Hi,

I hav a log files with a bunch of lines more or less as follows:

/var/log/radius/radiusd-inner-tunnel-peap-20090812.log:Wed Aug 12 11:30:31 2009 : Auth: Login OK: [pepe] (from client Sarlanga-PVIII-I-bis port 0 via TLS tunnel)

/var/log/radius/radiusd-inner-tunnel-peap-20090812.log:Wed Aug 12 11:30:32 2009 : Auth: Login OK: [pepe2] (from client Sarlanga-PVIII-I port 0 via TLS tunnel)

I'd want to get only those lines that has Sarlanga-PVIII-I but *not* Sarlanga-PVIII-I-bis


If I use '-w' grep flag is not useful because it considers "-" as part of the word. How can I prevent it?

Thanks in advance!!

Please bear in mind that use something as grep -w Sarlanga-PVIII-I| grep "I-bis" is not what I am looking for because I am using really in a for loop, something like:

for i in SS PB I I-bis II III IV V VI VII VIII IX
do echo -e PVIII-$i="$(grep OK /var/log/radius/radiusd-*-$(date -d 'yesterday' +%Y%m%d).log | grep -w PVIII-$i[A-Za-z]* | egrep -v 'anonymous | awk '{print $10}' | sort -fu | wc -l)\n"
done

So I am evaluating in turn PVIII-I and PVIII-bis

Thanks in advance

Slightly more robust is:

Even this fails if <pattern> is followed by a newline

Brute force: