In bash, I know to be able to find the unique values between two arrays can be found by:
echo "${array1[@]} ${array2[@]}" | tr ' ' '\n' | sort | uniq -u
However, this gives the unique values between BOTH arrays. What if I wanted something about the elements that are unique only to array1 and elements that are unique only to array2? For example:
array1=(1 2 3 4 5)
array2=(2 3 4 5 6)
original_command_output = 1 6
new_command_output1 = 1
new_command_output2 = 6
You could use the comm command.
To get elements unique to the first array:
comm -23 \
<(printf '%s\n' "${array1[@]}" | sort) \
<(printf '%s\n' "${array2[@]}" | sort)
and elements unique to the second array:
comm -13 \
<(printf '%s\n' "${array1[@]}" | sort) \
<(printf '%s\n' "${array2[@]}" | sort)
Or, more robust, allowing for any character including newlines to be part of the elements, split on the null byte:
comm -z -23 \
<(printf '%s\0' "${array1[@]}" | sort -z) \
<(printf '%s\0' "${array2[@]}" | sort -z)
comm is probably the way to go but if you're running bash >= 4 then you can do it with associative arrays:
#!/bin/bash
declare -a array1=(1 2 3 4 5) array2=(2 3 4 5 6)
declare -A uniq1=() uniq2=()
for e in "${array1[@]}"; do uniq1[$e]=; done
for e in "${array2[@]}"; do
if [[ ${uniq1[$e]-1} ]]
then
uniq2[$e]=
else
unset "uniq1[$e]"
fi
done
echo "new_command_output1 = ${!uniq1[*]}"
echo "new_command_output2 = ${!uniq2[*]}"
new_command_output1 = 1
new_command_output2 = 6
BASH builtins can handle this cleaner and quicker. This will read both arrays for each element, comparing if they exist in either. If no match is found, output unique elements
arr1=(1 2 3 4 5)
arr2=(1 3 2 4)
for i in "${arr1[@]}" "${arr2[@]}" ; do
[[ ${arr2[@]} =~ $i ]] || echo $i
[[ ${arr1[@]} =~ $i ]] || echo $i
done
output: 5
If one of yours arrays have multiple character elements, e.g. 152, then you must convert the arrays, adding a literal character before and after. This way regex can identify an exact match
arr1=(1 2 3 4 5)
arr2=(1 3 2 4 152)
for i in "${arr1[@]}" ; do
var1+="^$i$"
done
for i in "${arr2[@]}" ; do
var2+="^$i$"
done
for i in "${arr1[@]}" "${arr2[@]}" ; do
[[ $var1 =~ "^$i$" ]] || echo $i
[[ $var2 =~ "^$i$" ]] || echo $i
done
output: 5 152
${arr1[@]} would split array elements on whitespace. Also, the index field in an array is an arithmetic context, so you could shorten to ${arr2[i-1]}
${arr1[@]} over "${arr1[@]}", would you? The former splits (1 '2 3' 4) into 1 2 3 4, the latter into 1 2 3 4, which is almost certainly preferable.