functools — Higher-order functions and operations on callable objects

Source code: Lib/

The functools module is for higher-order functions: functions that act on or return other functions. In general, any callable object can be treated as a function for the purposes of this module.

The functools module defines the following functions:


Transform a method of a class into a property whose value is computed once and then cached as a normal attribute for the life of the instance. Similar to property(), with the addition of caching. Useful for expensive computed properties of instances that are otherwise effectively immutable.


class DataSet:
    def __init__(self, sequence_of_numbers):
        self._data = sequence_of_numbers

    def stdev(self):
        return statistics.stdev(self._data)

    def variance(self):
        return statistics.variance(self._data)

New in version 3.8.


This decorator requires that the __dict__ attribute on each instance be a mutable mapping. This means it will not work with some types, such as metaclasses (since the __dict__ attributes on type instances are read-only proxies for the class namespace), and those that specify __slots__ without including __dict__ as one of the defined slots (as such classes don’t provide a __dict__ attribute at all).


Transform an old-style comparison function to a key function. Used with tools that accept key functions (such as sorted(), min(), max(), heapq.nlargest(), heapq.nsmallest(), itertools.groupby()). This function is primarily used as a transition tool for programs being converted from Python 2 which supported the use of comparison functions.

A comparison function is any callable that accept two arguments, compares them, and returns a negative number for less-than, zero for equality, or a positive number for greater-than. A key function is a callable that accepts one argument and returns another value to be used as the sort key.


sorted(iterable, key=cmp_to_key(locale.strcoll))  # locale-aware sort order

For sorting examples and a brief sorting tutorial, see Sorting HOW TO.

New in version 3.2.

@functools.lru_cache(maxsize=128, typed=False)

Decorator to wrap a function with a memoizing callable that saves up to the maxsize most recent calls. It can save time when an expensive or I/O bound function is periodically called with the same arguments.

Since a dictionary is used to cache results, the positional and keyword arguments to the function must be hashable.

Distinct argument patterns may be considered to be distinct calls with separate cache entries. For example, f(a=1, b=2) and f(b=2, a=1) differ in their keyword argument order and may have two separate cache entries.

If user_function is specified, it must be a callable. This allows the lru_cache decorator to be applied directly to a user function, leaving the maxsize at its default value of 128:

def count_vowels(sentence):
    sentence = sentence.casefold()
    return sum(sentence.count(vowel) for vowel in 'aeiou')

If maxsize is set to None, the LRU feature is disabled and the cache can grow without bound. The LRU feature performs best when maxsize is a power-of-two.

If typed is set to true, function arguments of different types will be cached separately. For example, f(3) and f(3.0) will be treated as distinct calls with distinct results.

The wrapped function is instrumented with a cache_parameters() function that returns a new dict showing the values for maxsize and typed. This is for information purposes only. Mutating the values has no effect.

To help measure the effectiveness of the cache and tune the maxsize parameter, the wrapped function is instrumented with a cache_info() function that returns a named tuple showing hits, misses, maxsize and currsize. In a multi-threaded environment, the hits and misses are approximate.

The decorator also provides a cache_clear() function for clearing or invalidating the cache.

The original underlying function is accessible through the __wrapped__ attribute. This is useful for introspection, for bypassing the cache, or for rewrapping the function with a different cache.

An LRU (least recently used) cache works best when the most recent calls are the best predictors of upcoming calls (for example, the most popular articles on a news server tend to change each day). The cache’s size limit assures that the cache does not grow without bound on long-running processes such as web servers.

In general, the LRU cache should only be used when you want to reuse previously computed values. Accordingly, it doesn’t make sense to cache functions with side-effects, functions that need to create distinct mutable objects on each call, or impure functions such as time() or random().

Example of an LRU cache for static web content:

def get_pep(num):
    'Retrieve text of a Python Enhancement Proposal'
    resource = '' % num
        with urllib.request.urlopen(resource) as s:
    except urllib.error.HTTPError:
        return 'Not Found'

>>> for n in 8, 290, 308, 320, 8, 218, 320, 279, 289, 320, 9991:
...     pep = get_pep(n)
...     print(n, len(pep))

>>> get_pep.cache_info()
CacheInfo(hits=3, misses=8, maxsize=32, currsize=8)

Example of efficiently computing Fibonacci numbers using a cache to implement a dynamic programming technique:

def fib(n):
    if n < 2:
        return n
    return fib(n-1) + fib(n-2)

>>> [fib(n) for n in range(16)]
[0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610]

>>> fib.cache_info()
CacheInfo(hits=28, misses=16, maxsize=None, currsize=16)

New in version 3.2.

