Question

I’m trying to extract fields from a pipe-delimited file and provide them as arguments to an external program in a loop. The file contains lines like this:

value1|value2
value3|value4

So I came up with:

 while read line;

    do echo -n "${line}" | awk -F '|' '{print $1}';
       echo -n " something ";
       echo -n "${line}" | awk -F '|' '{print $2}';
       echo " somethingelse";

 done < <(cat $FILE)

I want to see the following output:

value1 something value2 somethingelse
value3 something value4 somethingelse

But instead I’m getting:

value1 
 something value2
 somethingelse
value3
 something value4
 somethingelse

Perhaps I shouldn’t be using echo?

Answer 1

If your awk supports printf, I would use that instead. Even though echo -n doesn’t add a newline, awk {print} does. The printf variant will not put the newline at the end like print does:

echo -n "${line}" | awk -F '|' '{printf "%s", $1}'

Here’s your version hex-dumped, note the \n at the end when you pass it through awk:

pax> echo -n "hello|goodbye" | od -xcb
0000000    6568    6c6c    7c6f    6f67    646f    7962    0065
          h   e   l   l   o   |   g   o   o   d   b   y   e
        150 145 154 154 157 174 147 157 157 144 142 171 145
0000015

pax> echo -n "hello|goodbye" | awk -F '|' '{print $1}' | od -xcb
0000000    6568    6c6c    0a6f
          h   e   l   l   o  \n
        150 145 154 154 157 012

and here’s the printf version:

pax> echo -n "hello|goodbye" | awk -F '|' '{printf "%s", $1}' | od -xcb
0000000    6568    6c6c    006f
          h   e   l   l   o
        150 145 154 154 157

And here it is, running on your real file (named FILE in this example to ease my testing):

pax> while read line;
     do
         echo -n "${line}" | awk -F '|' '{printf "%s", $1}';
         echo -n " something ";
         echo -n "${line}" | awk -F '|' '{printf "%s", $2}';
         echo " somethingelse";
     done < <(cat FILE)

value1 something value2 somethingelse
value3 something value4 somethingelse