "Consider using a set comprehension", "consider-using-set-comprehension", "Although there is nothing syntactically wrong with this code, " "it is hard to read and can be simplified to a set comprehension." "Also it is faster since you don't need to create another " "transient list", ), "R1719": ( "The if expression can be replaced with %s", "simplifiable-if-expression", "Used when an if expression can be replaced with 'bool(test)'. ", ), "R1720": ( 'Unnecessary "%s" after "raise"', "no-else-raise", "Used in order to highlight an unnecessary block of " "code following an if containing a raise statement. " "As such, it will warn when it encounters an else " "following a chain of ifs, all of them containing a " "raise statement.", ), "R1721": ( "Unnecessary use of a comprehension", "unnecessary-comprehension", "Instead of using an identity comprehension, " "consider using the list, dict or set constructor. " "It is faster and simpler.", ), "R1722": ( "Consider using sys.exit()", "consider-using-sys-exit", "Instead of using exit() or quit(), consider using the sys.exit().", ), "R1723": ( 'Unnecessary "%s" after "break"', "no-else-break", "Used in order to highlight an unnecessary block of " "code following an if containing a break statement. " "As such, it will warn when it encounters an else " "following a chain of ifs, all of them containing a " "break statement.", ), "R1724": ( 'Unnecessary "%s" after "continue"', "no-else-continue", "Used in order to highlight an unnecessary block of " "code following an if containing a continue statement. " "As such, it will warn when it encounters an else " "following a chain of ifs, all of them containing a " "continue statement.", ), "R1725": ( "Consider using Python 3 style super() without arguments", "super-with-arguments", "Emitted when calling the super() builtin with the current class " "and instance. On Python 3 these arguments are the default and they can be omitted.", ), "R1728": ( "Consider using a generator instead '%s(%s)'", "consider-using-generator", "If your container can be large using " "a generator will bring better performance.", ), "R1729": ( "Use a generator instead '%s(%s)'", "use-a-generator", "Comprehension inside of 'any' or 'all' is unnecessary. " "A generator would be sufficient and faster.", ), } options = ( ( "max-nested-blocks", { "default": 5, "type": "int", "metavar": "", "help": "Maximum number of nested blocks for function / method body", }, ), ( "never-returning-functions", { "default": ("sys.exit",), "type": "csv", "help": "Complete name of functions that never returns. When checking " "for inconsistent-return-statements if a never returning function is " "called then it will be considered as an explicit return statement " "and no message will be printed.", }, ), ) priority = 0 def __init__(self, linter=None): checkers.BaseTokenChecker.__init__(self, linter) self._return_nodes = {} self._init() self._never_returning_functions = None def _init(self): self._nested_blocks = [] self._elifs = [] self._nested_blocks_msg = None self._reported_swap_nodes = set() self._can_simplify_bool_op = False def open(self): # do this in open since config not fully initialized in __init__ self._never_returning_functions = set(self.config.never_returning_functions) @decorators.cachedproperty def _dummy_rgx(self): return lint_utils.get_global_option(self, "dummy-variables-rgx", default=None) @staticmethod def _is_bool_const(node): return isinstance(node.value, astroid.Const) and isinstance( node.value.value, bool ) def _is_actual_elif(self, node): """Check if the given node is an actual elif This is a problem we're having with the builtin ast module, which splits `elif` branches into a separate if statement. Unfortunately we need to know the exact type in certain cases. """ if isinstance(node.parent, astroid.If): orelse = node.parent.orelse # current if node must directly follow an "else" if orelse and orelse == [node]: if (node.lineno, node.col_offset) in self._elifs: return True return False def _check_simplifiable_if(self, node): """Check if the given if node can be simplified. The if statement can be reduced to a boolean expression in some cases. For instance, if there are two branches and both of them return a boolean value that depends on the result of the statement's test, then this can be reduced to `bool(test)` without losing any functionality. """ if self._is_actual_elif(node): # Not interested in if statements with multiple branches. return if len(node.orelse) != 1 or len(node.body) != 1: return # Check if both branches can be reduced. first_branch = node.body[0] else_branch = node.orelse[0] if