Changed in version 3.3: Added the typed option.

Changed in version 3.8: Added the user_function option.

New in version 3.9: Added the function cache_parameters()


Given a class defining one or more rich comparison ordering methods, this class decorator supplies the rest. This simplifies the effort involved in specifying all of the possible rich comparison operations:

The class must define one of __lt__(), __le__(), __gt__(), or __ge__(). In addition, the class should supply an __eq__() method.

For example:

class Student:
    def _is_valid_operand(self, other):
        return (hasattr(other, "lastname") and
                hasattr(other, "firstname"))
    def __eq__(self, other):
        if not self._is_valid_operand(other):
            return NotImplemented
        return ((self.lastname.lower(), self.firstname.lower()) ==
                (other.lastname.lower(), other.firstname.lower()))
    def __lt__(self, other):
        if not self._is_valid_operand(other):
            return NotImplemented
        return ((self.lastname.lower(), self.firstname.lower()) <
                (other.lastname.lower(), other.firstname.lower()))


While this decorator makes it easy to create well behaved totally ordered types, it does come at the cost of slower execution and more complex stack traces for the derived comparison methods. If performance benchmarking indicates this is a bottleneck for a given application, implementing all six rich comparison methods instead is likely to provide an easy speed boost.

New in version 3.2.

Changed in version 3.4: Returning NotImplemented from the underlying comparison function for unrecognised types is now supported.

functools.partial(func, /, *args, **keywords)

Return a new partial object which when called will behave like func called with the positional arguments args and keyword arguments keywords. If more arguments are supplied to the call, they are appended to args. If additional keyword arguments are supplied, they extend and override keywords. Roughly equivalent to:

def partial(func, /, *args, **keywords):
    def newfunc(*fargs, **fkeywords):
        newkeywords = {**keywords, **fkeywords}
        return func(*args, *fargs, **newkeywords)
    newfunc.func = func
    newfunc.args = args
    newfunc.keywords = keywords
    return newfunc

The partial() is used for partial function application which “freezes” some portion of a function’s arguments and/or keywords resulting in a new object with a simplified signature. For example, partial() can be used to create a callable that behaves like the int() function where the base argument defaults to two:

>>> from functools import partial
>>> basetwo = partial(int, base=2)
>>> basetwo.__doc__ = 'Convert base 2 string to an int.'
>>> basetwo('10010')
class functools.partialmethod(func, /, *args, **keywords)

Return a new partialmethod descriptor which behaves like partial except that it is designed to be used as a method definition rather than being directly callable.

func must be a descriptor or a callable (objects which are both, like normal functions, are handled as descriptors).

When func is a descriptor (such as a normal Python function, classmethod(), staticmethod(), abstractmethod() or another instance of partialmethod), calls to __get__ are delegated to the underlying descriptor, and an appropriate partial object returned as the result.

When func is a non-descriptor callable, an appropriate bound method is created dynamically. This behaves like a normal Python function when used as a method: the self argument will be inserted as the first positional argument, even before the args and keywords supplied to the partialmethod constructor.


>>> class Cell:
...     def __init__(self):
...         self._alive = False
...     @property
...     def alive(self):
...         return self._alive
...     def set_state(self, state):
...         self._alive = bool(state)
...     set_alive = partialmethod(set_state, True)
...     set_dead = partialmethod(set_state, False)
>>> c = Cell()
>>> c.alive
>>> c.set_alive()
>>> c.alive

New in version 3.4.

functools.reduce(function, iterable[, initializer])

Apply function of two arguments cumulatively to the items of iterable, from left to right, so as to reduce the iterable to a single value. For example, reduce(lambda x, y: x+y, [1, 2, 3, 4, 5]) calculates ((((1+2)+3)+4)+5). The left argument, x, is the accumulated value and the right argument, y, is the update value from the iterable. If the optional initializer is present, it is placed before the items of the iterable in the calculation, and serves as a default when the iterable is empty. If initializer is not given and iterable contains only one item, the first item is returned.

