How to combine POSIX character classes of a regex?

Question

I need to combine different character classes in a regular expresion used by sed. I need to match [:word:] and the minus symbol -. How does that look like? All my attempts failed at trying or searching a solution.

In the following strings I would like to match everything until the first space:

foo-bar     |
baz-xyz-123 |

@cuonglm see http://www.regular-expressions.info/posixbrackets.html — SpaceTrucker, Aug 20 '15 at 13:31
@SpaceTrucker: I doubt `[:word:]` is supported in all locale. — cuonglm, Aug 20 '15 at 13:40
Possibly relevant for BSD systems: https://unix.stackexchange.com/q/356234/135943 — Wildcard, Sep 10 '17 at 20:08

score 5 · Accepted Answer · answered Aug 20 '15 at 13:44

POSIXly:

sed 's/[^[:alnum:]_-]//g'

will remove everything is not alpha numeric characters in your current locale, _ and -.

$ echo 'foo-bar     |' | sed -e 's/[^[:alnum:]_-]//g'
foo-bar

But if you want to print everything until first space:

sed -e 's/^\([^ ]*\) .*/\1/'

or awk:

awk '{print $1}'

How to combine POSIX character classes of a regex?

1 Answers1