Simple per-minute tty activity tracker.
It works like this:
- periodically check all ttys for activity;
- scan process list for what process is run on each active tty;
- log into
~/tt/tt.logtimestamp (with minute resolution) and process command line.
Then you can view statistics of how much you are worked at time interval (day, hour, marked interval) and what you did. This is intentionally is not very precise, as precision is not always best thing, but it is good enough for understanding and sometimes billing. Smallest time unit is minute. If many commands was recorded at the same minute they will be accounted just as one minute of work time.
You will need ruby as prerequisite.
workbox:~/$ git clone http://github.com/aabc/tt.git
This will clone tracker code into your ~/tt/ directory. Note that ~/tt/ name is built-in. Add autostart of tracker into crontab, example:
* * * * * ~/tt/timetrack.rb --daemon 2>/dev/null
To start every minute is normal, because tracker will let itself run only once and in case of some fault it ill restart quickly.
Check that you have ~/bin/ in your PATH env variable, if it's there: ln -s ~/tt/timetrack.rb ~/bin/tt. Alternatively you can create bash alias into your .bashrc: alias tt=~/tt/timetrack.rb. Done.
Now to view statistics just run tt. To get some command line help run tt --help.
Example output of tt:
Total 19 minutes (0.32 hours). From 2014-10-06 04:25 to 2014-10-06 05:29. --today
1 0.25 -bash
21 16.75 bash
2 2.00 vim README.md
- Total minutes is how much minutes you was active for specified interval (which is
--todayin this case, it's also default and could be changed with command line options). - From ... to ... first and last registered activity time for your interval, so you can check that all your work time (ex. for today) is accounted. Timestamp is in localtime.
- Then goes per-command statistics:
- First column is total active minutes accounted for this command.
- Second column is minutes without shared time. For example, if you are used two command in same minute they may have time 0.5. Or, if processes are scanned each 10 seconds (changeable default) and you used command A once per minute and command B is registered 4 times per same minute, then they will have share 0.2 and 0.8 accounted in second column. Smaller you set scan interval, more precise is calculation is. (But, I think 5-10 sec is enough in any case.)
- Third column is detected process command line.
Note, that sum of the first column should not give total work time, because some commands usually have shared minutes (i.e. many commands is run in same minute). But, sum of the second column should approximate to total time, with some imprecision. In general you should rely on totals as more precise time estimation.
You may mark some point in time with tt --mark=markname, then subtotals will be shown like this (here mark is 'test'):
subtotal 2014-10-06 04:25 -- 'test' 20 (0.33)
subtotal 'test' -- 2014-10-06 05:43 1 (0.02)
- First row is work time between beginning of time interval and your mark in minutes and hours.
- If multiple marks are used they will be accounted too, each interval per row.
- Last row is work time accounted from the last mark to current time.