Import code from file using awk in bash

Question

We'd like to import code between two strings and put them in an variable.

We have:

job=1

listen="$(awk -v a="BEGIN-$job" -v e="END-$job" '/BEGIN {print a}/{p=1;next} /BEGIN {print e}/{p=0;exit} p' whatsthat.file)"

whatsthat.file (edited):

BEGIN-1

echo "some code"

END-1

BEGIN-2

echo "next job"

END-2

Final result for variable $listen should be:

echo "some code"

But the command above leaves $listen empty.

Of course it does not work -- it's a no-sense. What is /BEGIN {print a}/ for? Do you know what /···/ means? — Jdamian
– Jdamian, Commented Mar 30, 2017 at 14:44
E.g. if you have awk /BEGIN/{p=1;next} /END/{p=0;exit} p' whatsthat.file and only BEGIN and END in whatsthat.file it works. So the // are for checking the code between. — realKSMC
– realKSMC, Commented Mar 30, 2017 at 14:48

mklement0 · Accepted Answer · 2017-03-30 15:39:35Z

Your awk command has several problems, what you're looking for is this:

job=1

listen="$(
awk -v a="BEGIN-$job" -v e="END-$job" '$0==a {p=1;next} $0==e {exit} p && NF' whatsthat.file
)"

Patterns $0==a and $0==e check the input line at hand ($0) against the variables a and b for equality. If so, the associated action ({...}) is executed.
{p=1;next}, which is executed for the start of the range of interest, sets (conceptually) Boolean flag p and moves to the next input line; its purpose is to indicate that subsequent lines - due to being inside the range of interest - should (potentially) be printed.
{exit}, which is executed for the end of the range, unconditionally ends processing (the range-ending line, just like the range-starting one, shouldn't be printed).
p && NF is a pattern that tests flag p and whether the input line is nonempty (NF contains the number of whitespace-separated fields on the line, and that number is 0 for empty or blank lines); if both conditions are met, the input line at hand is implicitly printed (which is Awk's default action in the absence of an action associated with a pattern).

1 Answer 1