未翻訳のページ

このページはまだ日本語に翻訳されていないため、英語の原文を表示しています。 翻訳に協力する

callable() Function Complexity

The callable() function checks whether an object is callable (can be invoked with arguments). It reads one slot on the object's type rather than looking up __call__ by name, so its cost does not depend on the object's attributes or the depth of its class hierarchy.

Complexity Analysis

Case	Time	Space	Notes
Check callable status	O(1)	O(1)	Reads the call slot on type(obj); no attribute lookup
Object with __call__	O(1)	O(1)	__call__ must be defined on the class or a base; one set on the instance is ignored
Built-in types	O(1)	O(1)	Every class is callable: its metaclass fills the slot

Basic Usage

Checking Functions and Methods

# O(1) - constant time check
def my_function():
    pass

callable(my_function)      # True
callable(len)              # True
callable("hi".upper)       # True (bound method)
callable(str.split)        # True (method descriptor)

Checking Non-Callable Objects

# O(1) - returns False
callable(42)               # False
callable("string")         # False
callable([1, 2, 3])        # False
callable({"key": "value"}) # False
callable(None)             # False

Classes and Instances

# O(1) - classes are callable
class MyClass:
    pass

callable(MyClass)          # True (can instantiate)

instance = MyClass()
callable(instance)         # False (unless __call__ defined)

# With __call__ method
class Callable:
    def __call__(self):
        return "called"

obj = Callable()
callable(obj)              # True (has __call__)

Complexity Details

How callable() Works

callable() never looks __call__ up by name. It reads the call slot on type(obj), which Python fills when the class or one of its bases defines __call__, so __getattr__ and __getattribute__ are never called. A __call__ set on the instance does not count, which is why hasattr(obj, "__call__") is not an equivalent:

class Plain:
    pass

obj = Plain()
obj.__call__ = lambda: "instance attribute"
assert hasattr(obj, "__call__")
assert not callable(obj)  # O(1) - the instance attribute is ignored

Plain.__call__ = lambda self: "class attribute"
assert callable(obj)  # O(1) - the class now fills the slot

A True Result Does Not Promise a Successful Call

callable() only says the type has a call slot. The call can still fail, for example on the wrong arguments or when the class sets __call__ to something that is not callable:

class Broken:
    __call__ = None

assert callable(Broken())  # O(1) - the slot is filled

try:
    Broken()()
except TypeError as e:
    assert "not callable" in str(e)
else:
    raise AssertionError("expected TypeError")

Common Patterns

Function Dispatch

# O(1) - check before calling
def execute_if_callable(func, *args):
    if callable(func):
        return func(*args)
    else:
        return func

result = execute_if_callable(len, [1, 2, 3])  # 3
value = execute_if_callable(42)  # 42

Callback Validation

# O(1) - validate callback parameter
def register_callback(callback):
    if not callable(callback):
        raise TypeError(f"callback must be callable, got {type(callback)}")
    # Register callback
    return callback

# Valid
register_callback(lambda x: x + 1)  # OK

# Invalid
try:
    register_callback(42)  # TypeError
except TypeError as e:
    print(e)

Polymorphic Processing

# O(n) for n items, plus the cost of each call
def process(items):
    results = []
    for item in items:
        if callable(item):  # O(1)
            results.append(item())  # Call it
        else:
            results.append(item)    # Use as-is
    return results

values = [42, lambda: 100, "string", dict]
assert process(values) == [42, 100, "string", {}]

Callables in Python

Functions

# All are callable O(1)

# Regular function
def func():
    pass
callable(func)  # True

# Lambda
f = lambda x: x + 1
callable(f)  # True

# Built-in function
callable(len)  # True
callable(print)  # True

# Method
obj = "hello"
callable(obj.upper)  # True

Classes

# O(1) - classes are callable (they instantiate)

class MyClass:
    pass

callable(MyClass)  # True

# Built-in types
callable(int)      # True
callable(str)      # True
callable(list)     # True
callable(dict)     # True

Objects with call

# O(1) - check for __call__ method

class Callable:
    def __call__(self):
        return "I was called"

obj = Callable()
callable(obj)  # True

result = obj()  # "I was called"

