Shell Programming and Scripting

I have a two files
>cat file1
jjjjj 10.345 6.673
ppp 9.000 5.883

>cat file2
mmm 80 10
jjjjj 10.305 6.873
ppp 9.000 5.883

I am reading file 1 line by line , and look for the string jjjj and then read the line in file 2 with jjjj
I want to get the flaoting point difference for colum2 and column 3 of file1 and file2, if both has the same column1.

I have the snippet code like this
while read line
do
name=`echo $line | awk ''{ print $1 }`
matchline=`grep -m 1 -e"$name" file2`
f1col2=`echo $line | awk ''{ print $2 }`
f2col2=`echo $matchline | awk ''{ print $2 }`
diffcol2=$(( $f1col2 - $f2col2 ))
echo $diffcol2
done < file1

However I get error at the line diffcol2=$(( $f1col2 - $f2col2 )),
10.345 - 10.305 : syntax error in expression (error token is ".10.345 - 10.305 ")

Do I ahve to do anything special for floating point substarction .
I tried bracing the variable like the following , but still di not work
diffcol2=$(( ${f1col2} - ${f2col2} ))

I got it, it look like bash does not support floating point arithmetics

I have to use awk, like the following

diff=`echo $f1col1 $f2col2 | awk '{print $1-$2}'

Hi.

Good work -- if one tool doesn't work, find another.

For small pieces of arithmetic, you can use bc:

Code:

#!/bin/sh

# @(#) s1       Demonstrate floating point arithmetic with bc.

set -o nounset
echo
echo "GNU bash $BASH_VERSION" >&2
bc --version | head -1 >&2
echo

echo 'scale=3 ; 10.345 - 10.305' |
bc

exit 0

which produces:

Code:

% ./s1

GNU bash 2.05b.0(1)-release
bc 1.06

.040

See man bc (and relative dc) for details ... cheers, drl

wouldn't it be easier to do it with 'awk':

Code:

nawk '
  FNR==NR { arr[$1]=$2; next }
  $1 in arr { printf("%s %f\n", arr[$1] - $2) }
' file1 file2

Thansk , But I also have to print the originla values , with diffeences 0, incase I did not find a matching line.

ok, so modify the proposed solution.