Remove more than one key from Python dict

Use the del statement:

x = {'a': 5, 'b': 2, 'c': 3}
del x['a'], x['b']
print x
{'c': 3}

The general form I use is this:

1. Produce a list of keys to delete from the mapping;
2. Loop over the list and call del for each.

Example:

Say I want to delete all the string keys in a mapping. Produce a list of them:

>>> x={'a':5,'b':2,'c':3,1:'abc',2:'efg',3:'xyz'}
>>> [k for k in x if type(k) == str]
['a', 'c', 'b']

Now I can delete those:

>>> for key in [k for k in x if type(k) == str]: del x[key]
>>> x
{1: 'abc', 2: 'efg', 3: 'xyz'}

Remove a number of keys

I have tested the performance of three methods:

d = dict.fromkeys('abcdefghijklmnopqrstuvwxyz')
remove_keys = set('abcdef')

# Method 1
for key in remove_keys:
    del d[key]

# Method 2
for key in remove_keys:
    d.pop(key)

# Method 3
{key: v for key, v in d.items() if key no in remove_keys}

Here are the results of 1M iterations:

1. 1.88s 1.9 ns/iter (100%)
2. 2.41s 2.4 ns/iter (128%)
3. 4.15s 4.2 ns/iter (221%)

So del is the fastest.

Remove a number of keys safely

However, if you want to delete safely, so that it does not fail with KeyError, you have to modify the code:

# Method 1
for key in remove_keys:
    if key in d:
        del d[key]

# Method 2
for key in remove_keys:
    d.pop(key, None)

# Method 3
{key: v for key, v in d.items() if key no in remove_keys}

1. 2.03s 2.0 ns/iter (100%)
2. 2.38s 2.4 ns/iter (117%)
3. 4.11s 4.1 ns/iter (202%)

Still, del is the fastest.