python - 如何用可调用类装饰类方法？

Question

我希望类中的函数将它们的返回值存储在某种数据结构中。为此，我想使用装饰器:

results = []
instances = []

class A:
    def __init__(self, data):
        self.data = data

    @decorator
    def f1(self, a, b):
        return self.data + a + b

    @decorator
    def f2(self, a):
        return self.data + a + 1

x = A(1)
x.f1(1, 2)
x.f2(3)

print(results)

问题是，如何实现这个装饰器。

主要思想如下:

class Wrapper:

    def __init__(self, func):
        self.func = func

    def __call__(self, *args):
        res = self.func(*args)
        results.append(res)
        instances.append(args[0])

def decorator(func):
    return Wrapper(func)

但随后我收到一条错误消息:

TypeError: f1() missing 1 required positional argument: 'b'

这个问题和其他人问的类似(How to decorate a method inside a class?，Python functools.wraps equivalent for classes)，但是不清楚在哪里放置@functools.wraps或者调用@functools。 update_wrapped() 在我的例子中。

Answer 1

我认为这就是您的要求。

In [8]: from functools import wraps 
   ...: results = [] 
   ...: instances = [] 
   ...:  
   ...: class A: 
   ...:     def __init__(self, data): 
   ...:         self.data = data 
   ...:  
   ...:     def decorator(f): 
   ...:         @wraps(f) 
   ...:         def inner(*args, **kwargs): 
   ...:             retval = f(*args, **kwargs) 
   ...:             results.append(retval) 
   ...:             return retval  
   ...:         return inner 
   ...:         
   ...:  
   ...:     @decorator 
   ...:     def f1(self, a, b): 
   ...:         return self.data + a + b 
   ...:  
   ...:     @decorator 
   ...:     def f2(self, a): 
   ...:         return self.data + a + 1 
   ...:                                                                                         

In [9]: a = A(3)                                                                                

In [10]: a.f1(2,3)                                                                              
Out[10]: 8

In [11]: results                                                                                
Out[11]: [8]

Decorator 不必在类中，但当您需要访问类实例本身时，它会很方便。您无需更改任何代码即可将其取出。

In [8]: from functools import wraps 
   ...: results = [] 
   ...: instances = [] 

   ...: def decorator(f): 
   ...:         @wraps(f) 
   ...:         def inner(*args, **kwargs): 
   ...:             retval = f(*args, **kwargs) 
   ...:             results.append(retval) 
   ...:             return retval  
   ...:         return inner 
   ...:  
   ...: class A: 
   ...:     def __init__(self, data): 
   ...:         self.data = data 
   ...:  
   ...:         
   ...:  
   ...:     @decorator 
   ...:     def f1(self, a, b): 
   ...:         return self.data + a + b 
   ...:  
   ...:     @decorator 
   ...:     def f2(self, a): 
   ...:         return self.data + a + 1 
   ...:                                                                                         

In [9]: a = A(3)                                                                                

In [10]: a.f1(2,3)                                                                              
Out[10]: 8

In [11]: results                                                                                
Out[11]: [8]

效果相同。