How can I use a variable in awk command

Question

With my code I am trying to sum up the values with the specific name of a column in a csv file, depending on the input of the name. Here's my code:

#!/bin/bash

updatedata() {

    index=0
    while IFS="" read -r line
    do
        IFS=';' read -ra array <<< "$line"
        for arrpos in "${array[@]}"
        do
            if [ "$arrpos" == *"$1"* ] || [ "$1" == "$arrpos" ]
            then
                break
            else
                let index=index+1
            fi
        done
        break
       
    done < data.csv
    ((index=$index+1))


       
    if [ $pos -eq 0 ]
    then
        v0=$(awk -F";", -v index=$index '{x+=$index}END{print x}' ./data.csv )
    elif [ $pos -eq 1 ]
    then
        v1=$(awk -F";" '{x+=$index}END{print x}' ./data.csv )
    elif [ $pos -eq 2 ]
    then
        v2=$(awk -F";" '{x+=$index}END{print x}' ./data.csv )
    elif [ $pos -eq 3 ]
    then
        v3=$(awk -F";" '{x+=$index}END{print x}' ./data.csv )
    fi
               
                   
         
}

In the middle of the code you can see in v0=, I was trying to experiment a little, but I just keep getting errors:

First I tried this:

v0=$(awk -F";" '{x+=$index}END{print x}' ./data.csv)

but it gave me this error:

'awk: line 1: syntax error at or near }'

so then I decided to try this(as you can see in the code)

v0=$(awk -F";", -v index=$index '{x+=$index}END{print x}' ./data.csv )

And I got this error: 'awk: run time error: cannot command line assign to index type clash or keyword FILENAME="" FNR=0 NR=0'

I don't know what to do. Can you guys help me.

Answer 1

Given some CSV data in data.csv,

A;B;C
1;2;3
4;5;6
-1.2;3;3.3

the following script would calculate the sum of the column named by the colname variable given on the command line:

BEGIN {
        FS = ";"

        if (colname == "") {
                print "Did not get column name (colname) to work with" >"/dev/stderr"
                exit 1
        }
}

FNR == 1 {
        colnum = 0

        for (i = 1; i <= NF; ++i)
                if ($i == colname) {
                        colnum = i
                        break
                }

        if (colnum == 0) {
                printf "Did not find named column (colname = \"%s\")\n", colname >"/dev/stderr"
                exit 1
        }

        sum = 0
        next
}

{
        sum += $colnum
}

END {
        print sum
}

Testing it:

$ awk -v colname='A' -f script.awk data.csv
3.8
$ awk -v colname='B' -f script.awk data.csv
10
$ awk -v colname='C' -f script.awk data.csv
12.3
$ awk -v colname='D' -f script.awk data.csv
Did not find named column (colname = "D")

---

Shorter variant of the script without so much error checking:

BEGIN { FS = ";" }

FNR == 1 {
        for (i = 1; i <= NF; ++i)
                if ($i == colname) break

        if (i > NF) exit 1
        next
}

{ sum += $i }

END { print sum }

or, as a "one-liner":

$ awk -v colname='A' -F ';' 'FNR == 1 { for (i = 1; i <= NF; ++i) if ($i == colname) break; if (i > NF) exit 1; next } { sum += $i } END { print sum }' data.csv

---

Ideally, though, you'd use some form of CSV parser, like CSVkit:

$ csvstat --sum -c A data.csv
3.8

The csvstat utility calculats several different statistics for any given CSV file. Here, it figures out that the delimiter is ; on its own. In this example I ask it for the sum of the column named A.