isinstance(first_branch, astroid.Return): if not isinstance(else_branch, astroid.Return): return first_branch_is_bool = self._is_bool_const(first_branch) else_branch_is_bool = self._is_bool_const(else_branch) reduced_to = "'return bool(test)'" elif isinstance(first_branch, astroid.Assign): if not isinstance(else_branch, astroid.Assign): return # Check if we assign to the same value first_branch_targets = [ target.name for target in first_branch.targets if isinstance(target, astroid.AssignName) ] else_branch_targets = [ target.name for target in else_branch.targets if isinstance(target, astroid.AssignName) ] if not first_branch_targets or not else_branch_targets: return if sorted(first_branch_targets) != sorted(else_branch_targets): return first_branch_is_bool = self._is_bool_const(first_branch) else_branch_is_bool = self._is_bool_const(else_branch) reduced_to = "'var = bool(test)'" else: return if not first_branch_is_bool or not else_branch_is_bool: return if not first_branch.value.value: # This is a case that can't be easily simplified and # if it can be simplified, it will usually result in a # code that's harder to understand and comprehend. # Let's take for instance `arg and arg <= 3`. This could theoretically be # reduced to `not arg or arg > 3`, but the net result is that now the # condition is harder to understand, because it requires understanding of # an extra clause: # * first, there is the negation of truthness with `not arg` # * the second clause is `arg > 3`, which occurs when arg has a # a truth value, but it implies that `arg > 3` is equivalent # with `arg and arg > 3`, which means that the user must # think about this assumption when evaluating `arg > 3`. # The original form is easier to grasp. return self.add_message("simplifiable-if-statement", node=node, args=(reduced_to,)) def process_tokens(self, tokens): # Process tokens and look for 'if' or 'elif' for index, token in enumerate(tokens): token_string = token[1] if token_string == "elif": # AST exists by the time process_tokens is called, so # it's safe to assume tokens[index+1] # exists. tokens[index+1][2] is the elif's position as # reported by CPython and PyPy, # tokens[index][2] is the actual position and also is # reported by IronPython. self._elifs.extend([tokens[index][2], tokens[index + 1][2]]) elif _is_trailing_comma(tokens, index): if self.linter.is_message_enabled("trailing-comma-tuple"): self.add_message("trailing-comma-tuple", line=token.start[0]) def leave_module(self, _): self._init() @utils.check_messages("too-many-nested-blocks") def visit_tryexcept(self, node): self._check_nested_blocks(node) visit_tryfinally = visit_tryexcept visit_while = visit_tryexcept def _check_redefined_argument_from_local(self, name_node): if self._dummy_rgx and self._dummy_rgx.match(name_node.name): return if not name_node.lineno: # Unknown position, maybe it is a manually built AST? return scope = name_node.scope() if not isinstance(scope, astroid.FunctionDef): return for defined_argument in scope.args.nodes_of_class( astroid.AssignName, skip_klass=(astroid.Lambda,) ): if defined_argument.name == name_node.name: self.add_message( "redefined-argument-from-local", node=name_node, args=(name_node.name,), ) @utils.check_messages("redefined-argument-from-local", "too-many-nested-blocks") def visit_for(self, node): self._check_nested_blocks(node) for name in node.target.nodes_of_class(astroid.AssignName): self._check_redefined_argument_from_local(name) @utils.check_messages("redefined-argument-from-local") def visit_excepthandler(self, node): if node.name and isinstance(node.name, astroid.AssignName): self._check_redefined_argument_from_local(node.name) @utils.check_messages("redefined-argument-from-local") def visit_with(self, node): for _, names in node.items: if not names: continue for name in names.nodes_of_class(astroid.AssignName): self._check_redefined_argument_from_local(name) def _check_superfluous_else(self, node, msg_id, returning_node_class): if not node.orelse: # Not interested in if statements without else. return if self._is_actual_elif(node): # Not interested in elif nodes; only if return if _if_statement_is_always_returning(node, returning_node_class): orelse = node.orelse[0] followed_by_elif = (orelse.lineno, orelse.col_offset) in self._elifs self.add_message( msg_id, node=node, args="elif" if followed_by_elif else "else" ) def _check_superfluous_else_return(self, node): return self._check_superfluous_else( node, msg_id="no-else-return", returning_node_class=astroid.Return ) def _check_superfluous_else_raise(self, node): return self._check_superfluous_else( node, msg_id="no-else-raise", returning_node_class=astroid.Raise ) def _check_superfluous_else_break(self, node): return self._check_superfluous_else( node, msg_id="no-else-break", returning_node_class=astroid.Break ) def _check_superfluous_else_continue(self, node): return self._check_superfluous_else( node, msg_id="no-else-continue", returning_node_class=astroid.Continue ) @staticmethod def _type_and_name_are_equal(node_a, node_b): for _type in [astroid.Name, astroid.AssignName]: if all(isinstance(_node, _type) for _node in [node_a, node_b]): return node_a.name == node_b.name if all(isinstance(_node, astroid.Const) for _node in [node_a, node_b]): return node_a.value == node_b.value return False def _is_dict_get_block(self, node): # "if " if not isinstance(node.test, astroid.Compare): return False # Does not have a single statement in the guard's body if len(node.body) != 1: return False # Look for a single variable assignment on the LHS and a subscript on RHS stmt = node.body[0] if not ( isinstance(stmt, astroid.Assign) and len(node.body[0].targets) == 1 and isinstance(node.body[0].targets[0], astroid.AssignName) and isinstance(stmt.value, astroid.Subscript) ): return False # The subscript's slice needs to be the same as the test variable. # Python 3.9 we no longer have the `Index` node. slice_value = stmt.value.slice if isinstance(slice_value, astroid.Index): slice_value = slice_value.value if not ( self._type_and_name_are_equal(stmt.value.value, node.test.ops[0][1]) and self._type_and_name_are_equal(slice_value, node.test.left) ): return False # The object needs to be a dictionary instance return isinstance(utils.safe_infer(node.test.ops[0][1]), astroid.Dict) def _check_consider_get(self, node): if_block_ok = self._is_dict_get_block(node) if if_block_ok and not node.orelse: self.add_message("consider-using-get", node=node) elif ( if_block_ok and len(node.orelse) == 1 and isinstance(node.orelse[0], astroid.Assign) and self._type_and_name_are_equal( node.orelse[0].targets[0], node.body[0].targets[0] ) and len(node.orelse[0].targets) == 1 ): self.add_message("consider-using-get", node=node) @utils.check_messages( "too-many-nested-blocks", "simplifiable-if-statement", "no-else-return", "no-else-raise", "no-else-break", "no-else-continue", "consider-using-get", ) def visit_if(self, node): self._check_simplifiable_if(node) self._check_nested_blocks(node) self._check_superfluous_else_return(node) self._check_superfluous_else_raise(node) self._check_superfluous_else_break(node) self._check_superfluous_else_continue(node) self._check_consider_get(node) @utils.check_messages("simplifiable-if-expression") def visit_ifexp(self, node): self._check_simplifiable_ifexp(node) def _check_simplifiable_ifexp(self, node): if not isinstance(node.body, astroid.Const) or not isinstance( node.orelse, astroid.Const ): return if not isinstance(node.body.value, bool) or not isinstance( node.orelse.value, bool ): return if isinstance(node.test, astroid.Compare): test_reduced_to = "test" else: test_reduced_to = "bool(test)" if (node.body.value, node.orelse.value) == (True, False): reduced_to = f"'{test_reduced_to}'" elif (node.body.value, node.orelse.value) == (False, True): reduced_to = "'not test'" else: return self.add_message("simplifiable-if-expression", node=node, args=(reduced_to,)) @utils.check_messages( "too-many-nested-blocks", "inconsistent-return-statements", "useless-return" ) def leave_functiondef(self, node): # check left-over nested blocks stack self._emit_nested_blocks_message_if_needed(self._nested_blocks) # new scope = reinitialize the stack of nested blocks self._nested_blocks = [] # check consistent return statements self._check_consistent_returns(node) # check for single return or return None at the end self._check_return_at_the_end(node) self._return_nodes[node.name] = [] @utils.check_messages("stop-iteration-return") def visit_raise(self, node): self._check_stop_iteration_inside_generator(node) def _check_stop_iteration_inside_generator(self, node): """Check if an exception of type StopIteration is raised inside a generator""" frame = node.frame() if not isinstance(frame, astroid.FunctionDef) or not frame.is_generator(): return if utils.node_ignores_exception(node, StopIteration): return if not node.exc: return exc = utils.safe_infer(node.exc) if not exc or not isinstance(exc, (astroid.Instance, astroid.ClassDef)): return if self._check_exception_inherit_from_stopiteration(exc): self.add_message("stop-iteration-return", node=node) @staticmethod def _check_exception_inherit_from_stopiteration(exc): """Return True if the exception node in argument inherit from StopIteration""" stopiteration_qname = f"{utils.EXCEPTIONS_MODULE}.StopIteration" return any(_class.qname() == stopiteration_qname for _class in exc.mro()) def _check_consider_using_comprehension_constructor(self, node): if ( isinstance(node.func, astroid.Name) and node.args and isinstance(node.args[0], astroid.ListComp) ): if node.func.name == "dict" and not isinstance( node.args[0].elt, astroid.Call ): message_name = "consider-using-dict-comprehension" self.add_message(message_name, node=node) elif node.func.name == "set": message_name = "consider-using-set-comprehension" self.add_message(message_name, node=node) def _check_consider_using_generator(self, node): # 'any' and 'all' definitely should use generator, while 'list' and 'tuple' need to be considered first # See https://github.com/PyCQA/pylint/pull/3309#discussion_r576683109 checked_call = ["any", "all", "list", "tuple"] if ( isinstance(node, astroid.Call) and node.func and isinstance(node.func, astroid.Name) and node.func.name in checked_call ): # functions in checked_calls take exactly one argument # check whether the argument is list comprehension if len(node.args) == 1 and isinstance(node.args[0], astroid.ListComp): # remove square brackets '[]' inside_comp = node.args[0].as_string()[1:-1] call_name = node.func.name if call_name in ["any", "all"]: self.add_message( "use-a-generator", node=node, args=(call_name, inside_comp), ) else: self.add_message( "consider-using-generator", node=node, args=(call_name, inside_comp), ) @utils.check_messages( "stop-iteration-return", "consider-using-dict-comprehension", "consider-using-set-comprehension", "consider-using-sys-exit", "super-with-arguments", "consider-using-generator", ) def visit_call(self, node): self._check_raising_stopiteration_in_generator_next_call(node) self._check_consider_using_comprehension_constructor(node) self._check_quit_exit_call(node) self._check_super_with_arguments(node) self._check_consider_using_generator(node) @staticmethod def _has_exit_in_scope(scope): exit_func = scope.locals.get("exit") return bool( exit_func and isinstance(exit_func[0], (astroid.ImportFrom, astroid.Import)) ) def _check_quit_exit_call(self, node): if isinstance(node.func, astroid.Name) and node.func.name in BUILTIN_EXIT_FUNCS: # If we have `exit` imported from `sys` in the current or global scope, exempt this instance. local_scope = node.scope() if self._has_exit_in_scope(local_scope) or self._has_exit_in_scope( node.root() ): return self.add_message("consider-using-sys-exit", node=node) def _check_super_with_arguments(self, node): if not isinstance(node.func, astroid.Name) or node.func.name != "super": return # pylint: disable=too-many-boolean-expressions if ( len(node.args) != 2 or not isinstance(node.args[1], astroid.Name) or node.args[1].name != "self" or not isinstance(node.args[0], astroid.Name) or not isinstance(node.args[1], astroid.Name) or node_frame_class(node) is None or node.args[0].name != node_frame_class(node).name ): return self.add_message("super-with-arguments", node=node) def _check_raising_stopiteration_in_generator_next_call(self, node): """Check if a StopIteration exception is raised by the call to next function If the next value has a default value, then do not add message. :param node: Check to see if this Call node is a next function :type node: :class:`astroid.node_classes.Call` """ def _looks_like_infinite_iterator(param): inferred = utils.safe_infer(param) if inferred: return inferred.qname() in KNOWN_INFINITE_ITERATORS return False if isinstance(node.func, astroid.Attribute): # A next() method, which is now what we want. return inferred = utils.safe_infer(node.func) if getattr(inferred, "name", "") == "next": frame = node.frame() # The next builtin can only have up to two # positional arguments and no keyword arguments has_sentinel_value = len(node.args) > 1 if ( isinstance(frame, astroid.FunctionDef) and frame.is_generator() and not has_sentinel_value and not utils.node_ignores_exception(node, StopIteration) and not _looks_like_infinite_iterator(node.args[0]) ): self.add_message("stop-iteration-return", node=node) def _check_nested_blocks(self, node): """Update and check the number of nested blocks""" # only check block levels inside functions or methods if not isinstance(node.scope(), astroid.FunctionDef): return # messages are triggered on leaving the nested block. Here we save the # stack in case the current node isn't nested in the previous one nested_blocks = self._nested_blocks[:] if node.parent == node.scope(): self._nested_blocks = [node] else: # go through ancestors from the most nested to the less for ancestor_node in reversed(self._nested_blocks): if