Home > Article > Backend Development > Summarize the knowledge points of decorators in Python
This article brings you relevant knowledge about python, which mainly introduces related issues about decorators, including closures, decorators, using multiple decorators, and parameters. Decorators and other contents, let’s take a look at them below. I hope it will be helpful to everyone.
Recommended learning: python video tutorial
To understand what is Decorator (decorator), we first need to know the concept of closure (closure).
Closure, also known as closure function or closed function, generally speaking, when a function is returned as an object and also entrains external variables, a closure is formed.
Taking printing Hello World as an example, let's first take a look at what the structure of the nested function should look like:
def print_msg(msg):
def printer():
print(msg)
printer()print_msg('Hello World')# Hello World
Execution print_msg('Hello World') It is equivalent to executing printer(), that is, executing print(msg), so Hello World will be output.
Let’s take a look at what kind of structure it would be if it were a closure:
def print_msg(msg):
def printer():
print(msg)
return printer
my_msg = print_msg('Hello World')my_msg()# Hello World
The
printerfunction in this example is a closure.
Executionprint_msg('Hello World') actually returns a function like the following, which entrains external variables 'Hello World':
def printer():
print('Hello World')
So calling my_msg is equivalent to executing printer().
So how to determine whether a function is a closure function? The __closure__ attribute of the closure function defines a tuple for storing all cell objects. Each cell object stores all external variables in the closure. The __closure__ attribute of a normal function is None.
def outer(content):
def inner():
print(content)
return innerprint(outer.__closure__)
# Noneinner = outer('Hello World')print(inner.__closure__)
# (<cell at 0x0000023FB1FD0B80: str object at 0x0000023FB1DC84F0>,)
It can be seen that the outer function is not a closure, and the inner function is a closure.
We can also view the external variables carried by the closure:
print(inner.__closure__[0].cell_contents)# Hello World
Having said so much, what is the use of closures? The meaning of the existence of a closure is that it carries external variables (private goods). If it does not carry private goods, then it is no different from an ordinary function.
The advantages of closures are as follows:
Let’s first consider such a scenario. Assume that a previously written function has implemented 4 functions. For simplicity, we use print statement to represent each specific function:
def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')
Now, for some reason, you need to add a function 5 to the module function, you It can be modified like this:
def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')
print('功能5')
But in real business, it is often dangerous to make such modifications directly (it will become difficult to maintain). So How to add a new function to it without modifying the original function?
You may have thought of using the previous closure knowledge:
def func_5(original_module):
def wrapper():
original_module()
print('功能5')
return wrapper
func_5 means that the function is mainly used to implement function 5 , we will next pass module in to observe the effect:
new_module = func_5(module)new_module()# 功能1# 功能2# 功能3# 功能4# 功能5
It can be seen that our new module: new_module has implemented function 5 .
In the above example, function
func_5is a decorator, which decorates the original module (adds a new function to it).
Of course, Python has a more concise way of writing (called syntactic sugar), we can use the @ symbol with the name of the decorator function and place it in the definition of the function to be decorated Above:
def func_5(original_module):
def wrapper():
original_module()
print('功能5')
return wrapper@func_5def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')module()# 功能1# 功能2# 功能3# 功能4# 功能5
Based on this, we can complete the timing task (calculate the running time of the original function) without modifying the original function, as follows:
def timer(func):
def wrapper():
import time
tic = time.time()
func()
toc = time.time()
print('程序用时: {}s'.format(toc - tic))
return wrapper@timerdef make_list():
return [i * i for i in range(10**7)]my_list = make_list()# 程序用时: 0.8369960784912109s
In fact, my_list is not a list. Direct printing will display None. This is because our wrapper function does not set a return value. If you need to get the return value of make_list, you can modify the wrapper function like this:
def wrapper():
import time
tic = time.time()
a = func()
toc = time.time()
print('程序用时: {}s'.format(toc - tic))
return a
If we want to module Newly added function 5 and function 6 (in numerical order), what should I do?
Fortunately, Python allows the use of multiple decorators at the same time:
def func_5(original_module):
def wrapper():
original_module()
print('功能5')
return wrapperdef func_6(original_module):
def wrapper():
original_module()
print('功能6')
return wrapper@func_6@func_5def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')module()# 功能1# 功能2# 功能3# 功能4# 功能5# 功能6
The above process is actually equivalent to:
def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')new_module = func_6(func_5(module))new_module()
In addition, it should be noted that when using multiple decorators When decorating a decorator, the decorator closest to the function definition will decorate the function first . If we change the decoration order, the output result will also change:
@func_5@func_6def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')module()# 功能1# 功能2# 功能3# 功能4# 功能6# 功能5
If the decorated function has parameters, how to construct the decorator?
