I got the example idea from a Stack Overflow question where a method for getting the IP address of the primary DNS server was used like this

$ cat /etc/resolv.conf | grep -i '^nameserver' | head -n1 | cut -d ' ' -f2

I the snippet reads pretty clear, but the pipeline is rather long. At first I thought I could at least drop the head part, as I can limit grep amount of matched lines with -m

$ cat /etc/resolv.conf | grep -i -m 1 '^nameserver' | cut -d ' ' -f2

But I think I can do better. grep -P makes use of Perl-compatible regular expression (PCRE) patterns and opens up for use of capture groups with \K. To return only the contents of the capture group, use -o. And I know the nameserver portion is lowercase, so I don't want case-insensitive search; dropping -i

$ cat /etc/resolv.conf | grep -P -o -m 1 '^nameserver \K\S+'

Where the regex pattern reads as follows:

#  `^` start of line
#  `nameserver ` match exact string contents
#  `\K` start capture group
#  `\S+` one or more non-whitespace characters

But.. I think I can do even better. Since the beginning this snippet, it has been subject to a useless use of cat. grep takes a file argument after the pattern

$ grep -Pom 1 '^nameserver \K\S+' /etc/resolv.conf

References

  • man grep
  • https://unix.stackexchange.com/a/192852/28043