dufferzafar/design-time-tracker.md

## design-time-tracker.md

      
    Raw
  

              design-time-tracker.md
            
          
    Time Tracker

Build a Manic Time alternative.
The application would be in two parts:

a tracker agent that dumps data into some backend

should be dumb and not do much processing of the data


a reporting dashboard which will show various timelines, statistics etc.

Features


Completely Open Source

Multiple Datastores


Stay backend agnostic


Application can be built using something like SQLite

But the design should be generic enough to support any thing (JSONs, MySQL etc.)


This lets the user stay in control of whether data goes over the wire to some remote connectordb instance or stays locally in plain text files.


Export data as CSV, Excel etc.?


Not just active Windows


Collect other metrics from the system


Internet usage

For all interfaces?


Keyboard activity


Mouse activity


Setup


Should be straight-forward

pip install time-tracker
a shell script, if dependencies are needed etc.
might not be easy on Windows


Platforms

Linux


Try to be as Distro/DE agnostic as possible


View issues on Thyme to see what stuff can go wrong


External tools will do the heavy lifting

xdpyinfo
xwininfo
xprop
xdotool
wmctrl


If using Python, try to find pure python ways of doing the things

as it'll be faster than executing processes

do a performance test for this


macOS


Applescripts can do a lot of the heavy lifting

Windows


Windows API will make things quite easy

Manic Time is a proof of what all can happen


Prior Art

A list of tools (opensource & otherwise) that already exist, what they do, and what we can steal from them.
Thyme


Written in Golang


New kid in town, so not mature enough


Has decent cross-platform support


Simple but neat UI

Made with Google Charts


Stores data in plain JSON files


SelfSpy


Written in Python


Not actively developed (last commits in 2015)


macOS: Pretty legit looking support using PyObjC


Windows: pyHook, pythoncom


Linux: Pure Python XLib


uLogMe


Written in Python by Karpathy


Tailored to Ubuntu/macOS


Feels pretty hacky

Bunch of shell scripts etc.


Takes screenshots too


Linux: xdotool, wmctrl etc.


macOS: Custom application that is called by shell scripts


Windows: No support yet


laptoplogger


Written in Python


Nice, clean code


macOS: No support yet


Windows: PyQt


Linux: Pure Python XLib


Too coupled with connectordb as a backend


Manic Time


Stores data in a .sdf databse

Schema can be stolen from


Rescue Time


Find what's special.

Waka Time


Time Tracker, but for Programming Editors only


Find out if there's anything special to what they do

something that can't be done via Title grabbing & some regex-fu


Reports Dashboard

Read and aggregate data from the datastore and display in the UI.
Use latest web technologies for building the dashboard:

React
d3.js charts
Material UI etc.

Allow searching, filtering, grouping etc.
The reporting UI will need a datastore of it's own:

to store things like group of windows.
so a user can club windows with titles "Word", "Excel" into a group called "Work Related" etc.

Misc.


Write an email to Stephen Wolfram to find out what tools/scripts he uses to track all this.

Away Time

One of Manic Time features, where it detects inactivity and marks that time as away.
uLogMe has screensaver detection for various DEs.
Application Specific Tracking & Reporting

The current Window title may not be enough for every application, and some other metrics might be nice too!
Current URL of the browser is a nice example, but after enough googling, I found that it's not at all trivial (ManicTime has custom plugins that do this though.)
A workaround for this might be to use browser extensions like this one (for Firefox) that appends the page URL to window title.
This data will then have to be taken into account while performing aggregation in the dashboard, so for eg. regex: ([^\|]*) \| (.*) - .* will extract the page title & url from Firefox windows using the above plugin & | as a separator between title & url.
A similar thing will be needed for Sublime Text. It's window titles are usually of the form: {file-path} • ({project-name}) - Sublime Text and both paths & project name are interesting information to have when focusing on insights. regex: ([^•]*) • \(([^)]*)\) -
Links that might be useful

http://unix.stackexchange.com/questions/5478/what-process-created-this-x11-window
http://superuser.com/questions/382616/detecting-currently-active-window