Extract specific text out of bash variable containing TSV string

echo "$tsv_data" | awk '{print $2 " " $5}' | sort | awk '{print $2}' | paste -sd ',' -

This does the following steps in order:

• Print the second and 5th argument (Product and Description) with a space between them.
• Sort the input with sort (use gnu-sort if it can contain numbers)
• Print only the description (in each line)
• Join the lines together with paste

which will produce the following output:

Desc1,Desc2,Desc3

Here's a function that I suppose should do it:

get_descriptions() {
    local tsvstring="$1"
    local prodnames="$2"
    local result=()
    # read tsv line by line, splitting into variables
    while IFS=$'\t' read -r tags prodname val1 val2 desc || [[ -n ${prodname} && -n ${desc} ]]; do
        # check if the line matches the query, and if, append to array
        if grep -iq "${prodname}" <<< "${prodnames}"; then
            result+=("${desc}")
        fi
    done <<< "${tsvstring}"
    # echo the result-array with field separator set to comma
    echo $(IFS=,; echo "${result[*]}")
}

Then you can just use it like:

get_descriptions "${tsv_string_var}" "product1,product2"

echo "$var" | sort -k2 tags | cut -f5 | paste -sd,

sort + awk + paste pipeline:

echo "$tsv" | sort -nk2 | awk '{print $5}' | paste -sd',' -

The output:

Desc1,Desc2,Desc3

sort -nk2 - sorts the input by the second column numerically

awk '{print $5}' - prints out each fifth column

paste -sd',' - merge lines with ,