元类编程

一. property动态属性
 
1. 首先来个例子,需求是根据出生年月日,得到某人的年龄
from datetime import date, datetime
 
 
class User:
    def __init__(self, name, birthday):
        self.name = name
        self.birthday = birthday
 
    def get_age(self):
        return datetime.now().year - self.birthday.year
 
if __name__ == "__main__":
    user1 = User("jack", date(1982, 3, 21))
    print(user1.get_age())
输出结果为36
 
注意:这里学习一下取时间的方法
>>> from datetime import date, datetime
 
#使用datetime来取值
>>> datetime.now().year
2018
>>> datetime.now().month
10
 
# 使用date来取值,所得的值为int
>>> a=date(1982,3,21)
>>> a.year
1982
>>> type(a.year)
<class 'int'>
 
 
 
2. 下面使用@property装饰器来优化代码,它会把下面定义的函数变成一个属性描述符,也就是把函数当作一个属性来用
from datetime import date, datetime
 
 
class User:
    def __init__(self, name, birthday):
        self.name = name
        self.birthday = birthday
 
    @property
    def age(self):
        return datetime.now().year - self.birthday.year
 
if __name__ == "__main__":
    user1 = User("jack", date(1982, 3, 21))
    print(user1.age)
输出结果同样为36,注意最后我们用的是user1.age来得到年龄
 
 
 
3. 上面的@propety相当于获取年龄,下面的例子中,装饰器@age.setter可用来设置年龄
from datetime import date, datetime
 
 
class User:
    def __init__(self, name, birthday):
        self.name = name
        self.birthday = birthday
 
    @property
    def myage(self):
        return datetime.now().year - self.birthday.year
 
    @myage.setter
    def myage(self, value):
        self._age_set = value
 
 
if __name__ == "__main__":
    user1 = User("jack", date(1982, 3, 21))
    # 给@age.setter下的myage函数设置value值
    user1.myage = 30
    # 得到设置的value值
    print(user1._age_set)
    # 下面代码得到的是@property下的age
    print(user1.myage)
输出结果如下
30
36
 
说明:
1) _age_set中前面的单下划线只是一种编程规范,表示这个变量不想对外暴漏。但并不能真正做到隐藏,实际上还是可以通过user1._age_set访问到;只有双下划线__age才能隐藏
2)@property以及@myage.setter下的函数名myage必须相同,@myage.setter中的myage也必须和函数名myage相同
 
 
 
 
 
二. __getattr__和__getattribute
 
1. __getattr__ :在查找不到属性的时候,python就会调用这个魔法函数,可避免出现报错信息
from datetime import date
 
class User:
    def __init__(self, name, birthday):
        self.name = name
        self.birthday = birthday
 
    def __getattr__(self, item):
        return "not find attr"
 
if __name__ == "__main__":
    user1 = User("jack", date(1982, 3, 21))
    print(user1.age)
输出结果如下
not find attr
如果不加__getattr__函数,就会报找不到age属性的错误
 
 
1. 2 我们也可以在__getattr__中添加自己的逻辑,改写上例,添加一个参数info, 类型为字典,使对象可访问字典中的键得到相应的值
from datetime import date
 
class User:
    def __init__(self, name, birthday, info={}):
        self.name = name
        self.birthday = birthday
        self.info = info
 
    def __getattr__(self, item):
        return self.info[item]
 
if __name__ == "__main__":
    user1 = User("jack", date(1982, 3, 21), info={"nickname": "monkey", "gender": "male"})
    print(user1.nickname)
输出结果为monkey
 
 
2. __getattribute__ :对象只要调用属性,无论能否找到这个属性,都会先调用这个魔法函数,比__getattr__的优先级高
在上面代码不变的基础上添加一个__getattribute__函数
from datetime import date
 
class User:
    def __init__(self, name, birthday, info={}):
        self.name = name
        self.birthday = birthday
        self.info = info
 
    def __getattr__(self, item):
        return self.info[item]
 
    def __getattribute__(self, item):
        return "hong"
 
if __name__ == "__main__":
    user1 = User("jack", date(1982, 3, 21), info={"nickname": "monkey", "gender": "male"})
    print(user1.nickname)
输出结果为hong
说明:这个魔法函数最好别随便重写,不然会造成整个属性系统崩溃,写框架的时候用比较好
 
 
 
 
 
