Extracting Substring with SED

[chuckp123]

I have a text file(replicateTotals.txt) with a bunch of lines in this form:

2009/10/21-12:15:58 avtar Stats <6559>: 111.5 KB in 1,838 messages sent to server, 4.541 MB in 1,832 messages received from server.
2009/10/21-19:44:40 avtar Stats <6559>: 22.88 GB in 2,781,764 messages sent to server, 157.2 MB in 2,779,968 messages received from server.
2009/10/21-19:44:40 avtar Stats <6559>: 72.55 MB in 1,338,246 messages sent to server, 22.80 GB in 1,336,448 messages received from server.

What I would like to do is extract just the data amount which appears right after the '<6559>'. So, the output I would want from the lines above would be:

111.5 KB
22.88 GB
72.55 MB

I'm not very experienced with sed, but here's the command that I came up with to grab this data:

# cat replicateTotals.txt | grep \<6559\> | sed 's/.*\<6559\>\: \([[:digit:]]*\.[[:digit:]]*(KB|MB|GB)\).*/\1/'

This command is not working. Can anyone tell me where I'm going wrong?

[lomcevak]

grep '<6559>' replicateTotals.txt | sed -r 's/.*<6559>: ([[:digit:]]*.[[:digit:]]* (KB|MB|GB)).*/\1/'

Several problems:

When you have a string within single quotes you don't have to escape characters. They \'s were being searched for as part of the string.

You were missing a space between [[:digit:]]*.[[:digit:]]* and (KB|MB|GB).

Use of (KB|MB|GB) requires extended regular expressions so you have to use the -r option.

[chuckp123]

You rock!!

Thanks, man.