一、封装(属性/私有方法/公有方法/静态方法/构造函数...)
# 定义一个类 class Animal: # 私有成员(用_开头的约定为私有成员 - 注:仅仅是君子协定) _age = 0 # 构造函数 def __init__(self, name): # 建议所有私有成员在这里初始化(不管是已定义的,还是"动态"添加的) self.name = name # 动态添加的私有成员 self._age = 1 # 已经定义好的私有成员 # 静态方法 @staticmethod def eat(): print("Animal can eat food") # 公有方法 def to_string(self): return "name:" + self.name # 私有方法(用_开头的约定为私有方法 - 注:仅仅是君子协定) def _some_method(self): return "this is a private method of :" + self.name # setter示例 def set_age(self, age): if age < 0 or age > 100: print("age is invalid!") self._age = age # getter示例 def get_age(self): return self._age def run(self): print(self._some_method()) print("I am running") def test(self): print("test1") # 注:这样会覆盖上面的test(self版本) def test(self, hello): print("test2:" + hello) animal = Animal("A new animal") print(animal.to_string()) # 方法调用 animal.eat() # 静态方法调用1(不推荐) Animal.eat() # 静态方法调用2 print("animal._age=" + str(animal.get_age())) animal.set_age(10) # 调用setter print(animal.get_age()) # 调用getter,输出10 animal.run() # 公有方法里调用私有方法 animal._age = 30 # 直接修改私有成员(不推荐) animal._age2 = 40 # 注意:这里实际上给实例animal"动态"添加了一个_age2属性 print(animal.get_age()) # 这里输出的仍然是10 print("_age2:" + str(animal._age2)) # 这里输出的动态添加的_age2属性值 print(animal.test("hi")) #print(animal.test()) # 这里会报错,因为test(self,hello)这个版本要求hello参数有值 print(animal._some_method()) # 直接调用私有方法(不推荐)
输出:
name:A new animal Animal can eat food Animal can eat food animal._age=1 10 this is a private method of :A new animal I am running 30 _age2:40 test2:hi None this is a private method of :A new animal
几个要点:
1、约定大于配置:比如构造函数约定为__init__;私有方法/成员约定为"_"开头(注:只是君子协定,硬要调用的话,外部是可以直接调用的);实例方法约定第1个参数为self
2、重载的处理,不象java里要定义多个方法签名,python里就一个版本,但是通过可变参数来实现(后面还要细讲)
3、动态语言,实例在运行过程中,可随时动态添加属性
另外注意:不要轻易使用__(二个连续的下划线)做为方法或变量的前缀或后缀,"__XXX__"在python里通常有特定含义,如果使用"__"开头的变量或写法,可能会遇到各种奇葩问题。
二、重载
传统的OOP语言,比如java,重载只能是定义多个不同方法签名的method,来实现重载,但是python可以做得更简单,用参数默认值就可以变相实现。
# 定义一个类 class Calculator: # 加法 def add(self, a=0, b=0, c=0, d=0): print("args:", a, b, c, d) return int(a) + int(b) + int(c) + int(d) cal = Calculator() print(cal.add(1, 2)) print(cal.add(1, 2, 3)) print(cal.add(1, "2", 3, "4")) print(" ") cal = Calculator #注意这里,如果漏了(),结果会大不同 print(cal.add(1, 2))
输出:
args: 1 2 0 0 3 args: 1 2 3 0 6 args: 1 2 3 4 10 args: 2 0 0 0 2
注意:16行,如果漏了(),cal = Calculator,实际上只是cal只是相当于类Calculator的别名,这样调用add(1,2)时,类似于静态方法调用,1会认为是第一个参数self,所以输出就成了2.
