Question

I noticed that a shell script variable can be used inside an awk script like this:

var="help"
awk 'BEGIN{print "'$var'" }'

Can anyone tell me how to change the value of var inside awk while retaining the value outside of awk?

Similarly to accessing a variable of shell script inside awk, can we access shell array inside awk? If so, how?

Answer 1

It is impossible; the only variants you have:

• use command substitution and write output of awk to the variable;
• write data to file and then read from the outer shell;
• produce shell output and then execute it with eval.

Examples.

Command substitution, one variable:

$ export A=10
$ A=$(awk 'END {print 2*ENVIRON["A"]}' < /dev/null)
$ echo $A
20

Here you multiple A by two and write the result of multiplication back.

eval; two variables:

$ A=10
$ B=10
$ eval $(awk 'END {print "A="2*ENVIRON["A"]"; B="2*ENVIRON["B"]}' < /dev/null)
$ echo $A
20
$ echo $B
20
$ awk 'END {print "A="2*ENVIRON["A"]"; B="2*ENVIRON["B"]}' < /dev/null
A=40; B=40