Shell Programming and Scripting

I am working on a script to install a software distribution which ships as a *.tar.gz file. The company doesn't name the dist consistently, but the resulting directory out of the *.tar.gz file (ie: XXX_X_X_X) is consistent and shows the software version.

I'm writing this with BASH-3.25.x (RHEL).

I'm trying to determine how to reliably grab that root directory name for use inside a script.

I have hacked this:

Code:

tar ztvf $DISTFILE | head -1 | sed -e "s/.*F/F/g" | cut -f1 -d"/"

(the dir always starts with F)

This is a crappy hack, but it works. I looked at the manpage for pax and tar and I didn't see anything obvious that accomplishes what I require.

Anyone have a more elegant/reliable way to do this?


Thanks.

Show the input data it gives and the output data you want.

Alright...

tar ztvf:

drwxr-xr-x 0/0 0 2011-07-26 17:20:44 FMS_4_0_3_r4011/conf/_defaultRoot_/
drwxr-xr-x 0/0 0 2011-07-26 16:51:28 FMS_4_0_3_r4011/conf/_defaultRoot_/_defaultVHost_/
-rw-r--r-- 0/0 43801 2011-07-26 16:51:28 FMS_4_0_3_r4011/conf/Logger.xml
-rw-r--r-- 0/0 46191 2011-07-26 16:51:28 FMS_4_0_3_r4011/conf/Server.xml

I only want:

FMS_4_0_3_r4011

So whatever release it is, I need to grab that name as a variable for installation work. Their installation stuff is crappy.

You don't need the -v if you don't need the details, and you don't.

Code:

BASE=$(tar -ztf filename.tar.gz | awk -F"/" '{ print $1 ; exit }' )

Not the most efficient but given we have to run tar anyway...

Thanks, that is a much more elegant version of my messy solution above.

I need to learn awk/perl. Been putting it off too long.