将可变变量传递给第二个变量并不总是能按预期工作。 为什么？ 有没有变通方法来保留原始变量引用？

Question

Passing a mutable variable to a second variable doesn't always work as expected. Why? Is there a workaround to keep the original variable reference?

Example:

>>> listA = [10,11,12,13,14]

# this works as I would expect it:
>>> ref = listA
>>> ref[0] = 20
>>> listA
[20, 11, 12, 13, 14]

# but this doesn't
>>> ref = listA[1:3]    # still making reference to a mutable (only part of it)
>>> ref
[11, 12]
>>> ref[0]=30
>>> listA
[20, 11, 12, 13, 14]

Answer 1

Your assumption in the comment:

# still making reference to a mutable (only part of it)

Is wrong .

When you use slicing on list s , you create a shallow copy , not a view . So those are two independent lists that operate differently. Therefore changes to ref will not reflect on listA and vice versa.

Note that the elements in the list , are still the same. So the memory will look like:

         +-------------------+         +-------+
listA -> |       list        |  ref -> |  list |
         +---+---+---+---+---+         +---+---+
         | o | o | o | o | o |         | o | o |
         +-|-+-|-+-|-+-|-+-|-+         +-|-+-|-+
           v   |   |   v   v             |   |
           20  |   |   13  14            |   |
               |   v                     |   |
               v   12 <------------------|---/
               11 <----------------------/

Here that does not matter, since int s are immutable. But if you would for instance call a method on an element of listA that changes the state of that object, the change will also reflect when you refer to that element through ref[0] for instance.

This behavior is not always the case, for instance numpy arrays use views when slicing is done. For instance:

>>> import numpy as np
>>> a = np.arange(10)
>>> b = a[1:3] # slicing in numpy generates a "view"
>>> a
array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
>>> b
array([1, 2])
>>> b[0] = 5
>>> a
array([0, 5, 2, 3, 4, 5, 6, 7, 8, 9])
>>> b
array([5, 2])

You should always consult the documentation, like here for numpy arrays:

(...)

All arrays generated by basic slicing are always views of the original array.

(...)

Creating a listview

You could construct a view on a list, by defining a class like:

class listview:

    def __init__(self,data,frm=None,len=None):
        self.data = data
        self.frm = frm
        self.len = len

    def __getitem__(self,idx):
        if self.frm is not None:
            idx += self.frm
        if self.len is not None and idx >= self.len:
            raise Exception('Index out of range!')
        return self.data[idx]

    def __setitem__(self,idx,value):
        if self.frm is not None:
            idx += self.frm
        if self.len is not None and idx >= self.len:
            raise Exception('Index out of range!')
        self.data[idx] = value

    def __len__(self):
        frm = 0
        if self.frm is not None:
            frm = self.frm
        if self.len is not None:
            return min(self.len,len(self.data)-frm)
        return len(self.data)

    def __repr__(self):
        return 'listview(%s,%s,%s)'%(self.data,self.frm,self.len)

    # ... and so on

Then you can construct a view like:

>>> listA = [10,11,12,13,14]
>>> ref = listview(listA,1,2)
>>> ref[0] = 5
>>> listA
[10, 5, 12, 13, 14]