从内部追加和删除实例到列表

Question

The following code defines a class (Wall) which when it's instantiated the object is added to a list (in_progress) and once it's attribute (progress) reaches 3 it is removed from that list and moved to another one (built).

in_progress = []
built = []

class Wall:
    global in_progress, built

    def __init__(self):
        self.progress = 0
        in_progress.append(self)

    def build(self):
        self.progress += 1

        if self.progress == 3:
            in_progress.remove(self)
            built.append(self)

This is convenient since no matter how many Walls there are in the list "in_progress" I can run:

for wall in in_progress:
    wall.build()

and eventually "in_progress" will be empty. However I've done some tests and something strange happens when an instance in in_progress reaches progress = 3.

For example. Let's instantiate three walls:

Wall()
Wall()
Wall() 

#check in_progress
in_progress
--->
[<__main__.Wall at 0x7f4b84e68cf8>,
 <__main__.Wall at 0x7f4b84e68c50>,
 <__main__.Wall at 0x7f4b84e68f28>]

#check attribute progress

for wall in in_progress:
    print(f'{wall}: {wall.progress}')
--->
<__main__.Wall object at 0x7f4b84e68cf8>: 0
<__main__.Wall object at 0x7f4b84e68c50>: 0
<__main__.Wall object at 0x7f4b84e68f28>: 0

#'build' on them 2 times
for wall in in_progress:
    wall.build()

for wall in in_progress:
    print(f'{wall}: {wall.progress}')
--->
<__main__.Wall object at 0x7f4b84e68cf8>: 2
<__main__.Wall object at 0x7f4b84e68c50>: 2
<__main__.Wall object at 0x7f4b84e68f28>: 2

If we run the last code one more time we expect to find the list in_progress empty but what we find is this:

#'build' on them once more
for wall in in_progress:
    wall.build()

for wall in in_progress:
    print(f'{wall}: {wall.progress}')
--->
<__main__.Wall object at 0x7f4b84e68c50>: 2

If we check the list built we find the 2 walls left but there should be 3. Why does this happen?

Answer 1

The problem in your build function is that you are trying to modify the same list you are iterating on, which is causing this weird issue to happen, try as follows and you should not see the issue. I am copying the list to another variable via copy.copy https://docs.python.org/3/library/copy.html

import copy
in_progress = []
built = []

class Wall:
    global in_progress, built

    def __init__(self):
        self.progress = 0
        in_progress.append(self)

    def build(self):
        global in_progress
        self.progress += 1
        #Make a copy of the list and operate on that
        copy_in_progress = copy.copy(in_progress)
        if self.progress == 3:
            copy_in_progress.remove(self)
            built.append(self)
        in_progress = copy_in_progress

Wall()
Wall()
Wall()

print(in_progress)
#[<__main__.Wall object at 0x108259908>, 
#<__main__.Wall object at 0x108259940>, 
#<__main__.Wall object at 0x1082599e8>]

for wall in in_progress:
    print(f'{wall}: {wall.progress}')

#<__main__.Wall object at 0x108259908>: 0
#<__main__.Wall object at 0x108259940>: 0
#<__main__.Wall object at 0x1082599e8>: 0

for wall in in_progress:
    wall.build()
    wall.build()

for wall in in_progress:
    print(f'{wall}: {wall.progress}')


#<__main__.Wall object at 0x108259908>: 2
#<__main__.Wall object at 0x108259940>: 2
#<__main__.Wall object at 0x1082599e8>: 2
for wall in in_progress:
    wall.build()

for wall in in_progress:
    print(f'{wall}: {wall.progress}')
#Nothing is printed

Answer 2

Traversing a list and altering it during traversal can cause some counterintuitive behavior, such as this. By doing the remove() of the element you are currently positioned at, the list is altered so that the next time through the loop, the next element is one beyond where you think you should be, since the list got shifted back one by the operation of remove() .

>>> q = ['a', 'ab', 'abc', 'again', 'b', 'a1', 'c', 'a2', 'ack']
>>> for pos in q:
...     if pos.startswith('a'):
...             q.remove(pos)
... 
>>> q
['ab', 'again', 'b', 'c', 'ack']

Here, when the first element is removed, the list shifts down so the first element becomes 'ab'. Then at the top of the loop, the "next" element is 'abc', since it is now in the second position, so 'ab' never is tested for removal. Similarly, 'again' and 'ack' don't get removed because they were never tested. In fact, 'b' and 'c' remain in the list not because they do not start with 'a', but they also were never tested, since the list shifted and the loop skipped them over too!

If you iterate over a copy or a slice of your original list, that will probably get you what you need, but be careful about any scenario where you are iterating through something that is being updated at the same time.

>>> q = ['a', 'ab', 'abc', 'again', 'b', 'a1', 'c', 'a2', 'ack']
>>> for pos in q[:]:
...     if pos.startswith('a'):
...             q.remove(pos)
... 
>>> q
['b', 'c']