mjhjmtu

I have an array like this:

array=(1 2 7 6)

and would like to search for the second largest value, with the output being

secondGreatest=6

Is there any way to do this in bash?

printf '%sn' "$array[@]" | sort -n | tail -2 | head -1

Print each value of the array on it's own line, sort it, get the last 2 values, remove the last value

secondGreatest=$(printf '%sn' "$array[@]" | sort -n | tail -2 | head -1)

Set that value to the secondGreatest variable.

Glenn Jackman had an excellent point about duplicate numbers that I didn't consider. If you only care about unique values you can use the -u flag of sort:

secondGreatest=$(printf '%sn' "$array[@]" | sort -nu | tail -2 | head -1)

A bash-specific loop through the array could do it; you have to keep track of the largest and second-largest. The only other tricky part is to be careful about initializing those values; the largest value is initialized to the first element; the second-largest value is initialized the first time we see a value that's smaller than the largest value. Subsequently for the second-largest value, we only update it if it's strictly less than the current largest value:

#!/bin/bash

array=(7 7 6 2 1)

if [ "$#array[@]" -lt 2 ]
then
 echo Incoming array is not large enough >&2
 exit 1
fi

largest=$array[0]
secondGreatest='unset'

for((i=1; i < $#array[@]; i++))
do
 if [[ $array[i] > $largest ]]
 then
 secondGreatest=$largest
 largest=$array[i]
 elif (( $array[i] != $largest )) && [[ "$secondGreatest" = "unset" ]] 
 then
 secondGreatest=$array[i]
 fi
done

echo "secondGreatest = $secondGreatest"

It's still slower than calling out to sort, but it has the added benefit of picking the strictly-smaller second-largest value in the face of multiple high values (such as 7 and 7 above).

It's a good job for dc :

array=(1 2 7 6)
echo $array[*] | dc -f - -e '
 [lasbdsa]sB
 [dla!>Bsc1z>A]sA
 lAx
 [secondGreatest=]nlbp'