bash script to get the spent time from a file - linux

Question

I have a log file that shows switch time between my scripts:
Tue Oct 24 11:57:54 IRST 2017 Script switched from abc to XYZ
Tue Oct 24 14:03:41 IRST 2017 Script switched from XYZ to ZEN
Tue Oct 24 15:43:16 IRST 2017 Script switched from ZEN to XYZ
Tue Oct 24 17:07:25 IRST 2017 Script switched from XYZ to ZEN
Tue Oct 24 18:40:48 IRST 2017 Script switched from ZEN to XLS
Tue Oct 24 19:52:26 IRST 2017 Script switched from XLS to XYZ
Tue Oct 24 20:20:30 IRST 2017 Script switched from XYZ to ZEN
Tue Oct 24 20:36:06 IRST 2017 Script switched from ZEN to XLS
Tue Oct 24 21:01:03 IRST 2017 Script switched from XLS to XYZ
Tue Oct 24 21:47:47 IRST 2017 Script switched from XYZ to ZEN
How do I get total time spent on each script with bash
So the output shows like this:
abc 2 hours 30 min 40 sec
XYZ 3 hours 23 min 45 sec
zen ...
XLS ...

Answer 1

Assuming you have a log file named test.txt, following script should work,
#!/bin/bash
dtime=0
sname=""
while read line
do
_dtime=$(echo "$line" | awk '{print $1,$2,$3,$4}')
_sname=$(echo "$line" | awk '{print $10}')
_dtimesec=$(date +%s -d "$_dtime")
_timediff=$(( _dtimesec - dtime ))
[ "x$sname" != "x" ] && printf "$sname %d hours %d minutes %d seconds\n" $(($_timediff/3600)) $(($_timediff%3600/60)) $(($_timediff%60))
dtime=$_dtimesec
sname=$_sname
done < test.txt
This will produce an output like the following:
]$ ./test
abc 2 hours 5 minutes 47 seconds
XYZ 1 hours 39 minutes 35 seconds
ZEN 1 hours 24 minutes 9 seconds
XYZ 1 hours 33 minutes 23 seconds
ZEN 1 hours 11 minutes 38 seconds
XLS 0 hours 28 minutes 4 seconds
XYZ 0 hours 15 minutes 36 seconds
ZEN 0 hours 24 minutes 57 seconds
XLS 0 hours 46 minutes 44 seconds
EDIT
In order to find total amount of time spent by each script, this modified script should do the job:
#!/bin/bash
dtime=0
sname=""
namearr=()
timearr=()
while read line
do
_dtime=$(echo "$line" | awk '{print $1,$2,$3,$4}')
_sname=$(echo "$line" | awk '{print $10}')
_dtimesec=$(date +%s -d "$_dtime")
_timediff=$(( _dtimesec - dtime ))
_rc=1
for n in "${!namearr[#]}"
do
if [ "${namearr[$n]}" == "$_sname" ]; then
export _rc=$?
export ind=$n
break;
else
export _rc=1
fi
done
if [ $_rc -eq 0 ]; then
timearr[$ind]=$(( ${timearr[$ind]} + _timediff ))
else
if [ $dtime -eq 0 ] && [ "x$sname" == "x" ]; then
:
else
namearr+=($_sname)
timearr+=($_timediff)
fi
fi
dtime=$_dtimesec
sname=$_sname
done < test.txt
echo "Total time spent by each script:"
echo
for i in "${!namearr[#]}"
do
_gtime=${timearr[$i]}
printf "${namearr[$i]} %d hours %d minutes %d seconds\n" $(($_gtime/3600)) $(($_gtime%3600/60)) $(($_gtime%60))
done
Result:
$ ./test
Total time spent by each script:
XYZ 4 hours 44 minutes 44 seconds
ZEN 3 hours 28 minutes 34 seconds
XLS 1 hours 36 minutes 35 seconds

Answer 2

You can use the following gawk program:
time_spent.awk
BEGIN {
months["Jan"] = "01"
months["Feb"] = "02"
months["Mar"] = "03"
months["Apr"] = "04"
months["May"] = "05"
months["Jun"] = "06"
months["Jul"] = "07"
months["Aug"] = "08"
months["Seb"] = "09"
months["Oct"] = "10"
months["Nov"] = "11"
months["Dec"] = "12"
}
{
split($4, time, ":")
# mktime() manual: https://www.gnu.org/software/gawk/manual/html_node/Time-Functions.html
now = mktime($6" "months[$2]" "$3" "time[1]" "time[2]" "time[3])
prv = $(NF-2)
cur = $(NF)
start[cur] = now
spent[prv]+=start[prv]?now-start[prv]:0
}
END {
for(i in spent) {
printf "%s seconds spent in %s\n", spent[i], i
}
}
Save it into a file time_spent.awk and execute it like this:
gawk -f time_spent.awk input.log
Output from the above example:
5795 seconds spent in XLS
0 seconds spent in abc
17084 seconds spent in XYZ
12514 seconds spent in ZEN

Answer 3

#!/usr/bin/env python
import sys
from time import strptime
from datetime import datetime
intervals = (
('weeks', 604800), # 60 * 60 * 24 * 7
('days', 86400), # 60 * 60 * 24
('hours', 3600), # 60 * 60
('minutes', 60),
('seconds', 1),
)
def display_time(seconds, granularity=2):
result = []
for name, count in intervals:
value = seconds // count
if value:
seconds -= value * count
if value == 1:
name = name.rstrip('s')
result.append("{} {}".format(value, name))
return ' '.join(result[:granularity])
with open(sys.argv[1], "rb") as df:
lines = df.readlines()
totals = {}
for i in range(len(lines)-1):
(_,t1,t2,t3,_,t4,_,_,_,_,_,scr) = lines[i].strip().split(' ')
st = datetime.strptime(' '.join([t1,t2,t3,t4]), "%b %d %H:%M:%S %Y")
(_,t1,t2,t3,_,t4,_,_,_,_,_,_) = lines[i+1].strip().split(' ')
et = datetime.strptime(' '.join([t1,t2,t3,t4]), "%b %d %H:%M:%S %Y")
if scr not in totals:
totals[scr] = 0
totals[scr] += (et-st).seconds
print("{} {}".format(scr,display_time((et-st).seconds, 3)))
print("\nTotals:")
for scr in totals:
print("{} {}".format(scr,display_time(totals[scr], 3)))
Here is the output, assuming your times are in a file named logfile:
$ ./times.py logfile
XYZ 2 hours 5 minutes 47 seconds
ZEN 1 hour 39 minutes 35 seconds
XYZ 1 hour 24 minutes 9 seconds
ZEN 1 hour 33 minutes 23 seconds
XLS 1 hour 11 minutes 38 seconds
XYZ 28 minutes 4 seconds
ZEN 15 minutes 36 seconds
XLS 24 minutes 57 seconds
XYZ 46 minutes 44 seconds
Totals:
XLS 1 hour 36 minutes 35 seconds
XYZ 4 hours 44 minutes 44 seconds
ZEN 3 hours 28 minutes 34 seconds
$
Note: I lifted the handy display_time function from here: Python function to convert seconds into minutes, hours, and days.