Why Bash Returns Zero When Summing a Column and How to Fix It

Background

We want to calculate the sum of the numbers in the second column of a file test.data. A one‑liner with awk '{s+=$2} END {print s}' test.data works, but the article focuses on implementing the same task by processing the file line by line in Bash.

Attempt One: Using a Pipe

The first script pipes the file into a while read loop:

#!/usr/bin/env bash
sum=0
cat test.data | while read line
do
    temp_num=$(echo "$line" | cut -d ' ' -f 2)
    sum=$(( $sum + $temp_num ))
done
echo "we get sum:$sum"

The output is we get sum:0. The reason is that the pipeline creates a subshell; the variable sum is modified inside that subshell and its value is lost when the loop finishes.

Running shellcheck on the script produces warnings such as SC2002: Useless cat, SC2030: Modification of sum is local (to subshell), and SC2031: sum was modified in a subshell, confirming the problem.

Attempt Two: Using for with Command Substitution

Because pipelines are discouraged, the next version uses a for loop:

#!/usr/bin/env bash
sum=0
for line in $(cat test.data)
do
    echo "get line :$line"
    temp_num=$(echo "$line" | cut -d ' ' -f 2)
    sum=$(( $sum + $temp_num ))
done
echo "we get sum:$sum"

The result shows each token separated by spaces (e.g., get line :1, get line :12, …) and the final sum is 135, but the loop stops at any whitespace, not at line boundaries. Setting IFS=$'\n' is suggested to treat only newlines as delimiters, although the author notes it did not work on their own system.

Attempt Three: Redirecting Input Directly

The reliable solution redirects the file into the while read loop, avoiding a pipeline:

#!/usr/bin/env bash
sum=0
while read line
do
    echo "line $line"
    temp_num=$(echo "$line" | cut -d ' ' -f 2)
    sum=$(( $sum + $temp_num ))
done < "test.data"
echo "we get sum:$sum"

This script produces the correct sum. A variant that reads two columns directly is also shown:

#!/usr/bin/env bash
sum=0
while read col1 col2
do
    sum=$(( $sum + $col2 ))
done < "test.data"
echo "we get sum:$sum"

Here col1 and col2 represent the first and second columns respectively.

Handling Escape Sequences

If the input contains escaped characters such as \n, the default read interprets them. Adding the -r option prevents backslash interpretation:

while read -r line
    # processing …

With -r, the literal \n strings are preserved.

Summary of Pitfalls

Using a pipeline ( cat … | while read …) runs the loop in a subshell, causing variable updates to be lost.

Iterating with for line in $(cat …) splits on any whitespace, not just newlines.

Lines containing spaces, tabs, or escaped characters need careful handling (e.g., setting IFS or using read -r). shellcheck recommends avoiding for … in $(cat …) and prefers a direct while read loop with input redirection.