Roughly equivalent to:

def reduce(function, iterable, initializer=None):
    it = iter(iterable)
    if initializer is None:
        value = next(it)
        value = initializer
    for element in it:
        value = function(value, element)
    return value

See itertools.accumulate() for an iterator that yields all intermediate values.


Transform a function into a single-dispatch generic function.

To define a generic function, decorate it with the @singledispatch decorator. Note that the dispatch happens on the type of the first argument, create your function accordingly:

>>> from functools import singledispatch
>>> @singledispatch
... def fun(arg, verbose=False):
...     if verbose:
...         print("Let me just say,", end=" ")
...     print(arg)

To add overloaded implementations to the function, use the register() attribute of the generic function. It is a decorator. For functions annotated with types, the decorator will infer the type of the first argument automatically:

>>> @fun.register
... def _(arg: int, verbose=False):
...     if verbose:
...         print("Strength in numbers, eh?", end=" ")
...     print(arg)
>>> @fun.register
... def _(arg: list, verbose=False):
...     if verbose:
...         print("Enumerate this:")
...     for i, elem in enumerate(arg):
...         print(i, elem)

For code which doesn’t use type annotations, the appropriate type argument can be passed explicitly to the decorator itself:

>>> @fun.register(complex)
... def _(arg, verbose=False):
...     if verbose:
...         print("Better than complicated.", end=" ")
...     print(arg.real, arg.imag)

To enable registering lambdas and pre-existing functions, the register() attribute can be used in a functional form:

>>> def nothing(arg, verbose=False):
...     print("Nothing.")
>>> fun.register(type(None), nothing)

The register() attribute returns the undecorated function which enables decorator stacking, pickling, as well as creating unit tests for each variant independently:

>>> @fun.register(float)
... @fun.register(Decimal)
... def fun_num(arg, verbose=False):
...     if verbose:
...         print("Half of your number:", end=" ")
...     print(arg / 2)
>>> fun_num is fun

When called, the generic function dispatches on the type of the first argument:

>>> fun("Hello, world.")
Hello, world.
>>> fun("test.", verbose=True)
Let me just say, test.
>>> fun(42, verbose=True)
Strength in numbers, eh? 42
>>> fun(['spam', 'spam', 'eggs', 'spam'], verbose=True)
Enumerate this:
0 spam
1 spam
2 eggs
3 spam
>>> fun(None)
>>> fun(1.23)

Where there is no registered implementation for a specific type, its method resolution order is used to find a more generic implementation. The original function decorated with @singledispatch is registered for the base object type, which means it is used if no better implementation is found.

If an implementation registered to abstract base class, virtual subclasses will be dispatched to that implementation:

>>> from import Mapping
>>> @fun.register
... def _(arg: Mapping, verbose=False):
...     if verbose:
...         print("Keys & Values")
...     for key, value in arg.items():
...         print(key, "=>", value)
>>> fun({"a": "b"})
a => b

To check which implementation will the generic function choose for a given type, use the dispatch() attribute:

>>> fun.dispatch(float)
<function fun_num at 0x1035a2840>
>>> fun.dispatch(dict)    # note: default implementation
<function fun at 0x103fe0000>

To access all registered implementations, use the read-only registry attribute:

>>> fun.registry.keys()
dict_keys([<class 'NoneType'>, <class 'int'>, <class 'object'>,
          <class 'decimal.Decimal'>, <class 'list'>,
          <class 'float'>])
>>> fun.registry[float]
<function fun_num at 0x1035a2840>
>>> fun.registry[object]
<function fun at 0x103fe0000>

New in version 3.4.

Changed in version 3.7: The register() attribute supports using type annotations.

class functools.singledispatchmethod(func)

Transform a method into a single-dispatch generic function.

To define a generic method, decorate it with the @singledispatchmethod decorator. Note that the dispatch happens on the type of the first non-self or non-cls argument, create your function accordingly:

class Negator:
    def neg(self, arg):
        raise NotImplementedError("Cannot negate a")

    def _(self, arg: int):
        return -arg

    def _(self, arg: bool):
        return not arg

@singledispatchmethod supports nesting with other decorators such as @classmethod. Note that to allow for dispatcher.register, singledispatchmethod must be the outer most decorator. Here is the Negator class with the neg methods being class bound:

class Negator:
    def neg(cls, arg):
        raise NotImplementedError("Cannot negate a")

    def _(cls, arg: int):
        return -arg

    def _(cls, arg: bool):
        return not arg

The same pattern can be used for other similar decorators: staticmethod, abstractmethod, and others.

