I want to get a part in every line of a regular file. For this purpose I've used the program awk in my script. I need to put each part in an array. Something like this works:
declare -i j=1
awk 'BEGIN { FS="|" } { a[j]=$6; ((j++)) }' myFile
but I have a problem after in the script, when I need to use the array. Indeed the array in the awk block isn't seen by the remaining lines of the script -- the lines outside that block. How can this problem be solved?
As Tom Fenech points out, shell code and awk scripts are separate worlds with separate variables that know nothing of each other.
(More generally, any external utility runs in a child process and therefore cannot modify the calling shell's environment.)
In order to read awk output into a bash array, you must read awk's stdout output.
In Bash v4+, you can do the following:
readarray -t a < <(awk 'BEGIN { FS="|" } { print $6 }' myFile)
This creates Bash array "${a[@]}" from the individual lines in awk's stdout output.
Note the use of a process substitution (<(...)) to provide input to readline via stdin (<), which ensures that the array remains in scope in the rest of the script.
/bin/sh rather than /bin/bash.
jinside and outside of the awk script refer to two different things.((j++))works inawktoo, there's no reason for enclosing the expression in parentheses (unlike inbash).