regex to catch substring in a line and change output with sed - regex

Question

I am trying to put together a regex that will find a substring that has the following format:
[a-z].* ('x' number of lowercase letters)
A '-' sign [a-z].*
('x' number of lower case letters [0-9].*
('x' number of numbers from 0 to 9)
Anything else in the line that follows this substring (including space or ',') would not be caught by the regex and then I would add a new line to the results so they are in a list.
If this regex works the way I would like then from the following string
file.txt: hostname abcd-efg123, zfdh-eif23 , reox-bmo552, 'coor-dto201',
I would receive this output
abcd-efg123
zfdh-eif23
reox-bmo552
coor-dto201
This is what I have so far. I'm trying to use the regex and then store the result as as two variables which I can then put back into sed. I'm not getting the results I expected.
The regex/sed I am using is
sed 's/\([a-z].*\)-\([a-z].*[0-9].*\)/\2 \1 \n/g'
Here is the command straight from the prompt
macbook:~ user$ echo "file.txt: hostname abcd-efg123, zfdh-eif23 , reox-bmo552, 'coor-dto201'," | sed 's/\([a-z].*\)-\([a-z].*[0-9].*\)/\2 \1 \n/g'
dto201', file.txt: hostname abcd-efg123, zfdh-eif23 , reox-bmo552, 'coor n

Answer 1

Here's the regex I would use to match :
[a-z]*-[a-z]*[0-9]*
The main problem in your regex was the use of .* where you obviously meant *. As I commented, * is the quantifier that means "any number of times (including 0)", while . is the "any character" wildcard. You want to apply the quantifier to your previous character class rather than to ., the . has no reason to be here.
Note that using * includes 0 repetition, so the regex would match a single dash, which might not be to your taste.
Maybe you could be more specific, with a regex along those lines :
[a-z]{4}-[a-z]{3}[0-9]{2,3}
Here instead of using * as a quantifier, we use numbers between curly brackets : they give us the possibility to specify an exact number of repetitions (i.e. .{4} means "any 4 characters") or a range of repetitions (i.e. [0-9]{2,6} means "2 to 6 digits"). You could also use +, a quantifier that means "at least one time", as mentioned by Kenavoz.
And here's how I would use it in a linux command :
grep -o '[a-z]*-[a-z]*[0-9]*'
or
grep -Eo '[a-z]{4}-[a-z]{3}[0-9]{2,3}'
Here it is in action :
$ echo "file.txt: hostname abcd-efg123, zfdh-eif23 , reox-bmo552, 'coor-dto201'," | grep -o '[a-z]*-[a-z]*[0-9]*'
abcd-efg123
zfdh-eif23
reox-bmo552
coor-dto201
Or with the more specific regex :
$ echo "file.txt: hostname abcd-efg123, zfdh-eif23 , reox-bmo552, 'coor-dto201'," | grep -Eo "[a-z]{4}-[a-z]{3}[0-9]{2,3}"
abcd-efg123
zfdh-eif23
reox-bmo552
coor-dto201