New in version 3.8.

class functools.TopologicalSorter(graph=None)

Provides functionality to topologically sort a graph of hashable nodes.

A topological order is a linear ordering of the vertices in a graph such that for every directed edge u -> v from vertex u to vertex v, vertex u comes before vertex v in the ordering. For instance, the vertices of the graph may represent tasks to be performed, and the edges may represent constraints that one task must be performed before another; in this example, a topological ordering is just a valid sequence for the tasks. A complete topological ordering is possible if and only if the graph has no directed cycles, that is, if it is a directed acyclic graph.

If the optional graph argument is provided it must be a dictionary representing a directed acyclic graph where the keys are nodes and the values are iterables of all predecessors of that node in the graph (the nodes that have edges that point to the value in the key). Additional nodes can be added to the graph using the add() method.

In the general case, the steps required to perform the sorting of a given graph are as follows:

  • Create an instance of the TopologicalSorter with an optional initial graph.

  • Add additional nodes to the graph.

  • Call prepare() on the graph.

  • While is_active() is True, iterate over the nodes returned by get_ready() and process them. Call done() on each node as it finishes processing.

In case just an immediate sorting of the nodes in the graph is required and no parallelism is involved, the convenience method TopologicalSorter.static_order() can be used directly:

>>> graph = {"D": {"B", "C"}, "C": {"A"}, "B": {"A"}}
>>> ts = TopologicalSorter(graph)
>>> tuple(ts.static_order())
('A', 'C', 'B', 'D')

The class is designed to easily support parallel processing of the nodes as they become ready. For instance:

topological_sorter = TopologicalSorter()

# Add nodes to 'topological_sorter'...

while topological_sorter.is_active():
    for node in topological_sorter.get_ready():
        # Worker threads or processes take nodes to work on off the
        # 'task_queue' queue.

    # When the work for a node is done, workers put the node in
    # 'finalized_tasks_queue' so we can get more nodes to work on.
    # The definition of 'is_active()' guarantees that, at this point, at
    # least one node has been placed on 'task_queue' that hasn't yet
    # been passed to 'done()', so this blocking 'get()' must (eventually)
    # succeed.  After calling 'done()', we loop back to call 'get_ready()'
    # again, so put newly freed nodes on 'task_queue' as soon as
    # logically possible.
    node = finalized_tasks_queue.get()
add(node, *predecessors)

Add a new node and its predecessors to the graph. Both the node and all elements in predecessors must be hashable.

If called multiple times with the same node argument, the set of dependencies will be the union of all dependencies passed in.

It is possible to add a node with no dependencies (predecessors is not provided) or to provide a dependency twice. If a node that has not been provided before is included among predecessors it will be automatically added to the graph with no predecessors of its own.

Raises ValueError if called after prepare().


Mark the graph as finished and check for cycles in the graph. If any cycle is detected, CycleError will be raised, but get_ready() can still be used to obtain as many nodes as possible until cycles block more progress. After a call to this function, the graph cannot be modified, and therefore no more nodes can be added using add().


Returns True if more progress can be made and False otherwise. Progress can be made if cycles do not block the resolution and either there are still nodes ready that haven’t yet been returned by TopologicalSorter.get_ready() or the number of nodes marked TopologicalSorter.done() is less than the number that have been returned by TopologicalSorter.get_ready().

The __bool__() method of this class defers to this function, so instead of:

if ts.is_active():

if possible to simply do:

if ts:

Raises ValueError if called without calling prepare() previously.


Marks a set of nodes returned by TopologicalSorter.get_ready() as processed, unblocking any successor of each node in nodes for being returned in the future by a call to TopologicalSorter.get_ready().

Raises ValueError if any node in nodes has already been marked as processed by a previous call to this method or if a node was not added to the graph by using TopologicalSorter.add(), if called without calling prepare() or if node has not yet been returned by get_ready().


Returns a tuple with all the nodes that are ready. Initially it returns all nodes with no predecessors, and once those are marked as processed by calling TopologicalSorter.done(), further calls will return all new nodes that have all their predecessors already processed. Once no more progress can be made, empty tuples are returned.

Raises ValueError if called without calling prepare() previously.


