[python] Remove all special characters, punctuation and spaces from string

I need to remove all special characters, punctuation and spaces from a string so that I only have letters and numbers.

This will remove all non-alphanumeric characters except spaces.

string = "Special $#! characters   spaces 888323"
''.join(e for e in string if (e.isalnum() or e.isspace()))

Special characters spaces 888323

s = re.sub(r"[-()\"#/@;:<>{}`+=~|.!?,]", "", s)

Python 2.*

I think just filter(str.isalnum, string) works

In [20]: filter(str.isalnum, 'string with special chars like !,#$% etcs.')
Out[20]: 'stringwithspecialcharslikeetcs'

Python 3.*

In Python3, filter( ) function would return an itertable object (instead of string unlike in above). One has to join back to get a string from itertable:

''.join(filter(str.isalnum, string)) 

or to pass list in join use (not sure but can be fast a bit)

''.join([*filter(str.isalnum, string)])

note: unpacking in [*args] valid from Python >= 3.5

Shorter way :

import re
cleanString = re.sub('\W+','', string )

If you want spaces between words and numbers substitute '' with ' '

For other languages like German, Spanish, Danish, French etc that contain special characters (like German "Umlaute" as ü, ä, ö) simply add these to the regex search string:

Example for German:

re.sub('[^A-ZÜÖÄa-z0-9]+', '', mystring)

The most generic approach is using the 'categories' of the unicodedata table which classifies every single character. E.g. the following code filters only printable characters based on their category:

import unicodedata
# strip of crap characters (based on the Unicode database
# categorization:
# http://www.sql-und-xml.de/unicode-database/#kategorien

PRINTABLE = set(('Lu', 'Ll', 'Nd', 'Zs'))

def filter_non_printable(s):
    result = []
    ws_last = False
    for c in s:
        c = unicodedata.category(c) in PRINTABLE and c or u'#'
        result.append(c)
    return u''.join(result).replace(u'#', u' ')

Look at the given URL above for all related categories. You also can of course filter by the punctuation categories.

Removing Punctuations, Numbers, and Special Characters

Example :-

Code

combi['tidy_tweet'] = combi['tidy_tweet'].str.replace("[^a-zA-Z#]", " ") 

Result:-

Thanks :)

#!/usr/bin/python
import re

strs = "how much for the maple syrup? $20.99? That's ricidulous!!!"
print strs
nstr = re.sub(r'[?|$|.|!]',r'',strs)
print nstr
nestr = re.sub(r'[^a-zA-Z0-9 ]',r'',nstr)
print nestr

you can add more special character and that will be replaced by '' means nothing i.e they will be removed.

import re
my_string = """Strings are amongst the most popular data types in Python. We can create the strings by enclosing characters in quotes. Python treats single quotes the 

same as double quotes."""

# if we need to count the word python that ends with or without ',' or '.' at end

count = 0
for i in text:
    if i.endswith("."):
        text[count] = re.sub("^([a-z]+)(.)?$", r"\1", i)
    count += 1
print("The count of Python : ", text.count("python"))

string.punctuation contains following characters:

'!"#$%&\'()*+,-./:;<=>?@[\]^_`{|}~'

You can use translate and maketrans functions to map punctuations to empty values (replace)

import string

'This, is. A test!'.translate(str.maketrans('', '', string.punctuation))

Output:

'This is A test'

Differently than everyone else did using regex, I would try to exclude every character that is not what I want, instead of enumerating explicitly what I don't want.

For example, if I want only characters from 'a to z' (upper and lower case) and numbers, I would exclude everything else:

import re
s = re.sub(r"[^a-zA-Z0-9]","",s)

This means "substitute every character that is not a number, or a character in the range 'a to z' or 'A to Z' with an empty string".

In fact, if you insert the special character ^ at the first place of your regex, you will get the negation.

Extra tip: if you also need to lowercase the result, you can make the regex even faster and easier, as long as you won't find any uppercase now.

import re
s = re.sub(r"[^a-z0-9]","",s.lower())

Use translate:

import string

def clean(instr):
    return instr.translate(None, string.punctuation + ' ')

Caveat: Only works on ascii strings.

Assuming you want to use a regex and you want/need Unicode-cognisant 2.x code that is 2to3-ready:

>>> import re
>>> rx = re.compile(u'[\W_]+', re.UNICODE)
>>> data = u''.join(unichr(i) for i in range(256))
>>> rx.sub(u'', data)
u'0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\xaa\xb2 [snip] \xfe\xff'
>>>

After seeing this, I was interested in expanding on the provided answers by finding out which executes in the least amount of time, so I went through and checked some of the proposed answers with timeit against two of the example strings:

• string1 = 'Special $#! characters spaces 888323'
• string2 = 'how much for the maple syrup? $20.99? That s ricidulous!!!'

Example 1

'.join(e for e in string if e.isalnum())

• string1 - Result: 10.7061979771
• string2 - Result: 7.78372597694

Example 2

import re re.sub('[^A-Za-z0-9]+', '', string)

• string1 - Result: 7.10785102844
• string2 - Result: 4.12814903259

Example 3

import re re.sub('\W+','', string)

• string1 - Result: 3.11899876595
• string2 - Result: 2.78014397621

The above results are a product of the lowest returned result from an average of: repeat(3, 2000000)

Example 3 can be 3x faster than Example 1.

import re
abc = "askhnl#$%askdjalsdk"
ddd = abc.replace("#$%","")
print (ddd)

and you shall see your result as

'askhnlaskdjalsdk

Here is a regex to match a string of characters that are not a letters or numbers:

[^A-Za-z0-9]+

Here is the Python command to do a regex substitution:

re.sub('[^A-Za-z0-9]+', '', mystring)