Regex. Match words that contain special characters or 'http://'

Question

I want to match words that contain special characters or that begin with 'http://'

So this sentence

%he#llo, my website is: http://www.url.com/abcdef123

should turn into this

my website

So far, i have this

re.sub(r"^[^\w]", " ", "%he#llo, my website is: http://www.url.com/abcdef123")

This just removes the symbols, but it doesn't remove the words associated with the symbol (it also doesn't remove ':' and ','), nor does it remove the URL.

Answer 1

For the example string you give, the following regular expression works OK:

>>> a = '%he#llo, my website is: http://www.url.com/abcdef123'
>>> re.findall('(http://\S+|\S*[^\w\s]\S*)',a)
['%he#llo,', 'is:', 'http://www.url.com/abcdef123']

... or you can remove those words with re.sub

>>> re.sub('(http://\S+|\S*[^\w\s]\S*)','',a)
' my website  '

The | means alternation and will match the expression on either side within the group. The part on the left matches http:// followed by one or more non-space characters. The part on the right matches zero or more non-space characters, followed by anything that isn't a word or space character, followed by zero or more non-space characters -- that ensures that you have a string with at least one non-word character and no spaces.

Updated: Of course, as the other answers implicitly suggest, since the http:// prefix contains a non-word character (/) you don't need to have that as an alternative - you could simplify the regular expression to \S*[^\w\s]\S*. However, perhaps the example above with alternation is still useful.

Answer 2

You can use look aheads:

>>> re.findall(r"(?:\s|^)(\w+)(?=\s|$)", "Start %he#llo, my website is: http://www.url.comabcdef123 End")
['Start', 'my', 'website', 'End']

Explanation:

• (?:\s|^) means our word starts the regex or is preceeded by a space. (and the space does not belong to the word).
• (\w+) matches a word (and is what we are interested in).
• (?=\s|$) means our word is followed by space or end of the string. (and once again, the space does not belong to the word).