Returns an iterable of nodes in a topological order. Using this method does not require to call TopologicalSorter.prepare() or TopologicalSorter.done(). This method is equivalent to:

def static_order(self):
    while self.is_active():
        node_group = self.get_ready()
        yield from node_group

The particular order that is returned may depend on the specific order in which the items were inserted in the graph. For example:

>>> ts = TopologicalSorter()
>>> ts.add(3, 2, 1)
>>> ts.add(1, 0)
>>> print([*ts.static_order()])
[2, 0, 1, 3]

>>> ts2 = TopologicalSorter()
>>> ts2.add(1, 0)
>>> ts2.add(3, 2, 1)
>>> print([*ts2.static_order()])
[0, 2, 1, 3]

This is due to the fact that “0” and “2” are in the same level in the graph (they would have been returned in the same call to get_ready()) and the order between them is determined by the order of insertion.

If any cycle is detected, CycleError will be raised.

New in version 3.9.

functools.update_wrapper(wrapper, wrapped, assigned=WRAPPER_ASSIGNMENTS, updated=WRAPPER_UPDATES)

Update a wrapper function to look like the wrapped function. The optional arguments are tuples to specify which attributes of the original function are assigned directly to the matching attributes on the wrapper function and which attributes of the wrapper function are updated with the corresponding attributes from the original function. The default values for these arguments are the module level constants WRAPPER_ASSIGNMENTS (which assigns to the wrapper function’s __module__, __name__, __qualname__, __annotations__ and __doc__, the documentation string) and WRAPPER_UPDATES (which updates the wrapper function’s __dict__, i.e. the instance dictionary).

To allow access to the original function for introspection and other purposes (e.g. bypassing a caching decorator such as lru_cache()), this function automatically adds a __wrapped__ attribute to the wrapper that refers to the function being wrapped.

The main intended use for this function is in decorator functions which wrap the decorated function and return the wrapper. If the wrapper function is not updated, the metadata of the returned function will reflect the wrapper definition rather than the original function definition, which is typically less than helpful.

update_wrapper() may be used with callables other than functions. Any attributes named in assigned or updated that are missing from the object being wrapped are ignored (i.e. this function will not attempt to set them on the wrapper function). AttributeError is still raised if the wrapper function itself is missing any attributes named in updated.

New in version 3.2: Automatic addition of the __wrapped__ attribute.

New in version 3.2: Copying of the __annotations__ attribute by default.

Changed in version 3.2: Missing attributes no longer trigger an AttributeError.

Changed in version 3.4: The __wrapped__ attribute now always refers to the wrapped function, even if that function defined a __wrapped__ attribute. (see bpo-17482)

@functools.wraps(wrapped, assigned=WRAPPER_ASSIGNMENTS, updated=WRAPPER_UPDATES)

This is a convenience function for invoking update_wrapper() as a function decorator when defining a wrapper function. It is equivalent to partial(update_wrapper, wrapped=wrapped, assigned=assigned, updated=updated). For example:

>>> from functools import wraps
>>> def my_decorator(f):
...     @wraps(f)
...     def wrapper(*args, **kwds):
...         print('Calling decorated function')
...         return f(*args, **kwds)
...     return wrapper
>>> @my_decorator
... def example():
...     """Docstring"""
...     print('Called example function')
>>> example()
Calling decorated function
Called example function
>>> example.__name__
>>> example.__doc__

Without the use of this decorator factory, the name of the example function would have been 'wrapper', and the docstring of the original example() would have been lost.

partial Objects

partial objects are callable objects created by partial(). They have three read-only attributes:


A callable object or function. Calls to the partial object will be forwarded to func with new arguments and keywords.


The leftmost positional arguments that will be prepended to the positional arguments provided to a partial object call.


The keyword arguments that will be supplied when the partial object is called.

partial objects are like function objects in that they are callable, weak referencable, and can have attributes. There are some important differences. For instance, the __name__ and __doc__ attributes are not created automatically. Also, partial objects defined in classes behave like static methods and do not transform into bound methods during instance attribute look-up.


The functools module defines the following exception classes:

exception functools.CycleError

Subclass of ValueError raised by TopologicalSorter.prepare() if cycles exist in the working graph. If multiple cycles exist, only one undefined choice among them will be reported and included in the exception.

The detected cycle can be accessed via the second element in the args attribute of the exception instance and consists in a list of nodes, such that each node is, in the graph, an immediate predecessor of the next node in the list. In the reported list, the first and the last node will be the same, to make it clear that it is cyclic.