ancestor_node == node.parent: break self._nested_blocks.pop() # if the node is an elif, this should not be another nesting level if isinstance(node, astroid.If) and self._is_actual_elif(node): if self._nested_blocks: self._nested_blocks.pop() self._nested_blocks.append(node) # send message only once per group of nested blocks if len(nested_blocks) > len(self._nested_blocks): self._emit_nested_blocks_message_if_needed(nested_blocks) def _emit_nested_blocks_message_if_needed(self, nested_blocks): if len(nested_blocks) > self.config.max_nested_blocks: self.add_message( "too-many-nested-blocks", node=nested_blocks[0], args=(len(nested_blocks), self.config.max_nested_blocks), ) @staticmethod def _duplicated_isinstance_types(node): """Get the duplicated types from the underlying isinstance calls. :param astroid.BoolOp node: Node which should contain a bunch of isinstance calls. :returns: Dictionary of the comparison objects from the isinstance calls, to duplicate values from consecutive calls. :rtype: dict """ duplicated_objects = set() all_types = collections.defaultdict(set) for call in node.values: if not isinstance(call, astroid.Call) or len(call.args) != 2: continue inferred = utils.safe_infer(call.func) if not inferred or not utils.is_builtin_object(inferred): continue if inferred.name != "isinstance": continue isinstance_object = call.args[0].as_string() isinstance_types = call.args[1] if isinstance_object in all_types: duplicated_objects.add(isinstance_object) if isinstance(isinstance_types, astroid.Tuple): elems = [ class_type.as_string() for class_type in isinstance_types.itered() ] else: elems = [isinstance_types.as_string()] all_types[isinstance_object].update(elems) # Remove all keys which not duplicated return { key: value for key, value in all_types.items() if key in duplicated_objects } def _check_consider_merging_isinstance(self, node): """Check isinstance calls which can be merged together.""" if node.op != "or": return first_args = self._duplicated_isinstance_types(node) for duplicated_name, class_names in first_args.items(): names = sorted(name for name in class_names) self.add_message( "consider-merging-isinstance", node=node, args=(duplicated_name, ", ".join(names)), ) def _check_consider_using_in(self, node): allowed_ops = {"or": "==", "and": "!="} if node.op not in allowed_ops or len(node.values) < 2: return for value in node.values: if ( not isinstance(value, astroid.Compare) or len(value.ops) != 1 or value.ops[0][0] not in allowed_ops[node.op] ): return for comparable in value.left, value.ops[0][1]: if isinstance(comparable, astroid.Call): return # Gather variables and values from comparisons variables, values = [], [] for value in node.values: variable_set = set() for comparable in value.left, value.ops[0][1]: if isinstance(comparable, astroid.Name): variable_set.add(comparable.as_string()) values.append(comparable.as_string()) variables.append(variable_set) # Look for (common-)variables that occur in all comparisons common_variables = reduce(lambda a, b: a.intersection(b), variables) if not common_variables: return # Gather information for the suggestion common_variable = sorted(list(common_variables))[0] comprehension = "in" if node.op == "or" else "not in" values = list(collections.OrderedDict.fromkeys(values)) values.remove(common_variable) values_string = ", ".join(values) if len(values) != 1 else values[0] + "," suggestion = "%s %s (%s)" % (common_variable, comprehension, values_string) self.add_message("consider-using-in", node=node, args=(suggestion,)) def _check_chained_comparison(self, node): """Check if there is any chained comparison in the expression. Add a refactoring message if a boolOp contains comparison like a < b and b < c, which can be chained as a < b < c. Care is taken to avoid simplifying a < b < c and b < d. """ if node.op != "and" or len(node.values) < 2: return def _find_lower_upper_bounds(comparison_node, uses): left_operand = comparison_node.left for operator, right_operand in comparison_node.ops: for operand in (left_operand, right_operand): value = None if isinstance(operand, astroid.Name): value = operand.name elif isinstance(operand, astroid.Const): value = operand.value if value is None: continue if operator in ("<", "<="): if operand is left_operand: uses[value]["lower_bound"].add(comparison_node) elif operand is right_operand: uses[value]["upper_bound"].add(comparison_node) elif operator in (">", ">="): if operand is left_operand: uses[value]["upper_bound"].add(comparison_node) elif operand is right_operand: uses[value]["lower_bound"].add(comparison_node) left_operand = right_operand uses = collections.defaultdict( lambda: {"lower_bound": set(), "upper_bound": set()} ) for comparison_node in node.values: if isinstance(comparison_node, astroid.Compare): _find_lower_upper_bounds(comparison_node, uses) for _, bounds in uses.items(): num_shared = len(bounds["lower_bound"].intersection(bounds["upper_bound"])) num_lower_bounds = len(bounds["lower_bound"]) num_upper_bounds = len(bounds["upper_bound"]) if num_shared < num_lower_bounds and num_shared < num_upper_bounds: self.add_message("chained-comparison", node=node) break @staticmethod def _apply_boolean_simplification_rules(operator, values): """Removes irrelevant values or returns shortcircuiting values This function applies the following two rules: 1) an OR expression with True in it will always be true, and the reverse for AND 2) False values in OR expressions are only relevant if all values are false, and the reverse for AND""" simplified_values = [] for subnode in values: inferred_bool = None if not next(subnode.nodes_of_class(astroid.Name), False): inferred = utils.safe_infer(subnode) if inferred: inferred_bool = inferred.bool_value() if not isinstance(inferred_bool, bool): simplified_values.append(subnode) elif (operator == "or") == inferred_bool: return [subnode] return simplified_values or [astroid.Const(operator == "and")] def _simplify_boolean_operation(self, bool_op): """Attempts to simplify a boolean operation Recursively applies simplification on the operator terms, and keeps track of whether reductions have been made.""" children = list(bool_op.get_children()) intermediate = [ self._simplify_boolean_operation(child) if isinstance(child, astroid.BoolOp) else child for child in children ] result = self._apply_boolean_simplification_rules(bool_op.op, intermediate) if len(result) < len(children): self._can_simplify_bool_op = True if len(result) == 1: return result[0] simplified_bool_op = copy.copy(bool_op) simplified_bool_op.postinit(result) return simplified_bool_op def _check_simplifiable_condition(self, node): """Check if a boolean condition can be simplified. Variables will not be simplified, even in the value can be inferred, and expressions like '3 + 4' will remain expanded.""" if not utils.is_test_condition(node): return self._can_simplify_bool_op = False simplified_expr = self._simplify_boolean_operation(node) if not self._can_simplify_bool_op: return if not next(simplified_expr.nodes_of_class(astroid.Name), False): self.add_message( "condition-evals-to-constant", node=node, args=(node.as_string(), simplified_expr.as_string()), ) else: self.add_message( "simplifiable-condition", node=node, args=(node.as_string(), simplified_expr.as_string()), ) @utils.check_messages( "consider-merging-isinstance", "consider-using-in", "chained-comparison", "simplifiable-condition", "condition-evals-to-constant", ) def visit_boolop(self, node): self._check_consider_merging_isinstance(node) self._check_consider_using_in(node) self._check_chained_comparison(node) self._check_simplifiable_condition(node) @staticmethod def _is_simple_assignment(node): return ( isinstance(node, astroid.Assign) and len(node.targets) == 1 and isinstance(node.targets[0], astroid.node_classes.AssignName) and isinstance(node.value, astroid.node_classes.Name) ) def _check_swap_variables(self, node): if not node.next_sibling() or not node.next_sibling().next_sibling(): return assignments = [node, node.next_sibling(), node.next_sibling().next_sibling()] if not all(self._is_simple_assignment(node) for node in assignments): return if any(node in self._reported_swap_nodes for node in assignments): return left = [node.targets[0].name for node in assignments] right = [node.value.name for node in assignments] if left[0] == right[-1] and left[1:] == right[:-1]: self._reported_swap_nodes.update(assignments) message = "consider-swap-variables" self.add_message(message, node=node) @utils.check_messages( "simplify-boolean-expression", "consider-using-ternary", "consider-swap-variables", ) def visit_assign(self, node): self._check_swap_variables(node) if self._is_and_or_ternary(node.value): cond, truth_value, false_value = self._and_or_ternary_arguments(node.value) else: return if all( isinstance(value, astroid.Compare) for value in (truth_value, false_value) ): return inferred_truth_value = utils.safe_infer(truth_value) if inferred_truth_value in (None, astroid.Uninferable): truth_boolean_value = True else: truth_boolean_value = truth_value.bool_value() if truth_boolean_value is False: message = "simplify-boolean-expression" suggestion = false_value.as_string() else: message = "consider-using-ternary" suggestion = "{truth} if {cond} else {false}".format( truth=truth_value.as_string(), cond=cond.as_string(), false=false_value.as_string(), ) self.add_message(message, node=node, args=(suggestion,)) visit_return = visit_assign def _check_consider_using_join(self, aug_assign): """ We start with the augmented assignment and work our way upwards. Names of variables for nodes if match successful: result = '' # assign for number in ['1', '2', '3'] # for_loop result += number # aug_assign """ for_loop = aug_assign.parent if not isinstance(for_loop, astroid.For) or len(for_loop.body) > 1: return assign = for_loop.previous_sibling() if not isinstance(assign, astroid.Assign): return result_assign_names = { target.name for target in assign.targets if isinstance(target, astroid.AssignName) } is_concat_loop = ( aug_assign.op == "+=" and isinstance(aug_assign.target, astroid.AssignName) and len(for_loop.body) == 1 and aug_assign.target.name in result_assign_names and isinstance(assign.value, astroid.Const) and isinstance(assign.value.value, str) and isinstance(aug_assign.value, astroid.Name) and aug_assign.value.name == for_loop.target.name ) if is_concat_loop: self.add_message("consider-using-join", node=aug_assign) @utils.check_messages("consider-using-join") def visit_augassign(self, node): self._check_consider_using_join(node) @utils.check_messages("unnecessary-comprehension") def visit_comprehension(self, node): self._check_unnecessary_comprehension(node) def _check_unnecessary_comprehension(self, node): if ( isinstance(node.parent, astroid.GeneratorExp) or len(node.ifs) != 0 or len(node.parent.generators) != 1 or node.is_async ): return if ( isinstance(node.parent, astroid.DictComp) and isinstance(node.parent.key, astroid.Name) and isinstance(node.parent.value, astroid.Name) and isinstance(node.target, astroid.Tuple) and all(isinstance(elt, astroid.AssignName) for elt in node.target.elts) ): expr_list = [node.parent.key.name, node.parent.value.name] target_list = [elt.name for elt in node.target.elts] elif isinstance(node.parent, (astroid.ListComp, astroid.SetComp)): expr = node.parent.elt if isinstance(expr, astroid.Name): expr_list = expr.name elif isinstance(expr, astroid.Tuple): if any(not isinstance(elt, astroid.Name) for elt in expr.elts): return expr_list = [elt.name for elt in expr.elts] else: expr_list = [] target = node.parent.generators[0].target target_list = ( target.name if isinstance(target, astroid.AssignName) else ( [ elt.name for elt in target.elts if isinstance(elt, astroid.AssignName) ] if isinstance(target, astroid.Tuple) else [] ) ) else: return if expr_list == target_list != []: self.add_message("unnecessary-comprehension", node=node) @staticmethod def _is_and_or_ternary(node): """ Returns true if node is 'condition and true_value or false_value' form. All of: condition, true_value and false_value should not be a complex boolean expression """ return ( isinstance(node, astroid.BoolOp) and node.op == "or" and len(node.values) == 2 and isinstance(node.values[0], astroid.BoolOp) and not isinstance(node.values[1], astroid.BoolOp) and node.values[0].op == "and" and not isinstance(node.values[0].values[1], astroid.BoolOp) and len(node.values[0].values) == 2 ) @staticmethod def _and_or_ternary_arguments(node): false_value = node.values[1] condition, true_value = node.values[0].values return condition, true_value, false_value def visit_functiondef(self, node): self._return_nodes[node.name] = list( node.nodes_of_class(astroid.Return, skip_klass=astroid.FunctionDef) ) def _check_consistent_returns(self, node: astroid.FunctionDef) -> None: """Check that all return statements inside a function are consistent. Return statements are consistent if: - all returns are explicit and if there is no implicit return; - all returns are empty and if there is, possibly, an implicit return. Args: node (astroid.FunctionDef): the function holding the return statements. """ # explicit return statements are those with a not None value explicit_returns = [ _node for _node in self._return_nodes[node.name] if _node.value is not None ] if not explicit_returns: return if len(explicit_returns) == len( self._return_nodes[node.name] ) and self._is_node_return_ended(node): return self.add_message("inconsistent-return-statements", node=node) def _is_if_node_return_ended(self, node: astroid.If) -> bool: """Check if the If node ends with an explicit return statement. Args: node (astroid.If): If node to be checked. Returns: bool: True if the node ends with an explicit statement, False otherwise. """ # Do not check if inner function definition are return ended. is_if_returning = any( self._is_node_return_ended(_ifn) for _ifn in