Consider such a function:
def pide(a, b): return a / b
当b=0 时会出现 ZeropisionError。如何在避免修改该函数的基础上给出一个更加人性化的提醒呢？
因为我们的 pide 函数接收两个参数，所以我们的 wrapper 函数也应当接收两个参数：
def smart_pide(func): def wrapper(a, b): if b == 0: return '被除数不能为0!' else: return func(a, b) return wrapper
使用该装饰器进行装饰：
@smart_pidedef pide(a, b): return a / bprint(pide(3, 0))# 被除数不能为0!print(pide(3, 1))# 3.0
如果不知道要被装饰的函数有多少个参数，我们可以使用下面更为通用的模板：
def decorator(func): def wrapper(*args, **kwargs): # ... res = func(*args, **kwargs) # ... return res # 也可以不return return wrapper
我们之前提到的装饰器都没有带参数，即语法糖 @decorator 中没有参数，那么该如何写一个带参数的装饰器呢？
前面实现的装饰器都是两层嵌套函数，而带参数的装饰器是一个三层嵌套函数。
考虑这样一个场景。假如我们在为 module 添加新功能时，希望能够加上实现该功能的开发人员的花名，则可以这样构造装饰器（以 功能5 为例）：
def func_5_with_name(name=None):
def func_5(original_module):
def wrapper():
original_module()
print('功能5由{}实现'.format(name))
return wrapper return func_5
效果如下：
@func_5_with_name(name='若水')def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')module()# 功能1# 功能2# 功能3# 功能4# 功能5由若水实现
对于这种三层嵌套函数，我们可以这样理解：当为 func_5_with_name 指定了参数后，func_5_with_name(name='若水') 实际上返回了一个 decorator，于是 @func_5_with_name(name='若水') 就相当于 @decorator。
将类作为装饰器，我们需要实现 __init__ 方法和 __call__ 方法。
以计时器为例，具体实现如下：
class Timer:
def __init__(self, func):
self.func = func def __call__(self):
import time
tic = time.time()
self.func()
toc = time.time()
print('用时: {}s'.format(toc - tic))@Timerdef make_list():
return [i**2 for i in range(10**7)]make_list()# 用时: 2.928966999053955s
如果想要自定义生成列表的长度并获得列表（即被装饰的函数带有参数情形），我们就需要在 __call__ 方法中传入相应的参数，具体如下：
class Timer:
def __init__(self, func):
self.func = func def __call__(self, num):
import time
tic = time.time()
res = self.func(num)
toc = time.time()
print('用时: {}s'.format(toc - tic))
return res@Timerdef make_list(num):
return [i**2 for i in range(num)]my_list = make_list(10**7)# 用时: 2.8219943046569824sprint(len(my_list))# 10000000
如果要构建带参数的类装饰器，则不能把 func 传入 __init__ 中，而是传入到 __call__ 中，同时 __init__ 用来初始化类装饰器的参数。
接下来我们使用类装饰器来复现第五章节中的效果：
class Func_5:
def __init__(self, name=None):
self.name = name def __call__(self, func):
def wrapper():
func()
print('功能5由{}实现'.format(self.name))
return wrapper@Func_5('若水')def module():
print('功能1')
print('功能2')
print('功能3')
print('功能4')module()# 功能1# 功能2# 功能3# 功能4# 功能5由若水实现
Python中有许多内置装饰器，这里仅介绍最常见的三种：@classmethod、@staticmethod 和 @property。
@classmethod 用于装饰类中的函数，使用它装饰的函数不需要进行实例化也可调用。需要注意的是，被装饰的函数不需要 self 参数，但第一个参数需要是表示自身类的 cls 参数，它可以来调用类的属性，类的方法，实例化对象等。
cls代表类本身，self代表实例本身。
具体请看下例：
class A:
num = 100
def func1(self):
print('功能1')
@classmethod
def func2(cls):
print('功能2')
print(cls.num)
cls().func1()A.func2()# 功能2# 100# 功能1
@staticmethod 同样用来修饰类中的方法，使用它装饰的函数的参数没有任何限制（即无需传入 self 参数），并且可以不用实例化调用该方法。当然，实例化后调用该方法也是允许的。
具体如下：
class A: @staticmethod def add(a, b): return a + bprint(A.add(2, 3))# 5print(A().add(2, 3))# 5
使用 @property 装饰器，我们可以直接通过方法名来访问类方法，不需要在方法名后添加一对 () 小括号。
class A:
@property
def printer(self):
print('Hello World')a = A()a.printer# Hello World
除此之外，@property 还可以用来防止类的属性被修改。考虑如下场景
class A: def __init__(self): self.name = 'ABC'a = A()print(a.name)# ABCa.name = 1print(a.name)# 1
可以看出类中的属性 name 可以被随意修改。如果要防止修改，则可以这样做
class A: def __init__(self): self.name_ = 'ABC' @property def name(self): return self.name_ a = A()print(a.name)# ABCa.name = 1print(a.name)# AttributeError: can't set attribute
推荐学习：python视频教程
The above is the detailed content of Summarize the knowledge points of decorators in Python. For more information, please follow other related articles on the PHP Chinese website!