Shell - sum of column on user - shell

Question

Basically, I've two columns. First one stands for the users and the second one for the time they've spent on the server. So I'd like to sum for each client, how many minutes did he spend on the server.
user1 21:03
user2 19:55
user3 20:09
user1 18:57
user1 19:09
user3 21:05
user4 19:57
Let's say that I've this. I know how to split but there's one problem. Whenever I do awk -F: '{print $1} it prints users and the first parameter of the time (the number before :), and when I do awk -F: '{print $2} it prints only the numbers after :. After all of the sum, I'd like to get something like
user1 59:09
user2 19:55
user3 41:14
user4 19:57

Answer 1

Here's a possible solution:
perl -ne '/^(\S+) (\d\d):(\d\d)$/ or next; $t{$1} += $2 * 60 + $3; END { printf "%s %02d:%02d\n", $_, $t{$_} / 60, $t{$_} % 60 for sort keys %t }'
Or with better formatting:
perl -ne '
/^(\S+) (\d\d):(\d\d)$/ or next;
$t{$1} += $2 * 60 + $3;
END {
printf "%s %02d:%02d\n", $_, $t{$_} / 60, $t{$_} % 60
for sort keys %t;
}
'
We loop over all input lines (-n). We make sure every line matches the pattern \S+ \d\d:\d\d (i.e. a sequence of 1 or more non-space characters, a space, two digits, a colon, two digits) or else we skip it.
We accumulate the number of seconds per user in the hash %t. The keys are the user names, the values are the numbers.
At the end we print the contents of %t in a nicely formatted way.

Answer 2

this is an awk solution
cat 1.txt | awk '{a[$1]+=substr($2,0,2)*60+substr($2,4)} END {for(i in a) printf("%s %02d:%02d\n", i,a[i]/60,a[i]%60)}'
user1 59:09
user2 19:55
user3 41:14
user4 19:57
first construct an array with index=$1, and the value = convert the time to an integer by minutes * 60 + seconds
{a[$1]+=substr($2,0,2)*60+substr($2,4)}
then print the array in the desired format which converts integer to a mi:ss format.
printf("%s %02d:%02d\n", i,a[i]/60,a[i]%60)

Answer 3

If you want to use awk (and assuming the duration is always hh:mm, though their sizes can be arbitrary), the following will do the trick:
{
split($2, flds, ":") # Get hours and minutes.
mins[$1] += flds[1] * 60 + flds[2] # Add to initially zero array item.
}
END {
for (key in mins) { # For each key in array.
printf "%s %d:%02d\n", # Output specific format.
key, # Key, hours, and minutes.
mins[key] / 60,
mins[key] % 60
}
}
That's the expanded, readable variant, the compressed one is shown in the following transcript, along with the output as expected:
pax> awk '{split($2,flds,":");mins[$1] += flds[1] * 60 + flds[2]}END{for(key in mins){printf "%s %d:%02d\n",key,mins[key]/60,mins[key]%60}}' testprog.in
user1 59:09
user2 19:55
user3 41:14
user4 19:57
Just keep in mind you haven't specified the input format whenever a user entry has more than 24 hours. If it goes something like 25:42, the script will work as is.
If instead it decides to break out days (into something like 1:01:42 rather than 25:42), you'll need to adjust how the minutes are calculated. This can be done relatively easily (including the possibility for minutes-only entries) by checking the flds array size with (in the main body of the script, the non-END bit):
num = split($2, flds, ":")
if (num == 1) { add = flds[1] }
else if (num == 2) { add = flds[1] * 60 + flds[2] }
else { add = flds[1] * 1440 + flds[2] * 60 + flds[3] }
mins[$1] += add