How to pass column name of a file as variable in awk - linux

Question

I trying to printing the data in the particular columns by passing them int awk command
I have tried using "-v" to set is as variable but its considering "$" as string. And my delimiter is special character ^A (ctrl+v+a).
vi test_file.dat
a^Ab^Ac^Ad^Ae^Af^Ag^Ah^Ai^Aj^Ak^Al^Am^An^Ao^Ap
Working code
awk -F'^A' '{print $2,$5,$7}' test_file.dat
It's Printing
b e g
But if I try
export fields='$2,$5,$7'
export file='test_file.dat'
awk -v sample_file="$test_file.dat" -v columns="$fileds" -F'^A' '{print columns}' sample_file
It's printing
$2 $5 $7
I expect the output as
b e g
And I want to pass the delimiter, columns, file name as a parameter like
export fields='$2,$5,$7'
export file='test_file.dat'
export delimiter='^A'
awk -v sample_file="$test_file.dat" -v columns="$fields" -v file_delimiter="$delimiter" -F'file_delimiter' '{print columns}' sample_file

Answer 1

In awk, the $ symbol is effectively an operator which takes the field numbers as arguments. The field names are expressions, which is why $NF works for denoting the last field: NF is evaluated by the $ operator. So as you can see, we should not include the dollar sign in the field names.
If you're using the environment to pass material to Awk, the right thing to do is to have Awk pick it up from the environment.
The environment can be accessed using the ENVIRON associative array. If a variable called delimiter holds the field separator, we might do something like
BEGIN { FS = ENVIRON["delimiter"] }
in the Awk code. Then we aren't dealing with yet another round of shell parameter interpolation issues.
We can pick up the field numbers similarly. The split function can be used to get them into an array. Refer to this one-liner:
$ fields=1,3,9 awk 'BEGIN { split(ENVIRON["fields"], f, ",") ;
for (i in f)
printf("f[%d] = %d\n", i, f[i]) ; }'
f[1] = 1
f[2] = 3
f[3] = 9
GNU Awk, the expression length(f) gives the number of fields.

Answer 2

In order to get awk to see the special characters while reading the file you could use cat -v file (there might be a built-in method, although I'm not aware of it). Then the key to getting the character ^A (Control-A) delimiter to be recognized is to escape it with a \, otherwise the regex capability of awk,gawk, etc. is to treat ^ as start of line.
export fields='$2,$5,$7'
export test_file='test_file.dat'
export delimiter='\\^A'
awk -F $delimiter '{ print '$fields' }' < <(cat -v test_file)
There's also no need to set awk variables for bash variables that have already set — so you can eliminate all of them essentially.
One thing to note if you did want to set them in awk is that columns wouldn't work because usually setting an awk variable from a bash variable would be assigned individually. For example -v var1='$2' -v var2='$5' -v var3='$7', so you'd end up for { print var1, var2, var3 } in awk. It's doubtful a single string can translated it into three variables without additional steps.