> grep "^$1," database | sed -e "s/^$1,//" | tail -n 1 FWIW. Shell pipelines are very useful, but sometimes, you don't need them. Case in point: - sed can do

> grep "^$1," database | sed -e "s/^$1,//" | tail -n 1 FWIW. Shell pipelines are very useful, but sometimes, you don't need them. Case in point: - sed can do the matching: sed -n -e "/^$1,/s/^$1,//p" | tail -1 - sed can just take the last match: sed -n -e "/^$1,/h;\${g;s/^$1,//p}" - sed can be cryptic, so you might prefer awk: awk -F, '$1 == "'"$1"'" { val=$2 } END { print val }' although the fact that it's in a shell function makes the "'"$1"'" thing kind of confusing. It would be less confusing if using a named environment variable... maybe less bad with explicit escapes... awk -F, "\$1 == \"$1\" { val=\$2 } END { print val }"