bash help with file input to sed and awk command?

[bmanzeck]

Hello,
I have a script that works pretty good except one part where I try to use a file with one item on each line as an input to sed. Can someone tell me why this bash script is not functioning. It is something with my while loop that is not working I think.

sed 's/ //g' $input >$output
sed 's/filter3/filter2/g' $output >$output2

cat $output2 |while read line
do
sed '/$line/d' $input2 >diditwork.txt
done

Many Thanks in Advance.
Bryan

[secondmouse]

Change

sed '/$line/d' $input2 >diditwork.txt

to

CMD="sed '/$line/d' $input2 >diditwork.txt"
eval $CMD

[lomcevak]

A couple of things.

First, by using single quotes the $line won't be replaced by its value. Use double quotes.

Second, redirecting the sed output with '>' will create a new file every time sed is executed. May be you want to append to the file with '>>'?

sed "/$line/d" $input2 >> diditwork.txt

[bmanzeck]

secondmouse,lomcevak,
I tried the double quotes and the eval CMD suggestions and that gave me the same result.

Maybe I can tell you what I am trying to do and you can suggest the best course of action?

I have a file:
george=filter2
george=filter3
fred=filter2
fred=filter3
phil=filter2
phil=filter3
bill=filter3
mark=filter2

I have a second file with the users who are the duplicate entries I need deleted
george=filter2
phil=filter2


So, there you have it, I thought I was on the right track....
Anyone's help is surely appreciated

[bmanzeck]

I tried this suggestion from Andrew on another forum and it worked for me!

Your problem is on the sed line in your while loop..."$line" is inside
of single quotes, which means sed is searching for the literal string
"$line" rather than the variable you intend to be placed there. You
could try using double-quotes instead, or replace that command with:
cat >> tmpfile <<HERE
/$line/d
HERE
sed -f tmpfile $input2 > diditwork.txt

I know that's really ugly, but it should work.
-Andrew

--
Andrew Vandever