Problem description:
I want to print only the source and destination address from a tcpdump[1].
Have one working solution, but believe it could be improved a lot. An example that captures 5 packets, just as an example of what I'm looking for:
tcpdump -i eth1 -n -c 5 ip | \
cut -d" " -f3,5 | \
sed -e 's/^\([0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\)\..* \([0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\).*$/\1 > \2/'
Question:
Can this be done in any easier way? Performance is also an issue here.
[1] A part of a test if the snort home_net is correctly defined, or if we see traffic not defined in the home_net.
Solution:
Ok, thanks to everyone who have replied to this one. There have been two concerns related to the answers, one is the compatibility across different linux-versions and the second one is speed.
Here is the results on the speed test I did. First the grep-version:
time tcpdump -l -r test.dmp -n ip 2>/dev/null | grep -P -o '([0-9]+\.[0-9]+\.[0-9]+\.[0-9]+).*? > ([0-9]+\.[0-9]+\.[0-9]+\.[0-9]+)' | grep -P -o '[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+' | xargs -n 2 echo >/dev/null
real 0m5.625s
user 0m0.513s
sys 0m4.305s
Then the sed-version:
time tcpdump -n -r test.dmp ip | sed -une 's/^.* \(\([0-9]\{1,3\}\.\?\)\{4\}\)\..* \(\([0-9]\{1,3\}\.\?\)\{4\}\)\..*$/\1 > \3/p' >/dev/null
reading from file test.dmp, link-type EN10MB (Ethernet)
real 0m0.491s
user 0m0.496s
sys 0m0.020s
And the fastest one, the awk-version:
time tcpdump -l -r test.dmp -n ip | awk '{ print gensub(/(.*)\..*/,"\\1","g",$3), $4, gensub(/(.*)\..*/,"\\1","g",$5) }' >/dev/null
reading from file test.dmp, link-type EN10MB (Ethernet)
real 0m0.093s
user 0m0.111s
sys 0m0.013s
Unfortunately I have not been able to test how compatible they are, but the awk needs gnu awk to work due to the gensub function. Anyway, all three solutions works on the two platforms I have tested them on. :)
Here's one way using GNU awk
:
tcpdump -i eth1 -n -c 5 ip | awk '{ print gensub(/(.*)\..*/,"\\1","g",$3), $4, gensub(/(.*)\..*/,"\\1","g",$5) }'