Remove space from dictionary keys

Question

I have a dictionary with keys/values such as:

{'C  14': ['15263808', '13210478'], 'W   1': ['13122205']}

How can I remove the white space from all of the keys so they instead would be:

{'C14': ['15263808', '13210478'], 'W1': ['13122205']}

Do you want to edit the existing dictionary or create a new dictionary? Does anything need to be done about colliding keys? — Jared Goguen, Mar 02 '16 at 21:55
All of the keys are unique, ideally the spaces would be removed from the existing dictionary keys. — GreyTek, Mar 02 '16 at 21:58

score 12 · Accepted Answer · edited May 23 '17 at 11:46

12

You can process all keys in a dictionary comprehension; you could use str.translate() to remove the spaces; the Python 2 version for that is:

{k.translate(None, ' '): v for k, v in dictionary.iteritems()}

and the Python 3 version is:

{k.translate({32: None}): v for k, v in dictionary.items()}

You could also use str.replace() to remove the spaces:

{k.replace(' ', ''): v for k, v in dictionary.items()}

But this may be slower for longer keys; see Python str.translate VS str.replace

Note that these produce a new dictionary.

Demo on Python 2:

>>> d = {'C  14': ['15263808', '13210478'], 'W   1': ['13122205']}
>>> {k.translate(None, ' '): v for k, v in d.iteritems()}
{'W1': ['13122205'], 'C14': ['15263808', '13210478']}

And a 2.6 compatible version (using Alternative to dict comprehension prior to Python 2.7):

dict((k.translate(None, ' '), v) for k, v in dictionary.iteritems())

edited May 23 '17 at 11:46

Community

1
1

answered Mar 02 '16 at 21:36

Martijn Pieters

1,048,767
296
4,058
3,343

What about `iteritems()` for the Python 2 version? – pp_ Mar 02 '16 at 21:41
This is on Python 2.6.6 so dictionary comprehension isn't working – GreyTek Mar 02 '16 at 21:52
@GreyTek: then use [Alternative to dict comprehension prior to Python 2.7](https://stackoverflow.com/a/21069676) – Martijn Pieters Mar 02 '16 at 21:56

Padraic Cunningham · Answer 2 · 2016-03-02T22:10:09.793

4

You can also just use str.replace:

In [4]: d = {'C  14': ['15263808', '13210478'], 'W   1': ['13122205']}

In [5]: d = {k.replace(" ",""): v for k,v in d.items()}

In [6]: d
Out[6]: {'C14': ['15263808', '13210478'], 'W1': ['13122205']}

For python2.6:

d = dict((k.replace(" ",""),v)  for k,v in d.items())

edited Mar 02 '16 at 22:10

answered Mar 02 '16 at 21:42

Padraic Cunningham

176,452
29
245
321

This is on Python 2.6.6 so dictionary comprehension isn't working – GreyTek Mar 02 '16 at 21:54
@GreyTek just replace it with a gen exp and a dict call – Padraic Cunningham Mar 02 '16 at 22:11

Vader · Answer 3 · 2016-03-02T22:42:11.497

1

Without dict comprehension:

>>> d = {'C  14': ['15263808', '13210478'], 'W   1': ['13122205']}
>>> dict(((k.replace(' ',''),v) for k,v in d.items()))
{'W1': ['13122205'], 'C14': ['15263808', '13210478']}

edited Mar 02 '16 at 22:42

answered Mar 02 '16 at 21:59

Vader

3,675
23
40

Drop the square brackets; no point in building a list first where a generator expression will do. – Martijn Pieters Mar 02 '16 at 22:37

Jared Goguen · Answer 4 · 2016-03-02T22:50:03.057

1

Here's a solution without a dict comprehension, that edits the dictionary in-place, and assumes that no two old keys will map to the same new key.

d = {'C  14': ['15263808', '13210478'], 'W   1': ['13122205']}

for key in d.keys():
    if ' ' in key:
        d[key.translate(None, ' ')] = d[key]
        del d[key]

print d # {'W1': ['13122205'], 'C14': ['15263808', '13210478']}

After some timing, this is about 25% quicker than creating a new dictionary.

The loop iterates over d.keys() rather than d directly to avoid revisiting keys and avoid the whole "iterating over a changing object" issue.

edited Mar 02 '16 at 22:50

answered Mar 02 '16 at 22:03

Jared Goguen

8,772
2
18
36

Translate should actually be faster; perhaps I need to retime for single character replacements though. – Martijn Pieters Mar 02 '16 at 22:38
@StevenRumbalski I fixed the deleting key issue, the author is using Python 2.6, so no need to consider Python 3.X. – Jared Goguen Mar 02 '16 at 22:40
See http://stackoverflow.com/questions/31143290/python-str-translate-vs-str-replace – Martijn Pieters Mar 02 '16 at 22:45
@MartijnPieters From my tests, `replace` was running 1-2% faster than `translate` for a single character replacement in this scenario; good to know that in general `translate` will be faster. – Jared Goguen Mar 02 '16 at 22:46
Are your timsings still accurate with the updated code? Can you share your benchmark? – Martijn Pieters Mar 02 '16 at 22:46
@MartijnPieters No, they are not still accurate, now `translate` is running about 3X faster than `replace`. – Jared Goguen Mar 02 '16 at 22:50

Remove space from dictionary keys

4 Answers4