Python 类在每次调用时都继承了单例初始化实例 [英] Python class inherited singleton inits instance on every call
问题描述
我正在尝试按照here(方法2).回顾这个问题和广泛选择的答案,我尝试实现以下内容:
I'm trying to implement class inherited singleton as described here (Method 2). Going over the question and the extensive chosen answer I tried to implement the following:
class Singleton(object):
_instance = None
def __new__(cls, *args, **kwargs):
if not isinstance(cls._instance, cls):
cls._instance = object.__new__(cls, *args, **kwargs)
cls._instance._initialized = False
return cls._instance
class A(Singleton):
def __init__(self):
print "Init is called"
class B(Singleton):
def __init__(self):
print "Init is called"
正如您可能猜到的,每当我创建 Class A
时,我都会得到相同的对象,但会调用 __init__
.这是有问题的,因为 Class A
可能会因此更改其所有成员.
As you may guess, whenever I create Class A
I get the same object, but __init__
is called. This is problematic as the Class A
can have all it's members changed due to this.
正在做:
a1 = A()
a2 = A()
print a1 == a2
将导致:
>> Init is called
>> Init is called
>> True
这个问题提出了一个类似的问题,但我不想在那里使用解决方案,因为它不包括继承,我至少有 2 个类需要 Singleton
继承.我试图在 here 实施解决方案,但它没有工作.此处的解决方案有效,但它涉及更改Class A
和Class B
我不想这样做.
This question poses a similar issue but I would prefer not to use the solution there as it doesn't include the inheritance and I have at least 2 classes that needs the Singleton
inheritance. I tried to implement the solution here but it didn't work. The solution here works but it involves changing Class A
and Class B
which I would prefer not to.
有没有办法改变 Singleton
的实现,这样 __init__
就不会在每次创建时被调用?(我不能使用元类,因为 A
和 B
都继承了其他类,例如具有自己的元类的抽象类.
Is there a way to change the Singleton
implementation so that __init__
won't be called on every creation? (I can't use metaclasses as both A
and B
inherit other classes e.g. abstract classes with their own metaclasses).
谢谢
推荐答案
在现代 Python 中,这可以通过编写一个 __init_subclass__
方法来完成,该方法可以装饰 __init__
使其成为运行前检查cls._instance
.
In modern Python this could be done by writting a __init_subclass__
method that could decorate __init__
to make it check cls._instance
before running.
事实上,(即你也需要它用于 Python 2) - 我认为更简单的事情是 __new__
用 NOP 方法修补 __init__
如果实例已经存在:
As it is, (i.e. you need it for Python 2 as well) - I think the simpler thing is for __new__
to patch __init__
with a NOP method if the instance already exists:
_nop_init = lambda self, *args, **kw: None
class Singleton(object):
_instance = None
def __new__(cls, *args, **kwargs):
if not isinstance(cls._instance, cls):
cls._instance = object.__new__(cls, *args, **kwargs)
cls._instance._initialized = False
# Python 2 have to check in the cls.__dict__ - Py3 could check the attribute directly:
elif cls.__dict__.get("__init__", None) is not _nop_init:
cls.__init__ = _nop_init
return cls._instance
class A(Singleton):
def __init__(self):
print "Init is called"
class B(Singleton):
def __init__(self):
print "Init is called"
额外信息 调用__init__
的语言机制内置于type
的__call__
方法中——Python 中所有类的元类.在实例化目标类时,它会调用目标类的 __new__
和 __init__
方法,因此,使用元类,可以轻松地从自定义元类控制这些调用.有趣的是,当 __new__
没有返回目标类的实例时,__init__
不会被调用.(在这种单例情况下,单例是 te 类的一个实例,因此它被调用).
extra info The language mechanism to call __init__
is built-in the __call__
method of type
- the metaclass for all classes in Python. It will call the __new__
and __init__
method of a target class when instantiating it, and thus, with a metaclass, it is easy to control these calls from a custom metaclass. It is also interesting to note that when __new__
does not return an instance of the target class, __init__
is not called. (In this singleton case, the singleton is an instance of te class, and thus it is called).
真实世界示例:上次我编写单例时,我选择避免在自己的代码中重新运行 __init__
- 因为它是项目中唯一的单例,在 __new__
中不需要通用代码:https://github.com/jsbueno/pythonchain/blob1f9208dc8bd2741a574adc1bf745d218e4314e4a/pythonchain/block.py#L276
real world example: last time I coded a singleton, I opted for avoiding re-run __init__
in its own code - sine it is the only singleton in the project, tere was no need for generic code for that in a __new__
:
https://github.com/jsbueno/pythonchain/blob/1f9208dc8bd2741a574adc1bf745d218e4314e4a/pythonchain/block.py#L276
这篇关于Python 类在每次调用时都继承了单例初始化实例的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!