Shell Programming and Scripting

Fellows I ran in to a problem this morning and could not figure out a solution to it.
I have a file like

OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
DESCRIPTION="ABC MNCL JHDG "
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "


What I am trying to do is whenever there is a repition of pattern DESCRIPTION consecutively I want to delete all the lines but first.
On the outpur file i want to add consecutive lines to one line.

I couldn't figure out a way to do the first step.
But for secod i was trying this script

#! /bin/sh
awk 'BEGIN {
RS="";
FS="\n";
}
{ print $1 $2 }' filename

give the output record too long

Any thought why is it doing it more over Can i use paste command to do the second step??

See if this works for you:

$ more dup_test
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
DESCRIPTION="ABC MNCL JHDG "
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
$ sed -e '$!N; /^\(.*\)\n\1$/!P; D' dup_test
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU"
DESCRIPTION="ABC MNCL JHDG "

The reason why you get "line too long" is because when you unset the record separator like this: RS="" then you have told awk to separate records with blank lines. If there are no blank lines then the whole file is treated as a single record.

Try...

Code:

awk 'BEGIN {
  FS="=";
}
$1 != prev
{prev = $1}' filename | paste -d ' ' - -

Tested on the sample data...

OBJECT="ABC" GFT="JHU" DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU" DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU" DESCRIPTION="ABC MNCL JHDG "
OBJECT="ABC" GFT="JHU" DESCRIPTION="ABC MNCL JHDG "