Question

Ok, I have a pretty tricky regex problem.

I need to match and replace the + signs in the following strings with whitespace, thus the strings on the left must become the strings on the right.

1: word+word   =  word word
2: word+++word =  word + word
3: word.+word  =  word. word
4: word,+word  =  word, word

I’ve managed to nail most of ’em with this, which checks for strings before and after plus signs:

(?<=[\w\.,])\+(?=[\w])

However, I can only nail one set of the adjacent plus signs in (2) at a time:

(?<=[\w\.,\+])\+(?=[\w]) // gets the left plus sign
2: word+++word =  word++ word

(?<=[\w\.,])\+(?=[\w\+]) // gets the right plus sign
2: word+++word =  word ++word

In my current regex be-frazzled state, I’m wondering if I need to wrap the pattern in a larger condition, or if I need to implement an either or pattern in the lookarounds. Any regex mavens out there care to give this a whirl?

Answer 1

This should work on your examples:

(?<!\+)\+|\+(?!\+)

Replace that with spaces.

Example:

perl -E "$_='a+b++a+++b++++c+++++d'; s/(?<!\+)\+|\+(?!\+)/ /g; say"

Outputs:

a b  a + b ++ c +++ d