Hi all, i have encountered a strange problem with the double while loop. The aim of the two while loops is to compare 2 files, tmplist.txt is the reference file with the complete list of servers. Please see the attached picture for the logic and relationship between input and output.

The 2 while loops should search the servers in /test.csv in the /tmplist.txt, line by line. For example, search first item in /test.csv server2 in /tmplist.txt, starting from first line in /tmplist.txt (LINENUM=1), if not found then output a 'NA' and then go to next line in /tmplist.txt until it finds the matching server2, and update LINENUM at the same time. Else will proceed to search the next server server3 in /tmplist.txt from the current line onwards (since all servers are always listed in alphabatic order).

The variable LINENUM (indicator of line number) used in the inner while loop. It is predefined as 1, after the if condition if [ $SERVER = $REFS ] is true, LINENUM will increment and then break out the inner while loop and continue in the outer while loop. However, in the outer while loop, every time it loops and reads SERVER COUNT, the LINENUM starts from the initial value 1 again instead of from the incremented LINENUM coming from the inner while loop.


Below shows part of the program with the details of 2 while loops.
Could anyone help to explain to me what is wrong here? This is my first UNIX assignment, will really appreciate your help!!


#!/bin/bash
TYPE=P2PS
LINENUM=1

while read SERVER COUNT; do

cat tmplist.txt | while read LINE; do

REFS=$(head -$LINENUM tmplist.txt | tail -1)

if [ $SERVER = $REFS ]; then
let LINENUM++
echo -n ","$COUNT >>t.csv
break
else
let LINENUM++
echo -n ",NA" >> t.csv
continue
fi
done

done < test.csv

Attached Images

untitled.bmp (177.0 KB, 13 views)

The inner loop is a sub-process. Anything changed there has no effect on the parent process. It's the same situation as

X=1
bash -c "X=2"
echo $X

Both of your loops are reading from stdin. Each process only has 1 stdin so your two loops absolutely need to be separate processes.

There is no need for 2 loops, you have to use powerful unix commands in your scripts, like grep. This should work:
Of course, t.csv must be empty before executing the script.

while read LINE
do
aux=$(grep $LINE test.csv)
if [[ ! -z $aux ]]
then
echo "$aux" >> t.csv
else
echo "NA" >> t.csv
fi
done <tmplist.txt

the read in the inner loop is a subprocess. All variable values set or changed inside the inner loop are not known outside of it. If the files are small try to exchange the while-construct against a for-loop. If not, store meta-informations in the inner loop into files to have them outside. But this can decrease the speed very much.
But actually I do not see the need for double loops. Why not grep each line from tmplist.txt on the first file?

A solution with awk:

Thank you so much for your replies! Really powerful commands!
My script works now. Thanks all!