SUMMARY 2: grep for more than single lines

All,

After having it pointed out that this will only work for the first instance
of the string in the file by Dan Astoorian (Something I was already working
on) and thinking it was a bit crap having to edit the script everytime you
wanted to change the string or filename etc, I edited is as follows.  It now
does some fairly basic usage checking, and takes the parameters as command
line arguments, and will print the output for all instances of the string.
I called it 'greb' (Grep for Blocks) and stuck it in /usr/local/bin, so you
can use it just like you would grep (Except don't trying piping output to
it!)

#!/usr/bin/ksh

function usage {
echo "Incorrect arguments"
echo "Usage:   greb -b nn -a nn string file" 
echo "Example: greb -b 10 -a 19 ERROR /tmp/logfile" 
}

function main {
NPRINT=`expr $BEFORE + $AFTER + 1`
for LINENUM in `grep -n $STRING $FILE | awk -F: '{ print $1 }'`
do
        LINESTRT=`expr $LINENUM - $BEFORE | sed 's/^  *//'`
        tail +$LINESTRT $FILE | head -$NPRINT
        echo "--"
done
}

if [[ $1 = "-b" ]] && [[ $3 = "-a" ]]
then
        BEFORE=$2
        AFTER=$4
        STRING=$5
        FILE=$6
        main
else
        usage
        exit
fi

Regards

dAn

-----Original Message-----
From: Wilkinson, Daniel [mailto:Daniel.Wilkinson@capgemini.co.uk]
Sent: 18 June 2001 15:57
To: Sun Managers Mailing List
Subject: SUMMARY: grep for more than single lines


Thanks everyone - I shan't name you all, as I've has 20+ emails in 10
minutes already, and am aleady shamed by my pitifully woeful knowledge,
without having to name those cleverer than I...

The general opinion was to use GNU grep, but I wasn't keen on installing any
more packages etc which is why I was wandering down the script route.  If
you can, or have Gnu grep installed, then do this:

grep -A5 -B5 string file

Another way was with sed:

sed -n -e '/string/{=;x;1!p;g;$!N;p;D;}' -e h <inputfile

to print 1 line before and after, although I didn't play around with it to
do more than one line either way, because this weeks winner is Sarah
Eckhause, with:

	#!/bin/sh
	BEFORE=10
	AFTER=5
	#  Number of lines to print
	NPRINT=`expr $BEFORE + $AFTER + 1`

	# Find line number of the string in question

	LINENO=`grep -n 'string'  FILETOSEARCH | awk -F':' '{ print $1 }' |
head -1`

	LINESTRT=`expr $LINENO - $BEFORE | sed 's/^  *//'`

	tail +$LINESTRT  FILETOSEARCH | head -$NPRINT

Which does exactly what I want, in roundabout the way I was starting to try,
although I had temporarily forgotten about head/tail, and was trying to
trick more into doing something it'll never do...

Thanks Sarah!

Dan



-----Original Message-----
From: Wilkinson, Daniel [mailto:Daniel.Wilkinson@capgemini.co.uk]
Sent: 18 June 2001 14:46
To: Sun Managers Mailing List
Subject: grep for more than single lines


All,

I need a command or script to grep for a string within a file, and then
instead of just printing out that single line, getting it to print the
previous 10 lines, or the following 5 lines, or a combination of both, such
that I can cut out an error message in context, and see what happened before
or after etc.  Am I being stupid, and does this exist as a command in
Solaris, and if not, has anyone written anything already before I start?

Regards

dAn
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers