How to remove keys-values from dictionary 1 which are not in dictionary 2 based on common keys?-CodePudding

I have two large dictionaries and both dictionaries have same keys, (name of images) and have different values.

1st dict named train_descriptions which looks like this:

{'15970.jpg': 'Turtle Check Men Navy Blue Shirt',
 '39386.jpg': 'Peter England Men Party Blue Jeans',
 '59263.jpg': 'Titan Women Silver Watch',
 ....
 ....
 '1855.jpg': 'Inkfruit Mens Chain Reaction T-shirt'}

and a 2nd dict named train_features

{'31973.jpg': array([[0.00125694, 0.        , 0.03409385, ..., 0.00434341, 0.00728011,
         0.01451511]], dtype=float32),
 '30778.jpg': array([[0.0174035 , 0.04345186, 0.00772929, ..., 0.02230316, 0.        ,
         0.03104496]], dtype=float32),
 ...,
 ...,

 '38246.jpg': array([[0.00403965, 0.03701203, 0.02616892, ..., 0.02296285, 0.00930257,
         0.04575242]], dtype=float32)}

The length of both dictionaries are as follows:

len(train_descriptions) is 44424 and len(train_features) is 44441

As you can see length of train_description dict is less than length of train_features. train_features dictionary has more keys-values than train_descriptions. How do I remove the keys from train_features dictionary which are not in train_description? To make their length same.

CodePudding user response：

Using for loop

feat = train_features.keys()
desc = train_description.keys()
common = list(i for i in feat if i not in decc)

for i in common: del train_features[i]

Edit: See below

Above code works. But we can do this more efficiently by not converting dict_keys to list as follows:

for i in train_features.keys() - train_description.keys(): del train_features[i]

When python dict_keys are subtracted then it gives dict_keys of uncommon keys. First code was first converting into list which was neither efficient nor required.

CodePudding user response：

just pop() if not exist in another dict.

for key in list(train_descriptions.keys()):
    if key not in train_features.keys():
        train_features.pop(key)

CodePudding user response：

Use xor to get the difference between the dictionaries

diff = train_features.keys() ^ train_descriptions.keys()
for k in diff:
    del train_features[k]