topic Re: shell script in Operating System - HP-UX

shell script

Mark McDonald_2 — Fri, 19 Jan 2007 09:41:49 GMT

Hi all

I have a tarball of files per machine for about 1500 machines. I need a script to read each tarball and extract certain info from files within the tarball. I need this info to be put in a comma separated flat file so it can be read in to Excel.

so far I have
cd /export/home/Explorers
ls *.gz | while read FILE
do
cd /export/home/Explorers
# copy file to temp area
cp $FILE ./temporary
cd temporary

# extract the files
gunzip $FILE
tar xf *.tar
cd explorer.*
#extract the info
grep Hostname README
grep Hostid README
grep Release README
grep Company README
grep Kernel README
grep architecture README
grep Date README
grep Uptime README
#more info to be added
#clean up - remove contents of temp before next explorer
cd ..
rm -r /export/home/SUN/Explorers/temporary/*

done

The grep statements bring out the lines I require, but for each grep I need the data after the colon to be added to a csv file, and a new line to be started for each separate tar ball.

example output of the greps:
Hostname: ipcsoiwa
Hostid: 8075ac8c
Release: 5.6
Kernel architecture: sun4d
Kernel version: SunOS 5.6 Generic 105181-06
Kernel architecture: sun4d
Application architecture: sparc
Date: 2005.05.30.04.54
Uptime: 6:02am up 43 min(s), 2 users, load average: 1.34, 1.16, 0.66

Thanks

Mark

Re: shell script

Patrick Wallek — Fri, 19 Jan 2007 09:57:53 GMT

Well, this is probably ugly, but it should work:

HOST=$(grep Hostname README | awk -F : '{ print $2}')
HOSTID=$(grep Hostid README | awk -F : '{ print $2}')
RELEASE=$(grep Release README | awk -F : '{ print $2}')
COMPANY=$(grep Company README | awk -F : '{ print $2}')
KERNEL=$(grep Kernel README | awk -F : '{ print $2}')
ARCH=$(grep architecture README | awk -F : '{ print $2}')
DATE=$(grep Date README | awk -F : '{ print $2}')
UPTIME=$(grep Uptime README | awk -F : '{ print $2}')

echo "${HOSTNAME},${HOSTID},${RELEASE},${KERNEL},${ARCH},${DATE},${UPTIME}" >> /dir/afile

Another slight improvement is you can combine your gunzip and tar commands into a single line.

Instead of:

gunzip $FILE
tar xf *.tar

Try this:

gzcat $FILE | tar xf -

(Note there is a space between the 'xf' and the '-')

Re: shell script

Patrick Wallek — Fri, 19 Jan 2007 10:01:22 GMT

I just noticed one thing that could lead to some difficulty.

You have this grep statement:

grep Kernel README

which will return 2 lines of data:

Kernel architecture: sun4d
Kernel version: SunOS 5.6 Generic 105181-06

Then you also grep for architecture. Which will return the Kernel architecture line again. For those you should probably do this:

KERNEL=$(grep "Kernel version" README | awk -F : '{ print $2}')
ARCH=$(grep "Kernel architecture" README | awk -F : '{ print $2}')

This way you should only get one line per grep.

Re: shell script

Mark McDonald_2 — Fri, 19 Jan 2007 10:15:56 GMT

Thanks

Re: shell script

Peter Nikitka — Fri, 19 Jan 2007 11:46:08 GMT

Hi,

it's time for awk, IMHO. I see no info for 'Company', so I do not anchor that string and assume the requested value is the last field (like for the other pattern).

awk '/^Hostname:/ {hname=$NF}
/^Hostid:/ {hid=$NF}
/^Release/ {rel=$NF}
/Company/ {com=$NF}
/^Kernel architecture:/ {arch=$NF}
/^Kernel version:/ {ver=$3;for(i=4;i<=NF;i++) ver=ver" "$i }
/^Application architecture:/ {mach=$NF}
/^Date:/ {date=$NF}
/^Uptime:/ {upt=$2;for(i=3;i<=NF;i++) upt=upt" "$i }
END { printf("%s;%s;%s;%s;%s;%s;%s;%s;%s\n",hname,hid,rel,com,arch,ver,mach,date,upt)}' README

I would add some error checking as well, like instead
>>
gunzip $FILE
tar xf *.tar
cd explorer.*
<<
use in KSH
if ! gunzip<$file | tar xf -
then exit 1
fi
cd explorer.*

Re: shell script

Mark McDonald_2 — Thu, 11 Dec 2008 00:17:10 GMT