node.body if not isinstance(_ifn, astroid.FunctionDef) ) if not node.orelse: # If there is not orelse part then the if statement is returning if : # - there is at least one return statement in its siblings; # - the if body is itself returning. if not self._has_return_in_siblings(node): return False return is_if_returning # If there is an orelse part then both if body and orelse part should return. is_orelse_returning = any( self._is_node_return_ended(_ore) for _ore in node.orelse if not isinstance(_ore, astroid.FunctionDef) ) return is_if_returning and is_orelse_returning def _is_raise_node_return_ended(self, node: astroid.Raise) -> bool: """Check if the Raise node ends with an explicit return statement. Args: node (astroid.Raise): Raise node to be checked. Returns: bool: True if the node ends with an explicit statement, False otherwise. """ # a Raise statement doesn't need to end with a return statement # but if the exception raised is handled, then the handler has to # ends with a return statement if not node.exc: # Ignore bare raises return True if not utils.is_node_inside_try_except(node): # If the raise statement is not inside a try/except statement # then the exception is raised and cannot be caught. No need # to infer it. return True exc = utils.safe_infer(node.exc) if exc is None or exc is astroid.Uninferable or not hasattr(exc, "pytype"): return False exc_name = exc.pytype().split(".")[-1] handlers = utils.get_exception_handlers(node, exc_name) handlers = list(handlers) if handlers is not None else [] if handlers: # among all the handlers handling the exception at least one # must end with a return statement return any(self._is_node_return_ended(_handler) for _handler in handlers) # if no handlers handle the exception then it's ok return True def _is_node_return_ended(self, node: astroid.node_classes.NodeNG) -> bool: """Check if the node ends with an explicit return statement. Args: node (astroid.node_classes.NodeNG): node to be checked. Returns: bool: True if the node ends with an explicit statement, False otherwise. """ # Recursion base case if isinstance(node, astroid.Return): return True if isinstance(node, astroid.Call): try: funcdef_node = node.func.inferred()[0] if self._is_function_def_never_returning(funcdef_node): return True except astroid.InferenceError: pass # Avoid the check inside while loop as we don't know # if they will be completed if isinstance(node, astroid.While): return True if isinstance(node, astroid.Raise): return self._is_raise_node_return_ended(node) if isinstance(node, astroid.If): return self._is_if_node_return_ended(node) if isinstance(node, astroid.TryExcept): handlers = { _child for _child in node.get_children() if isinstance(_child, astroid.ExceptHandler) } all_but_handler = set(node.get_children()) - handlers return any( self._is_node_return_ended(_child) for _child in all_but_handler ) and all(self._is_node_return_ended(_child) for _child in handlers) # recurses on the children of the node return any(self._is_node_return_ended(_child) for _child in node.get_children()) @staticmethod def _has_return_in_siblings(node: astroid.node_classes.NodeNG) -> bool: """ Returns True if there is at least one return in the node's siblings """ next_sibling = node.next_sibling() while next_sibling: if isinstance(next_sibling, astroid.Return): return True next_sibling = next_sibling.next_sibling() return False def _is_function_def_never_returning(self, node: astroid.FunctionDef) -> bool: """Return True if the function never returns. False otherwise. Args: node (astroid.FunctionDef): function definition node to be analyzed. Returns: bool: True if the function never returns, False otherwise. """ try: return node.qname() in self._never_returning_functions except TypeError: return False def _check_return_at_the_end(self, node): """Check for presence of a *single* return statement at the end of a function. "return" or "return None" are useless because None is the default return type if they are missing. NOTE: produces a message only if there is a single return statement in the function body. Otherwise _check_consistent_returns() is called! Per its implementation and PEP8 we can have a "return None" at the end of the function body if there are other return statements before that! """ if len(self._return_nodes[node.name]) > 1: return if len(node.body) <= 1: return last = node.body[-1] if isinstance(last, astroid.Return): # e.g. "return" if last.value is None: self.add_message("useless-return", node=node) # return None" elif isinstance(last.value, astroid.Const) and (last.value.value is None): self.add_message("useless-return", node=node)