如果不确定参数个数,还可以这么做:
# 定义一个类 class Calculator: # 约定:*参数,表示参数个数不限定 def add(self, *args): result = 0 for arg in args: result += int(arg) return result cal = Calculator() print(cal.add(1, 2, 3)) print(cal.add("1", 2, "3"))
输出:
6 6
三、继承
3.1 基本示例
class Fruit: def __init__(self, name): print("Fruit constructor...") self.name = name def to_string(self): print("Fruit to_string...") return "name:" + self.name # 抽象方法 def get_color(self): print("Fruit get_color...") raise NotImplementedError class RedApple(Fruit): def __init__(self, name): print("Apple constructor...") # 调用父类的构造函数 Fruit.__init__(self, name) def get_color(self): return self.name + " is red" fruit = Fruit("unknown") print(fruit.to_string()) # print(fruit.get_color()) # 报错,因为没实现 print(" ") redApple = RedApple("red apple") print(redApple.get_color()) print(redApple.to_string()) print(" ") print("1、redApple is instance of RedApple ? ", isinstance(redApple, RedApple)) print("2、redApple is instance of Fruit ? ", isinstance(redApple, Fruit)) print("3、fruit is instance of Fruit ? ", isinstance(fruit, Fruit)) print("4、fruit is instance of RedApple ? ", isinstance(fruit, RedApple)) print("5、RedApple is subclass of Fruit ? ", issubclass(RedApple, Fruit)) print("6、Fruit is subclass of Fruit ? ", issubclass(Fruit, Fruit)) print("7、Fruit is subclass of RedApple ? ", issubclass(Fruit, RedApple))
输出:
Fruit constructor... Fruit to_string... name:unknown Apple constructor... Fruit constructor... red apple is red Fruit to_string... name:red apple 1、redApple is instance of RedApple ? True 2、redApple is instance of Fruit ? True 3、fruit is instance of Fruit ? True 4、fruit is instance of RedApple ? False 5、RedApple is subclass of Fruit ? True 6、Fruit is subclass of Fruit ? True 7、Fruit is subclass of RedApple ? False
注:抽象方法是通过抛出未实现的异常来实现的。如果想类似java定义抽象类,把__init__方法抛出未实现异常就行。
3.2 多继承
python支持多继承,这点与java有很大区别
class P1: def a(self): print("P1-a") def b(self): print("P1-b") def x(self): print("P1-x") class P2: def a(self): print("P2-a") def b(self): print("P2-b") def y(self): print("P2-y") # 多继承示例 class S1(P1, P2): def a(self): print("S1-a") class S2(P2, P1): def a(self): print("S2-a") s1 = S1() s1.a() s1.b() # P1-b s1.x() s1.y() print(" ") s2 = S2() s2.a() s2.b() # P2-b s2.x() s2.y() print(" ") print("s1 isinstance of P1:", isinstance(s1, P1)) print("s1 isinstance of P2:", isinstance(s1, P2)) print("s1 isinstance of S1:", isinstance(s1, S1)) print("s1 isinstance of S2:", isinstance(s1, S2))
输出:
S1-a P1-b P1-x P2-y S2-a P2-b P1-x P2-y s1 isinstance of P1: True s1 isinstance of P2: True s1 isinstance of S1: True s1 isinstance of S2: False
注意多承继的顺序,如果“爸爸们”之间有重名方法,将会按照继承顺序,谁在前面,就调用谁的。eg:def S(A,B) 子类S继承自A,B,如果A,B中都有方法x,调用S.x时,因为A排在B的前面,所以调用到的就是A.x方法。从上面的输出就可以得到印证。
3.3 接口、abc模块、属性
3.1中抽象类/方法是通过抛出异常来实现的,有点粗暴,下面介绍一种更优雅的方法,python内置的abc模块
from abc import * # 接口示例 class IRun(ABC): # 抽象方法 @abstractmethod def run(self): pass class Animal(ABC): # 抽象属性 @property @abstractmethod def name(self): pass class Dog(Animal, IRun): def __init__(self, name): self._name = name; # 属性的getter @property def name(self): return self._name # 属性的setter (注:每个property都会生成一个对应的@xxx.setter) @name.setter def name(self, value): self._name = value # 接口的方法实现 def run(self): print(self.name + " is running") dog1 = Dog("a") dog1.run() dog2 = Dog("b") dog2.run() print(isinstance(dog1, IRun)) print(id(dog1), id(dog2), id(dog1) == id(dog2)) # 判断2个实例是否相等
输出:
a is running b is running True 4314753736 4314753792 False
最后送一波福利: https://github.com/faif/python-patterns 这是python实现的所有设计模式,对oop感兴趣的推荐研究。