三. 属性描述符和属性查找过程
 
例子1,需求:验证属性类型,如果类型正确,保存起来;否则,报自定义错误信息
import numbers
 
class IntField:
    # 数据描述符可按照自定义的逻辑来检查对象
    def __get__(self, instance, owner):
        return self.value
 
    def __set__(self, instance, value):
        if not isinstance(value, numbers.Integral):
            raise ValueError("int value need")
        if value < 0:
            raise ValueError("positive value need")
        self.value = value
 
    def __delete__(self, instance):
        pass
 
class User:
    age = IntField() #这里的age是一个属性描述符的对象
 
if __name__ == "__main__":
    user = User()
    user.age = 30  #调用上面的__set__函数
    print(user.age) #调用上面的__get__函数
 
输出30
 
说明:
1) __get__函数中的对象属性value需要和__set__函数中的对象属性value相同
2) 如果user.age = "abc",就会报"int value need"的错误信息
3) numbers库中的Integral类可用来判断,对象是否为int类型
4) 要把一个类变成属性描述符,只需要在其中加入上面的三个魔法函数的任一个就可以了
 
 
1. 2 上面的IntField是数据非数据属性描述符如下,格式上的区别是只有一个__get__魔法函数,类的名字随意
class NonDataIntField:
    #非数据属性描述符
    def __get__(self, instance, owner):
        return self.value
 
 
2. 属性查找过程
如果user是某个类的实例,那么user.age(以及等价的getattr(user,’age’))首先调用__getattribute__。
如果类定义了__getattr__方法,那么在__getattribute__抛出 AttributeError 的时候就会调用到__getattr__,
而对于描述符(__get__)的调用,则是发生在__getattribute__内部的。
user = User(), 那么user.age 顺序如下:
1)如果“age”是出现在User或其基类的__dict__中, 且age是data descriptor, 那么调用其__get__方法(类的静态函数、类函数、普通函数、全局变量以及一些内置的属性都是放在类__dict__里的)
2)如果“age”出现在user对象的__dict__中, 那么直接返回 obj.__dict__[‘age’]
3)如果“age”出现在User或其基类的__dict__中,并且age是non-data descriptor,那么调用其__get__方法, 否则返回 __dict__[‘age’](例如age=1)
4)如果User有__getattr__方法,调用__getattr__方法,
5)  上面都不符合的话,抛出AttributeError
 
 
 
 
四. __init__和__new__的区别
 
1) __new__允许在生成类的对象之前添加逻辑,它传进来的参数cls表示类本身,可自定义类的生成过程
2) __init__传进来的参数self表示类的对象本身,是用来完善对象的
3) 调用__new__函数生成对象之后,并且__new__方法中要返回对象,才会调用__init__函数
class User:
 
    def __init__(self, name):
        self.name = name
        print("in init")
 
    def __new__(cls, *args, **kwargs):
        print("in new")
        #return super().__new__(cls)
 
if __name__ == "__main__":
    user = User(name="jack")
输出结果如下
in new
return super().__new__(cls)会返回一个对象,取消前面的注释后,输出如下
in new
in init
上例表明会先执行__new__魔法函数,并且没有返回对象时,不会调用__init__函数;
 
 
 
 
 
五.自定义元类
 
1. 一个简单的例子
# 类也是对象,这里函数中返回一个类,type创建类的类
def create_class(name):
    if name == "user":
        class User:
            def __str__(self):
                return "user"
        return User
 
    elif name == "company":
        class Company:
            def __str__(self):
                return "company"
        return Company
 
if __name__ == "__main__":
    MyClass = create_class("user")
    my_obj = MyClass()
    print(my_obj)
返回结果为user
 
 
2. type也可以动态创建类,定义类里面的变量,函数和继承其他类的简单例子如下
def say(self):
    # return self.name
    return "i am user"
 
class BaseClass:
    def answer(self):
        return "i am baseclass"
 
USER = type("User", (BaseClass, ), {"name": "jack", "age": "14", "say": say})
my_obj = USER()
 
print(type(my_obj))
print(my_obj.age)
print(my_obj.say())
print(my_obj.answer())
输出结果如下
<class '__main__.User'>
14
i am user
i am baseclass
说明:
1) 在type函数中,第一个参数User定义了要创建类的名字,第2个参数()表示要继承的类,如果为空则继承object类;第三个参数{}表示类的属性或方法
2) 在参数中写入方法时,只能写方法名,后面不能加()
3) type方法定义一个类继承另外一个类时,在父类名后面一定要加逗号
 
