Python: setting values of a list of nested dictionaries in class instances

Question

I have a burning question that I just can't figure out. Let's say you have a class, and this class takes in as input a list of nested dictionaries. It initializes one of the keys of each of these dictionaries with a new, empty dictionary. Later, I want to set one of the nested values of an object in the list as something. For some reason this seems to affect other objects in the list?

I know that's sounds very convoluted, so here's an example:

class Tester():
    def __init__(self, stuff):

        # stuff is a list of dictionaries
        self.stuff = stuff

        # Each item in the list should be initialized to this dictionary
        inside_init_dict = {'test': True}
        for x in self.stuff:
            x['info'] = inside_init_dict

if __name__ == '__main__':
    new_stuff = [{'info': {}}, {'info': {}}, {'info': {}}]
    mytest = Tester(new_stuff)

    print(mytest.stuff)
    # >>> [{'info': {'test': True}}, {'info': {'test': True}}, {'info': {'test': True}}]

    # I want to just set a value in the nested dict of the first item in the list
    mytest.stuff[0]['info']['test'] = False

    # However, all items in the list change
    print(mytest.stuff)
    # >>> [{'info': {'test': False}}, {'info': {'test': False}}, {'info': {'test': False}}]

This happens on both Python 2 and 3. The only way I can get around this is to not use the separate variable "inside_init_dict", and directly set the initialized dictionary:

class Tester():
    def __init__(self, stuff):

        # stuff is a list of dictionaries
        self.stuff = stuff

        # Each item in the list should be initialized to this dictionary
        for x in self.stuff:
            x['info'] = {'test': True}

if __name__ == '__main__':
    new_stuff = [{'info': {}}, {'info': {}}, {'info': {}}]
    mytest = Tester(new_stuff)

    print(mytest.stuff)
    # >>> [{'info': {'test': True}}, {'info': {'test': True}}, {'info': {'test': True}}]

    mytest.stuff[0]['info']['test'] = False

    # This is what I want
    print(mytest.stuff)
    # >>> [{'info': {'test': False}}, {'info': {'test': True}}, {'info': {'test': True}}]

What's going on here? I have tried setting the variable "inside_init_dict" in various places, like as a class variable or outside the class. The issue still occurs.

score 3 · Answer 1 · answered Oct 26 '16 at 23:57

In the first example you create a single dictionary inside_init_dict outside the loop and put it in multiple places. Every element of the list gets that same inside_init_dict. What you're seeing is not other objects in a list being affected, there's just one object being shown multiple times.

In the second example:

    for x in self.stuff:
        x['info'] = {'test': True}

Now each x gets its own dictionary. They all have the same value at first, but they are different instances, like identical twins.

Awesome, thank you for explaining it that way. Makes sense to me now. — deckardk, Oct 27 '16 at 00:00

score 2 · Answer 2 · edited May 23 '17 at 10:27

This happens because dicts are mutable, meaning that you can change their content without changing their identity. Here's a much simpler example of the behavior you're seeing:

my_dict = { "key" : "value" }
my_list = [ my_dict, my_dict ]
my_list[0]["key"] = "new_value"
print(my_list) # [ {"key" : "new_value"}, {"key": "new_value"} ]

Why this happens:

In the first line of this code, I create a new dictionary, {"key" : "value"}, and assign the name my_dict to it.

In the second line, I create a list, whose zeroth and first element both point to my_dict.

In the third line, I access my_dict (through my_list[0]), and I mutate it: changing the value associated with "key".

In the fourth line, I check the value of my_list. Both the zeroth and the first element of my_list still point to my_dict -- and we've changed my_dict. So the change is reflected in both elements of the list.

One way to fix it:

Instead pointing to the same dictionary twice, create two dictionaries that have the same value:

my_list = [ { "key" : "value" } , { "key" : "value" } ] 
my_list[0]["key"] = "new_value"
print(my_list) # [ {"key" : "new_value"}, {"key": "value"} ]

score 1 · Accepted Answer · answered Oct 26 '16 at 23:54

1

Assign the keys to different copies of the inside_init_dict dictionary instead of the same one:

    ...
    inside_init_dict = {'test': True}
    for x in self.stuff:
        x['info'] = inside_init_dict.copy()

answered Oct 26 '16 at 23:54

Moses Koledoye

77,341
8
133
139

@deckardk If I get what you're asking, you're essentially assigning them to the same object. Updating from one reference changes the object, and the change is reflected across all others. – Moses Koledoye Oct 27 '16 at 00:00

Python: setting values of a list of nested dictionaries in class instances

3 Answers3