I have the following variables :
list_m = ["a","b","c"]
list_s = ['x','y','z']
dict_m = dict.fromkeys(list_m[:])
dict_s = dict.fromkeys(list_s[:],copy.deepcopy(dict_m)) # empty dict of dicts
So I have
In[22]: dict_s
Out[22]:
{'x': {'a': None, 'b': None, 'c': None},
'y': {'a': None, 'b': None, 'c': None},
'z': {'a': None, 'b': None, 'c': None}}
On updating a value of dict_s like this
dict_s['x']['a']= np.arange(10)
I get
In[27]: dict_s
Out[27]:
{'x': {'a': array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]), 'b': None, 'c': None},
'y': {'a': array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]), 'b': None, 'c': None},
'z': {'a': array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]), 'b': None, 'c': None}}
instead of what i wanted/expected:
In[27]: dict_s
Out[27]:
{'x': {'a': array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]), 'b': None, 'c': None},
'y': {'a': None, 'b': None, 'c': None},
'z': {'a': None, 'b': None, 'c': None}}
I don't exactly understand if this is a deep/shallow copy issue or something else.
fromkeys
uses the same default value for each key. If you want separate values you can use dict comprehension and generate new dict for each value with fromkeys
:
>>> list_m = ["a","b","c"]
>>> list_s = ['x','y','z']
>>> dict_s = {x: dict.fromkeys(list_m) for x in list_s}
>>> dict_s
{'y': {'a': None, 'c': None, 'b': None}, 'x': {'a': None, 'c': None, 'b': None}, 'z': {'a': None, 'c': None, 'b': None}}
>>> dict_s['y']['a'] = 100
>>> dict_s
{'y': {'a': 100, 'c': None, 'b': None}, 'x': {'a': None, 'c': None, 'b': None}, 'z': {'a': None, 'c': None, 'b': None}}