focused-window-tracker

This project monitors active window names and visualizes them using Prometheus and Grafana. It's just like ActivityWatch.

Overview

This project consists of three main components:

1. backend: A Python FastAPI application that receives window name updates and exposes Prometheus metrics.
2. prometheus: A Prometheus server that scrapes metrics from the backend application.
3. grafana: A Grafana instance that visualizes the metrics stored in Prometheus as a timeline showing which window was active and how long.
4. Scripts/monitor.sh: A script that sends active window name to backend.

Setup Instructions

1. Build the Docker Images

   docker compose build

2. Start the Docker Containers

   docker compose up -d

   This will start all services (backend, prometheus, and grafana) in detached mode.

3. Start the window monitor script

   • Linux:

     bash ./Scripts/monitor.sh

   • Windows:

   python ./Scripts/monitor.py
   

   This script will start sending requests to the backend.

4. Access Grafana

   • Open your web browser and go to http://127.0.0.1:3000.
   • Log in with the default credentials:
     • Username: usr
     • Password: pwd

5. Configure Prometheus Data Source

   • In the Grafana sidebar, navigate to "Connections" (the plug icon).
   • Click "Data sources."
   • Click "Add data source."
   • Search and select "Prometheus."
   • Enter the Prometheus URL: http://prometheus:9090.
   • Click "Save & test." You should see a "Successfully queried the Prometheus API" message.

6. Create the Visualization

• Create a new dashboard by navigating to "Dashboards" and click "New Dashboard"
• Add a new panel of type "Time series".
• In the "Query" section, enter the following Prometheus query: current_window_name
• In the "Transform" tab, add the following transforms:
  • Select 'Reduce' in dropdown, and 'Last (Not NaN)' in function dropdown, then 'Group By' and 'window_name' in the Group By dropdown
  • Click 'Add Transform' and select 'Prepare Time Series'
  • Click 'Add Transform' and select 'Organize fields' and Rename 'Value' to 'is_active'. Uncheck Timestamp, Select is_active and set Display style to 'Hidden'. Select window_name and select 'Type' as 'String'
  • Click 'Add Transform' and select 'Filter by value'. Set the Filter to 'is_active' and select condition 'is equal to' and set value '1'
• Go to display, and in the section 'Band' Select 'Gradient Mode' to 'Scheme' and set Scheme to 'Cool to Warm'. * In the Legend set mode to 'Table'
• Save the dashboard.