 
3. 什么是元类? 
元类是创建类的类,比如type,常见的用法是定义一个类来继承type,那么这个新定义的类就是元类
 
python中类的实例化过程:
首先找自定义的metaclass, 通过metaclass来创建类对象,如果没有metaclass, 则会使用type来创建类对象
class MetaClass(type):
    def __new__(cls, *args, **kwargs):
        return super().__new__(cls, *args, **kwargs)
 
 
class User(metaclass=MetaClass):
    def __init__(self, name):
        self.name = name
 
    def __str__(self):
        return "user"
 
if __name__ == "__main__":
    my_obj = User(name="jack")
    print(my_obj)
输出为user
 
 
 
 
六,通过元类实现orm,先了解下代码吧,以后在细看
# 需求
import numbers
 
class Field:
    pass
 
class IntField(Field):
    # 数据描述符
    def __init__(self, db_column, min_value=None, max_value=None):
        self._value = None
        self.min_value = min_value
        self.max_value = max_value
        self.db_column = db_column
        if min_value is not None:
            if not isinstance(min_value, numbers.Integral):
                raise ValueError("min_value must be int")
            elif min_value < 0:
                raise ValueError("min_value must be positive int")
        if max_value is not None:
            if not isinstance(max_value, numbers.Integral):
                raise ValueError("max_value must be int")
            elif max_value < 0:
                raise ValueError("max_value must be positive int")
        if min_value is not None and max_value is not None:
            if min_value > max_value:
                raise ValueError("min_value must be smaller than max_value")
 
    def __get__(self, instance, owner):
        return self._value
 
    def __set__(self, instance, value):
        if not isinstance(value, numbers.Integral):
            raise ValueError("int value need")
        if value < self.min_value or value > self.max_value:
            raise ValueError("value must between min_value and max_value")
        self._value = value
 
 
 
class CharField(Field):
    def __init__(self, db_column, max_length=None):
        self._value = None
        self.db_column = db_column
        if max_length is None:
            raise ValueError("you must spcify max_lenth for charfiled")
        self.max_length = max_length
 
    def __get__(self, instance, owner):
        return self._value
 
    def __set__(self, instance, value):
        if not isinstance(value, str):
            raise ValueError("string value need")
        if len(value) > self.max_length:
            raise ValueError("value len excess len of max_length")
        self._value = value
 
 
# 定义元类
class ModelMetaClass(type):
    def __new__(cls, name, bases, attrs, **kwargs):
        if name == "BaseModel":
            return super().__new__(cls, name, bases, attrs, **kwargs)
        fields = {}
        for key, value in attrs.items():
            if isinstance(value, Field):
                fields[key] = value
        attrs_meta = attrs.get("Meta", None)
        _meta = {}
        db_table = name.lower()
        if attrs_meta is not None:
            table = getattr(attrs_meta, "db_table", None)
            if table is not None:
                db_table = table
        _meta["db_table"] = db_table
        attrs["_meta"] = _meta
        attrs["fields"] = fields
        del attrs["Meta"]
        return super().__new__(cls, name, bases, attrs, **kwargs)
 
 
class BaseModel(metaclass=ModelMetaClass):
    def __init__(self, *args, **kwargs):
        for key, value in kwargs.items():
            setattr(self, key, value)
        return super().__init__()
 
    def save(self):
        fields = []
        values = []
        for key, value in self.fields.items():
            db_column = value.db_column
            if db_column is None:
                db_column = key.lower()
            fields.append(db_column)
            value = getattr(self, key)
            values.append(str(value))
 
        sql = "insert {db_table}({fields}) value({values})".format(db_table=self._meta["db_table"],
                                                                   fields=",".join(fields), values=",".join(values))
        pass
 
 
 
 
class User(BaseModel):
    name = CharField(db_column="name", max_length=10)
    age = IntField(db_column="age", min_value=1, max_value=100)
 
    class Meta:
        db_table = "user"
 
 
if __name__ == "__main__":
    user = User(name="bobby", age=28)
    # user.name = "bobby"
    # user.age = 28
    user.save()
 
 
 
1. 当参数中有可变参数和默认参数时,要把可变参数写在前面
2. django中,db_column为数据表中的列名,如果不写的话,列名为前面定义的变量名
posted @ 2018-10-31 09:40  坚强的小蚂蚁  阅读(203)  评论(0编辑  收藏  举报