How to strip all whitespace from string

How do I strip all the spaces in a python string? For example, I want a string like strip my spaces to be turned into stripmyspaces, but I cannot seem to accomplish that with strip():

>>> 'strip my spaces'.strip()
'strip my spaces'

Solution 1:

Taking advantage of str.split's behavior with no sep parameter:

>>> s = " \t foo \n bar "
>>> "".join(s.split())
'foobar'

If you just want to remove spaces instead of all whitespace:

>>> s.replace(" ", "")
'\tfoo\nbar'

Premature optimization

Even though efficiency isn't the primary goal—writing clear code is—here are some initial timings:

$ python -m timeit '"".join(" \t foo \n bar ".split())'
1000000 loops, best of 3: 1.38 usec per loop
$ python -m timeit -s 'import re' 're.sub(r"\s+", "", " \t foo \n bar ")'
100000 loops, best of 3: 15.6 usec per loop

Note the regex is cached, so it's not as slow as you'd imagine. Compiling it beforehand helps some, but would only matter in practice if you call this many times:

$ python -m timeit -s 'import re; e = re.compile(r"\s+")' 'e.sub("", " \t foo \n bar ")'
100000 loops, best of 3: 7.76 usec per loop

Even though re.sub is 11.3x slower, remember your bottlenecks are assuredly elsewhere. Most programs would not notice the difference between any of these 3 choices.

Solution 2:

For Python 3:

>>> import re
>>> re.sub(r'\s+', '', 'strip my \n\t\r ASCII and \u00A0 \u2003 Unicode spaces')
'stripmyASCIIandUnicodespaces'
>>> # Or, depending on the situation:
>>> re.sub(r'(\s|\u180B|\u200B|\u200C|\u200D|\u2060|\uFEFF)+', '', \
... '\uFEFF\t\t\t strip all \u000A kinds of \u200B whitespace \n')
'stripallkindsofwhitespace'

...handles any whitespace characters that you're not thinking of - and believe us, there are plenty.

\s on its own always covers the ASCII whitespace:

  • (regular) space
  • tab
  • new line (\n)
  • carriage return (\r)
  • form feed
  • vertical tab

Additionally:

  • for Python 2 with re.UNICODE enabled,
  • for Python 3 without any extra actions,

...\s also covers the Unicode whitespace characters, for example:

  • non-breaking space,
  • em space,
  • ideographic space,

...etc. See the full list here, under "Unicode characters with White_Space property".

However \s DOES NOT cover characters not classified as whitespace, which are de facto whitespace, such as among others:

  • zero-width joiner,
  • Mongolian vowel separator,
  • zero-width non-breaking space (a.k.a. byte order mark),

...etc. See the full list here, under "Related Unicode characters without White_Space property".

So these 6 characters are covered by the list in the second regex, \u180B|\u200B|\u200C|\u200D|\u2060|\uFEFF.

Sources:

  • https://docs.python.org/2/library/re.html
  • https://docs.python.org/3/library/re.html
  • https://en.wikipedia.org/wiki/Unicode_character_property