Focus

A tool I wrote for myself to help me keep track of how much time I spend using different programs or browsing web.

Consists of 2 parts:

Polling script for data gathering
Jupyter Notebook for analysis

Polling script

The polling script continuosly records current window in focus and saves it to ./logs/ as .json.gz

It saves data such as pid, window name, process creation time, time in focus, executable path, command line arguments. Every switch of a window is a new data point (to make a sequence plot described below).

It also filters any idle time into a different data group and has keywords that won't set the idle timer.

If it lacks permissions or another error occurs, it will put window names with traceback in error log.

Notebook

The notebook Focus.ipynb looks like this:

It categorizes every data point by window_name and exe using custom OrderedDict tables.

Different plots are available:

plot_day_sequence_chart similar to Gantt chart but with top N categories (above picture).
plot_top_categories plots top N categories of all time.
plot_top_by_date same as above but splits each category by days.
plot_category_by_day plots one category usage each day.
plot_timeline_by_category_time same as above but for top N categories.

Installation

I recommend using Anaconda

For the polling script

To install necessary components for the polling script:

conda install psutil ujson

Additiinal components

If on Windows:

conda install pywin32

If on Linux:

sudo apt-get install xprintidle xdotool

Autostart

Windows

To run the logging script on startup, put a shortcut of the script focus_bg.pyw into C:\Users\USERNAME\AppData\Roaming\Microsoft\Windows\Start Menu\Programs\Startup\ .

Linux:

crontab -e
@reboot python /path/to/focus_bg.pyw

For the notebook

conda install pandas seaborn

RomanKornev/Focus