arrays - Gawk distinct and sum column -

i new linux , usage of awk , couldn't find answer following question:

i want use awk , file structured that:

date id size 2016-11-09 688 47 2016-11-09 688 56 2016-11-09 31640 55 

now want sum size each line has date , id , export .csv file. file should that:

date,id,size 2016-11-09,688,103 2016-11-09,31640 55 

i need help, because not figure out how on own, thank you.

if input sorted date , id in sample should use this:

$ cat tst.awk begin { ofs="," } nr==1 { $1=$1; print; next } { curr = $1 ofs $2 } (curr != prev) && (nr > 2) { print prev, sum; sum=0 } { prev = curr; sum += $3 } end { print prev, sum }  $ awk -f tst.awk file date,id,size 2016-11-09,688,103 2016-11-09,31640,55 

rather saving whole file in memory. note approach produce output in same order input whereas for .. in .. loop in end section print output in random (hash) order.