首页 > 学院 > 开发设计 > 正文

Python属性、方法和类管理系列之----元类

2019-11-14 17:02:55
字体:
来源:转载
供稿:网友

元类的介绍

请看位于下面网址的一篇文章,写的相当好。
http://blog.jobbole.com/21351/

实例补充

class Meta(type):    def __new__(meta, cls, parent, attr_dict):        res = super(Meta, meta).__new__(meta,cls, parent, attr_dict)        PRint('meta new stage, %s is %s, %s is %s' % (meta, type(meta), cls, type(cls)))         return res    def __init__(self,cls, parent, attr_dict):        super(Meta, self).__init__(cls,parent, attr_dict)        print('meta init stage, %s is %s, %s is %s' % (self, type(self), cls, type(cls)))    def __call__(self, *args, **kargs):        print('meta call stage, %s is %s' % (self, type(self)))        return super(Meta, self).__call__(*args, **kargs)def decorate(cls):    print('decorate cls, %s is %s' % (cls, type(cls)))    return cls@decorateclass A(metaclass=Meta):    def __new__(cls):        res = super(A, cls).__new__(cls)        print('A new stage, %s is %s' % (cls, type(cls)))        return res    def __init__(self):        super(A, self).__init__()        print('A init stage, %s is %s' % (self, type(self)))    def test(self):        passa=A()print(a)

运行结果如下:

meta new stage, <class '__main__.Meta'> is <class 'type'>, A is <class 'str'>meta init stage, <class '__main__.A'> is <class '__main__.Meta'>, A is <class 'str'>decorate cls, <class '__main__.A'> is <class '__main__.Meta'>meta call stage, <class '__main__.A'> is <class '__main__.Meta'>A new stage, <class '__main__.A'> is <class '__main__.Meta'>A init stage, <__main__.A object at 0x00000000022A74E0> is <class '__main__.A'><__main__.A object at 0x00000000022A74E0>

说明:
当我们自己创建一个类时,其实Python内部的运作机制如下:

  1. 看这个类中是否有设置元类,如果有,调用该元类进行初始化,如果没有,调用type进行初始化。
  2. 无论是我们自己定义的元类还是type,都有一个__new__方法,用来生成元类, 都有一个__init__用来初始化类。
  3. 查看是否有类的装饰器,如果有的话,调用之。
    其实,元类的__new____init__几乎什么都不做。
    当我们创建一个类的实例时,其实Python内部的运作机制如下:
  4. 调用元类的__call__方法,该方法会做两件事情:
  • 调用类自身的__new__方法用来创建类(如果有的话),如果我们没有显示的定义它,那么会调用从object继承过来的__new__方法。
  • 调用类自身的__init__方法(如果有的话)来初始化得到实例,如果我们没有显示的定义它,那么会调用从object继承过来的__init__方法。
    其实,object的__init__几乎什么都不做。

应用实例

由于我们经常在写类的内置拦截器方法时,少写下划线,或者出现拼写错误,从而怎么调试都不能发现问题所在,在浪费了很多时间以后才发现时犯的是多么低级的错误。
下面我写了这个元类来进行检查。

class AttrCheckMeta(type):    def __new__(meta, cls, parent, attr_dict):        import types        attrs_checking_list=['__init__', '__del__', '__call__', '__str__', '__repr__',                     '__getattr__', '__setattr__', '__delattr__', '__getattribute__',                    '__getitem__', '__setitem__', '__delitem__', '__iter__', '__next__',                    '__contains__', '__get__', '__set__', '__delete__', '__lt__',                     '__le__', '__gt__', '__ge__', '__eq__', '__add__', '__iadd__',                     '__radd__', '__sub__', '__isub__', '__rsub__', '__mul__', '__imul__',                    '__neg__', '__pos__', '__abs__', '__floordiv__', '__ifloordiv__',                     '__truediv__', '__itruediv__', '__mod__', '__imod__', '__imod__',                     '__pow__', '__ipow__', '__concat__', '__iconcat__', '__and__',                     '__iand__', '__or__', '__ior__', '__xor__', '__ixor__', '__inv__',                     '__invert__ ', '__lshift__', '__ilshift__', '__rshift__', '__irshift__ ',                    '__bool__', '__len__', '__nonzero__', '__enter__', '__exit__',                    '__new__', '__index__', '__oct__', '__hex__']        for attr,value in attr_dict.items():            #处理方法名前后都包含__,但是名字写错的情况。            if attr[:2]=='__' and attr[-2:]=='__' and isinstance(value, types.FunctionType):                if attr not in attrs_checking_list:                    print('found problem function: %s' % attr)            #处理漏写后面__的情况,此时Python会把这个方法吗当成是需要扩张的方法。            elif attr.startswith('_'+cls+'__') and isinstance(value, types.FunctionType):                print('maybe has problem: %s' % attr)        return super(AttrCheckMeta, meta).__new__(meta,cls, parent, attr_dict)    def __init__(self,cls, parent, attr_dict):        super(AttrCheckMeta, self).__init__(cls,parent, attr_dict)    def __call__(self, *args, **kargs):        return super(AttrCheckMeta, self).__call__(*args, **kargs)class A(metaclass=AttrCheckMeta):    def __new__(cls):        return super(A, cls).__new__(cls)    def __add(self, va, val):        pass    def __innit__(self):        super(A, self).__init__()a=A()

故意写了两个错误在类A中,运行结果如下:

found problem function name: __innit__maybe has problem: _A__add

当然,这个可以用装饰器来完成同样的任务,而且装饰器似乎更加直白、容易理解。
代码如下:

def check_ol(cls):    '''the overloading function name is easily to have spelling mistake.    It will be very hard to find the related mistakes, so i use this automethod to check    It will print the possible mistakes once found, will do nothing if passed'''    import types    attrs_checking_list=['__init__', '__del__', '__call__', '__str__', '__repr__',                 '__getattr__', '__setattr__', '__delattr__', '__getattribute__',                '__getitem__', '__setitem__', '__delitem__', '__iter__', '__next__',                '__contains__', '__get__', '__set__', '__delete__', '__lt__',                 '__le__', '__gt__', '__ge__', '__eq__', '__add__', '__iadd__',                 '__radd__', '__sub__', '__isub__', '__rsub__', '__mul__', '__imul__',                '__neg__', '__pos__', '__abs__', '__floordiv__', '__ifloordiv__',                 '__truediv__', '__itruediv__', '__mod__', '__imod__', '__imod__',                 '__pow__', '__ipow__', '__concat__', '__iconcat__', '__and__',                 '__iand__', '__or__', '__ior__', '__xor__', '__ixor__', '__inv__',                 '__invert__ ', '__lshift__', '__ilshift__', '__rshift__', '__irshift__ ',                '__bool__', '__len__', '__nonzero__', '__enter__', '__exit__',                '__new__', '__index__', '__oct__', '__hex__']    for attr,value in cls.__dict__.items():        #处理方法名前后都包含__,但是名字写错的情况。        if attr[:2]=='__' and attr[-2:]=='__' and isinstance(value, types.FunctionType):            if attr not in attrs_checking_list:                print('found problem function name: %s' % attr)        #处理漏写后面__的情况,此时Python会把这个方法吗当成是需要扩张的方法。        elif attr.startswith('_'+cls.__name__+'__') and isinstance(value, types.FunctionType):            print('maybe has problem: %s' % attr)    return cls

发表评论 共有条评论
用户名: 密码:
验证码: 匿名发表