The reason you need to use self. is because Python does not use special syntax to refer to instance attributes. Python decided to do methods in a way that makes the instance to which the method belongs be passed automatically, but not received automatically: the first parameter of methods is the instance the method is called on. That makes methods entirely the same as functions, and leaves the actual name to use up to you (although self is the convention, and people will generally frown at you when you use something else.) self is not special to the code, it's just another object.

Python could have done something else to distinguish normal names from attributes -- special syntax like Ruby has, or requiring declarations like C++ and Java do, or perhaps something yet more different -- but it didn't. Python's all for making things explicit, making it obvious what's what, and although it doesn't do it entirely everywhere, it does do it for instance attributes. That's why assigning to an instance attribute needs to know what instance to assign to, and that's why it needs self..

当对象实例化时，对象本身被传递到self参数中。

因此，对象的数据被绑定到对象上。下面是一个示例，说明您可能希望如何可视化每个对象的数据。注意“self”是如何被对象名称替换的。我并不是说下面这个示例图是完全准确的，但希望它能在可视化self的使用方面起到一定的作用。

对象被传递到self参数中，这样对象就可以保存自己的数据。

Although this may not be wholly accurate, think of the process of instantiating an object like this: When an object is made it uses the class as a template for its own data and methods. Without passing it's own name into the self parameter, the attributes and methods in the class would remain as a general template and would not be referenced to (belong to) the object. So by passing the object's name into the self parameter it means that if 100 objects are instantiated from the one class, they can all keep track of their own data and methods.

如下图所示:

我将用不使用类的代码演示:

def state_init(state):
    state['field'] = 'init'

def state_add(state, x):
    state['field'] += x

def state_mult(state, x):
    state['field'] *= x

def state_getField(state):
    return state['field']

myself = {}
state_init(myself)
state_add(myself, 'added')
state_mult(myself, 2)

print( state_getField(myself) )
#--> 'initaddedinitadded'

类只是一种避免始终传递这种“状态”的方法(以及其他一些不错的事情，如初始化、类组合、很少需要的元类，以及支持自定义方法来覆盖操作符)。

现在让我们使用内置的python类机制来演示上面的代码，以展示它们基本上是相同的东西。

class State(object):
    def __init__(self):
        self.field = 'init'
    def add(self, x):
        self.field += x
    def mult(self, x):
        self.field *= x

s = State()
s.add('added')    # self is implicitly passed in
s.mult(2)         # self is implicitly passed in
print( s.field )

[把我的答案从重复的封闭式问题中迁移过来]

以下摘自关于self的Python文档:

As in Modula-3, there are no shorthands [in Python] for referencing the object’s members from its methods: the method function is declared with an explicit first argument representing the object, which is provided implicitly by the call. Often, the first argument of a method is called self. This is nothing more than a convention: the name self has absolutely no special meaning to Python. Note, however, that by not following the convention your code may be less readable to other Python programmers, and it is also conceivable that a class browser program might be written that relies upon such a convention.

有关更多信息，请参阅Python类文档教程。

因为按照python的设计，其他的选择几乎行不通。Python被设计为允许在隐式this (a-la Java/ c++)或显式@ (a-la ruby)都不能工作的上下文中定义方法或函数。让我们看一个带有python约定的显式方法的例子:

def fubar(x):
    self.x = x

class C:
    frob = fubar

现在fubar函数不能工作了，因为它假定self是一个全局变量(在frob中也是如此)。另一种方法是使用替换的全局作用域(其中self是对象)执行方法。

隐式方法是

def fubar(x)
    myX = x

class C:
    frob = fubar

这意味着myX将被解释为fubar(以及frob)中的局部变量。这里的替代方案是使用替换的局部作用域执行方法，该作用域在调用之间保留，但这将消除方法局部变量的可能性。

然而，目前的情况很好:

 def fubar(self, x)
     self.x = x

 class C:
     frob = fubar

在这里，当作为方法调用时，frob将通过self参数接收它所调用的对象，fubar仍然可以以对象作为参数调用并且工作相同(我认为它与C.frob相同)。

让我们看一个简单的向量类:

class Vector:
    def __init__(self, x, y):
        self.x = x
        self.y = y

我们需要一个计算长度的方法。如果我们想在类中定义它会是什么样子呢?

    def length(self):
        return math.sqrt(self.x ** 2 + self.y ** 2)

当我们将它定义为全局方法/函数时，它应该是什么样子?

def length_global(vector):
    return math.sqrt(vector.x ** 2 + vector.y ** 2)

所以整个结构保持不变。我怎样才能利用这个呢?假设我们还没有为Vector类编写length方法，我们可以这样做:

Vector.length_new = length_global
v = Vector(3, 4)
print(v.length_new()) # 5.0

这是因为length_global的第一个参数可以重用为length_new中的self参数。如果没有显式的自我，这是不可能的。

另一种理解显式self需求的方法是查看Python在哪里添加了一些语法糖。当你记住的时候，基本上，一个电话就像

v_instance.length()

在内部转化为

Vector.length(v_instance)

很容易看出自我在哪里。实际上你不用在Python中编写实例方法;你所写的是类方法，它必须以一个实例作为第一个参数。因此，必须显式地将实例参数放置在某个位置。