Using AWK to define variables in a shell

Generally, to pass variables calculated or extracted by awk to the shell, you'd follow this coding patterns...

shell_var=$( awk '{ ...some processing... ; print awk_var }' )

If you have more than one variable you can do...

eval "$( awk '{ ...some processing... ; print "shell_var1="awk_var1, "shell_var2="awk_var2 }' )"

Explanation of the code: The shell pattern var=$( awk '...' ) will assign whatever the awk process will print to the shell variable. Since you can only assign one value by this means you need a different approach for two or more variable assignments. One approach is to let awk print the whole assignment, say awk '{ print "varname=" value }'; the shell construct $( ... ) will again be expanded to the output of the embedded awk call, in this case an assignment. And that expanded assignment shell command to become active you need the shell instruct to evaluate it; eval is the command that does this. The difference to the first (simple) case is that you can let awk print several assignments (separated by whitespace) and eval all in one go.

I'll give an example with an input and an output variable and some calculation...

number=42
square=$( awk -v NUM="$number" 'BEGIN{ print NUM*NUM }' )
echo "$square"

Hope that clarifies the parameter passing in both directions.