Performance Patterns

vs isinstance() Checks

# Both O(1), but different purposes
import types

obj = lambda x: x + 1

# callable() - high level
if callable(obj):
    pass

# isinstance() - more specific
if isinstance(obj, types.FunctionType):
    pass

# callable() is cleaner for just checking if callable

Batch Processing

# O(n) - check n items
# n calls to callable(), each O(1)

items = [42, lambda: 1, "str", len, [1, 2]]
callables = [x for x in items if callable(x)]
# O(n) total - [<lambda>, len]

Edge Cases

Partially Applied Functions

from functools import partial

# O(1) - partial objects are callable
add_five = partial(lambda x, y: x + y, 5)
callable(add_five)  # True

result = add_five(10)  # 15

Class Methods and Static Methods

# O(1) - each method, looked up on the class, is callable

class Example:
    @classmethod
    def class_method(cls):
        return "class"

    @staticmethod
    def static_method():
        return "static"

    def instance_method(self):
        return "instance"

callable(Example.class_method)      # True
callable(Example.static_method)     # True
callable(Example.instance_method)   # True

Generators and Iterators

# O(1) - functions are callable, but generators are not

def generator():
    yield 1

callable(generator)     # True (it's the function)
callable(generator())   # False (generator object not callable)

# Generator object has __iter__ and __next__, not __call__

Built-in Types

# O(1) - type check

class NotType:
    pass

# Instances - not callable
instance = NotType()
callable(instance)  # False

# Class - callable
callable(NotType)  # True

Practical Examples

Lazy Evaluation

# O(1) - check if value or factory
def lazy_value(obj):
    if callable(obj):
        return obj()  # Call factory
    else:
        return obj    # Return value directly

# Usage
value1 = lazy_value(42)              # 42
value2 = lazy_value(lambda: 42)      # 42 (factory called)

# Useful for configuration
def expensive_calculation():
    return sum(range(1_000))

config = {
    'static': 100,
    'dynamic': lambda: expensive_calculation()
}

for key, val in config.items():
    print(f"{key}: {lazy_value(val)}")

Function Composition

# O(1) - validate functions before composing
def compose(f, g):
    if not (callable(f) and callable(g)):
        raise TypeError("Both arguments must be callable")

    return lambda x: f(g(x))

square = lambda x: x ** 2
double = lambda x: x * 2

composed = compose(square, double)
result = composed(5)  # (5 * 2) ** 2 = 100

Event Handler Registration

# O(1) - validate event handler
class EventManager:
    def __init__(self):
        self.handlers = []

    def register(self, handler):
        if not callable(handler):
            raise TypeError("Handler must be callable")
        self.handlers.append(handler)

    def emit(self, event):
        for handler in self.handlers:
            handler(event)

manager = EventManager()
manager.register(lambda e: print(f"Event: {e}"))
manager.emit("click")  # Output: Event: click

Best Practices

✅ Do:

• Use callable() to check before invoking unknown objects
• Check in callback/handler validation
• Document which parameters should be callable
• Use in defensive programming

❌ Avoid:

• Assuming objects are callable without checking
• Using callable() for type discrimination (use isinstance() for that)
• Overcomplicating with __call__ when simple functions work
• Adding unnecessary callable() checks in performance-critical code

Related Functions

• hasattr() - Check for attribute existence
• isinstance() - Check object type
• type() - Get object type
• inspect.isfunction() - Detailed type checking

Version Notes

• Python 2.x: callable() available as built-in
• Python 3.0-3.1: Removed (controversial decision)
• Python 3.2+: Re-added as built-in
• All versions: O(1) constant time operation