I use in my bash script the following variables - val1 ,val2
val1 represented all disks that should be in OS linux
val2 represented the current disks in OS linux
my target is to delete the disks "$val2" from "$val1" variable
val1="sdb sdc sdd sde sdf sdg sdh sdi sdj sdk"
val2="sdb sdc sdf sdd sde sdg"
expected results:
val3=sdh sdi sdj sdk
and set the results as below in val3
With sed substitution:
val3=$(sed -E 's/\<('"$(tr ' ' '|' <<<"$val2")"')\> *//g' <<<"$val1")
echo "$val3"
sdh sdi sdj sdk
$(tr ' ' '|' <<<"$val2") - imitating regex alternation group, this will result in sdb|sdc|sdf|sdd|sde|sdg. The full construction should look as \<(sdb|sdc|sdf|sdd|sde|sdg)\>\< and \> - are word boundaries
" *". Or, alternatively, you may delete excessive spaces with additional tr command: sed -E 's/\<('"$(tr ' ' '|' <<<"$val2")"')\>//g' <<<"$val1" | tr -s ' '. That's all
One way:
$ val1="sdb sdc sdd sde sdf sdg sdh sdi sdj sdk"
$ val2="sdb sdc sdf sdd sde sdg"
$ comm -23 <(echo "$val1" | sed "s/ /\n/g" | sort) <(echo "$val2" | sed "s/ /\n/g"| sort) | paste -sd" "
sdh sdi sdj sdk
$
echo "$val1" | sed "s/ /\n/g" | sort :Flatten the list of variables and sort
comm -23 : Get the content unique to flattened val1
paste -sd" " : Concatenate the result using space
command
echo $val1| tr " " "\n" > val1.txt; echo $val2| tr " " "\n" > val2.txt; val3=`awk 'NR==FNR {a[$1];next}!($1 in a) {print $1}' val2.txt val1.txt` ;echo $val3
output
echo $val3
sdh sdi sdj sdk
With grep and process substitution:
#!/bin/bash
val1='sdb sdc sdd sde sdf sdg sdh sdi sdj sdk'
val2='sdb sdc sdf sdd sde sdg'
val3="$(grep -v -F -f <(printf "%s\n" $val2) <(printf "%s\n" $val1) | xargs)"
echo "$val1"
echo "$val2"
echo "$val3"
The process substitutions with printf "%s\n" are used to make the lists in var1 and var2 appear to be files with one entry per line to grep. var2 is the input "file" to the -f option, and var1 is the "file" to be grepped. xargs is used to convert the output of the grep from multiple lines to a single space separated string...a lazy but useful trick.
And here's the same thing using arrays:
#!/bin/bash
val1=(sdb sdc sdd sde sdf sdg sdh sdi sdj sdk)
val2=(sdb sdc sdf sdd sde sdg)
val3=($(grep -vFf <(printf "%s\n" "${val2[@]}") <(printf "%s\n" "${val1[@]}")))
echo "${val1[@]}"
echo "${val2[@]}"
echo "${val3[@]}"
