74
I usually start with grep, to get the regexp right.
# [multiple failed attempts here]
grep '[0-9]*\.[0-9]*\.[0-9]*\.[0-9]*' file # good?
grep -E '[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}' file # good enough
Then I'd try and convert it to sed to filter out the rest of the line. (After reading this thread, you and I aren't going to do that anymore: we're going to use grep -o instead)
sed -ne 's/.*\([0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\).*/\1/p # FAIL
That's when I usually get annoyed with sed for not using the same regexes as anyone else. So I move to perl.
$ perl -nle '/[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}/ and print $&'
Perl's good to know in any case. If you've got a teeny bit of CPAN installed, you can even make it more reliable at little cost:
$ perl -MRegexp::Common=net -nE '/$RE{net}{IPV4}/ and say $&' file(s)
I wrote a little script to see my log files better, it's nothing special, but might help a lot of the people who are learning perl. It does DNS lookups on the IP addresses after it extracts them.
I'd suggest perl. (\d+.\d+.\d+.\d+) should probably do the trick.
EDIT: Just to make it more like a complete program, you could do something like the following (not tested):
#!/usr/bin/perl -w
use strict;
while (<>) {
if (/(\d+\.\d+\.\d+\.\d+)/) {
print "$1\n";
}
}
This handles one IP per line. If you have more than one IPs per line, you need to use the /g option. man perlretut gives you a more detailed tutorial on regular expressions.