How can aggregate lines in CSV based on value of one column

miller is a pretty amazing tool:

mlr --csv --fs : group-by version then nest --ivar , -f device file.csv

version:device
1.0.0:dev1,dev2
1.2.3:dev3
1.3.4:dev4,dev5

try

awk -F: '$1 == before { printf ",%s",$2 ; } 
         $1 != before { printf "%s%s",nl,$0 ; before=$1 ; nl="\n" } 
         END {printf nl ; }' 

where

• -F: tell awk to use : as separator
• printf do not print newline, unless tol ("\n")
• this can be onelined (e.g.awk -F: '....' file), I split line for readability.

this use two variables to hold new line (nl) and previous $1's value.

Something like this should do what you want.

awk -F':' 'OFS=""; NR == 1; NR > 1 {x=$1;$1="";a[x]=a[x]$0","}END{for(x in a)print x":"a[x]}' file

awk -F: -v OFS=: '
    NR == 1{print;next} 
    !v {v=$1;d=$2;next} 
    v == $1 {d = d","$2}
    v != $1 {print v,d;v=$1;d=$2}
    END{print v,d}
' file

$ cat tst.awk
BEGIN { FS=OFS=":" }
$1 != prev {
    if ( NR == 1 ) {
        devs = $2 "s"
    }
    else {
        print prev, devs
        devs = $2
    }
    prev = $1
    next
}
{ devs = devs "," $2 }
END { print prev, devs }

$ awk -f tst.awk file
version:devices
1.0.0:dev1,dev2
1.2.3:dev3
1.3.4:dev4,dev5

Yet another solution with csvkit:

csvsql -d ':' --query \
    'SELECT version, GROUP_CONCAT(device) as devices FROM file GROUP BY version' file.txt \
    | csvformat -D ':'
version:devices
1.0.0:dev1,dev2
1.2.3:dev3
1.3.4:dev4,dev5