Question

I read this about Python classes (link) and it seems to be the issue I am having.

Here is an excerpt from my class and other code:

class s_board:

    def __init__(self):
        self.__board = [[n for n in range(1, 10)] for m in range(81)]
        self.__solved = [False for m in range(81)]

    def copy(self):
        b = s_board()
        b.__board = self.__board[:]
        b.__solved = self.__solved[:]
        return b

if __name__ == '__main__':
    A = s_board()
    B = A.copy()
    B.do_some_operation_on_lists()

When I call B’s method that does something to the list, A’s lists seem to be affected as well.

So my questions:

• Am I not copying the class or the lists correctly?
• Is there another issue here?
• How do I fix it so that I get a new copy of the class?

Answer 1

self.__board[:] creates a new list containing references to all the same objects that were in self.__board. Since self.__board contains lists, and lists are mutable, you end up with the two s_board instances with partially aliased data, and changing one affects the other.

As Raymond Hettinger suggested, you can use the copy.deepcopy to (mostly) guarantee that you take a true copy of an object and don’t share any data. I say mostly, as I believe there are some strange objects that deepcopy will not work on, but for normal things like lists and straightforward classes it will work fine.

I have an additional suggestion though. You call b = s_board(), which goes to all the effort of constructing the lists for the new blank board, and then you throw them away by assigning to b.__board and b.__solved. It seems to be like it would be better to do something like the following:

class s_board:

    def __init__(self, board=None, solved=None):
        if board is None:
            self.__board = [[n for n in range(1, 10)] for m in range(81)]
        else:
            self.__board = copy.deepcopy(board)
        if solved is None:
            self.__solved = [False for m in range(81)]
        else:
            self.__solved = copy.deepcopy(solved)

    def copy(self):
        b = s_board(self.__board, self.__solved)
        return b

Now if you call A = s_board() you get a new blank board, and if you call A.copy() you get a distinct copy of A, without having had to allocate and then discard a new blank board.