SED Help

[Jason Dixon]

On May 16, 2004, at 10:07 AM, Mike Vanecek wrote:

> Given this text in packet.test2
>
> May 16 21:35:35 www kernel: icmp_try IN=eth0 OUT=
> MAC=00:d0:09:3d:69:81:00:04:5a:ef:5e:1d:08:00  SRC=144.232.20.162
> DST=192.168.1.95 LEN=56 TOS=0x00 PREC=0xC0 TTL=241 ID=57923 PROTO=ICMP 
> TYPE=11
>  CODE=0 [SRC=66.76.12.5 DST=200.216.94.217 LEN=40 TOS=0x00 PREC=0x00 
> TTL=1
> ID=25653 PROTO=TCP INCOMPLETE [8 bytes] ]
> May 16 21:54:39 www kernel: icmp_try IN=eth0 OUT=
> MAC=00:d0:09:3d:69:81:00:04:5a:ef:5e:1d:08:00 SRC=144.232.7.98
> DST=192.168.1.95 LEN=56 TOS=0x00 PREC=0x00 TTL=239 ID=0 PROTO=ICMP 
> TYPE=11
> CODE=0 [SRC=66.76.12.5 DST=200.222.69.36 LEN=40 TOS=0x00 PREC=0x00 
> TTL=1
> ID=62986 PROTO=TCP INCOMPLETE [8 bytes] ]
>
> Why does this command
>
> $ grep -i "`date '+%b %_d'`" packet.test2  | sed -re
> 's/.*SRC=([0-9]+\.[0-9]+\.[0-9]+\.[0-9]).*/\1/'
> 66.76.12.5
> 66.76.12.5
>
> pick up the second SRC rather than the first?

I'm not a sed expert, but I'm guessing it's behaving "greedy", similar 
to perl regex.  The first SRC actually falls within the ".*" portion of 
your match, and the last match is what counts.

--
Jason Dixon, RHCE
DixonGroup Consulting
http://www.dixongroup.net