bash脚本从文件中获取花费的时间

Question

我有一个显示脚本之间切换时间的日志文件：

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

如何使用 bash 获取每个脚本花费的总时间 所以输出显示如下：

abc 2 hours 30 min 40 sec
XYZ 3 hours 23 min 45 sec
zen ...
XLS ...

Answer 1

假设您有一个名为 test.txt 的日志文件，以下脚本应该可以工作，

#!/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

这将产生如下输出：

 ]$ ./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

编辑

为了找到每个脚本花费的总时间，这个修改后的脚本应该可以完成这项工作：

   #!/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

结果：

 $ ./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

您可以使用以下gawk 程序：

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
    }
}

将它保存到一个文件 time_spent.awk 并像这样执行它：

gawk -f time_spent.awk input.log

上例的输出：

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)))

这是输出，假设您的时间在一个名为 logfile 的文件中：

$ ./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

$

注意：我从这里提取了方便的 display_time 函数：Python function to convert seconds into minutes, hours, and days。