Skip to content

mattlokes/touchkio

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

TouchKio

TouchKio is a Node.js application that utilizes Electron to create a kiosk mode specifically designed for a Home Assistant dashboard. This tool is packaged as a .deb file, making it easy to launch the kiosk mode on a Raspberry Pi equipped with the official 7" touch display.

display

This implementation addresses common issues encountered when using the built-in browser running in fullscreen mode on the Raspberry Pi touch display. Moreover, the device running the kiosk application also offers several Home Assistant MQTT sensors, enhancing it's functionality for automation purposes.

Features

  • Fast and easy setup.
  • Touch friendly web browser interface.
  • Configurable zoom and theme support.
  • Single-tap screen wake-up functionality.
  • Remote controllable via MQTT.
    • Manage kiosk window status.
    • Touch display power and brightness.
    • Execute system reboot and shutdown commands.
    • Monitor temperature, processor and memory usage.

The kiosk application can be executed with command line arguments to load a Home Assistant dashboard in fullscreen mode. Additionally, a MQTT endpoint can be defined, allowing the application to provide controls and sensors for the Raspberry Pi and the connected touch display.

Setup

Before you begin, ensure that you have a Raspberry Pi configured and running with a compatible touch display. This guide assumes you are using Raspberry Pi OS Lite (64-bit), which is recommended for kiosk applications due to its lightweight nature and minimal overhead.

Optional

To utilize the sensor features of your display through Home Assistant, it's essential to have a MQTT broker running and the MQTT integration installed on your Home Assistant instance. This setup allows seamless communication between your kiosk device and Home Assistant, enabling real-time data exchange.

mqtt

For a comprehensive guide on setting up MQTT with Home Assistant, please refer to the official documentation available here: https://www.home-assistant.io/integrations/mqtt.

Installation

On the first run of the application, you may encounter a setup procedure and the Home Assistant login screen. It's recommended to create a dedicated user (local access only) for your kiosk device. You might also need a keyboard or remote VNC access to input these credentials once.

Option 1 - The easy way

Run this command to download and install the release .deb file. It will also adjust some permissions, create a systemd file for auto-startup and will guide you through the setup process:

bash <(wget -qO- https://raw.githubusercontent.com/leukipp/touchkio/main/install.sh)

If you are paranoid, or smart, or both, have a look into the install.sh script before executing external code on your machine.

Option 2 - The hard way

The install.sh script mentioned above performs the following tasks (and you just have to do it manually):

  • Download the latest .deb file, which is specifically suitable for Raspberry Pi (64-bit) ARM architecture.
  • Open a terminal and execute the following command to install the application, e.g: sudo dpkg -i touchkio_1.x.x_arm64.deb
  • If you just want to load a Home Assistant dashboard without further control you are good to go, e.g: touchkio --web-url=https://demo.home-assistant.io

If you need the application to be automatically started on boot, create a systemd file.
If you need the Home Assistant MQTT sensors, create the udev rule for backlight control.

Configuration

Running touchkio --setup will prompt you to enter arguments that will be used when the application starts without any specified arguments. These default arguments are stored in ~/.config/touchkio/Arguments.json, where they can also be modified.

WEB

The available arguments to control the kiosk application via terminal are as follows:

Name Default Description
--web-url (Required) - Url of the Home Assistant instance (HTTP(S)://IP:PORT)
--web-theme (Optional) dark Theme settings of the web browser (light or dark)
--web-zoom (Optional) 1.25 Zoom settings of the web browser (1.0 is 100%)

These arguments allow you to customize the appearance of the web browser window.

For example:

touchkio --web-url=http://192.168.1.42:8123 --web-theme=dark --web-zoom=1.25

MQTT

To broadcast your local sensor data to Home Assistant, you can use the following arguments, which require a running MQTT broker:

Name Default Description
--mqtt-url (Required) - Url of the MQTT broker instance (MQTT(S)://IP:PORT)
--mqtt-user (Required) - Username which is available in Home Assistant (e.g. create a user named kiosk)
--mqtt-password (Required) - The password of the user (e.g. use password, because it's secure and easy to remember)
--mqtt-discovery (Optional) homeassistant The discovery prefix for MQTT (homeassistant works with default setups)

When you start the application with the MQTT arguments, the Home Assistant auto-discovery feature will automatically add controls and sensors that can be used for further integration. You can find them under Settings -> Devices and Services -> Devices by searching for "Raspberry Pi".

For example:

touchkio --web-url=http://192.168.1.42:8123 --mqtt-url=mqtt://192.168.1.42:1883 --mqtt-user=kiosk --mqtt-password=password

Development

If you want to create your own local build install nodejs and yarn on your Raspberry Pi.

Clone this repository and run yarn install to install the dependencies. Then use yarn start to execute the start script located in the package.json file. There you can adjust the --web-url and other arguments for development runs.

The nitty gritty

Don't waste your time reading this.

To enable write access to the /sys/class/backlight/10-0045/bl_power and /sys/class/backlight/10-0045/brightness files, you need to set up a udev rule. This is done within the install.sh script, which also creates a .service file. While creating a service file is optional, it's highly recommended if you want your Raspberry Pi to automatically boot into kiosk mode.

The Raspberry Pi's build-in screen blanking function uses the command swayidle -w timeout 600 'wlopm --off \*' resume 'wlopm --on \*' & inside ~/.config/labwc/autostart to blank the screen after 10 minutes. The wlopm --off \* command changes the bl_power value to 4, when setting the value to 0 the screen will turn on again. However, swayidle still seems to consider the screen to be off and as a result it will not turn off again unless there is some interaction in the meantime.

When using the MQTT integration, the kiosk application must be able to detect changes made on the device itself. I managed to achieve this for the brightness file by implementing a simple fs.watch(..) file listener. However, I found that it never triggered for the bl_power file. Although the file content changes, none of the filesystem listeners where fired. This could be due to swayidle/wlopm performing write actions at a deeper level that are not detectable by file listeners. As a result, I went for a polling solution, checking the state of both files every 500 milliseconds for any changes. While I understand this is not ideal, it's necessary to ensure proper functionality.

Keep in mind that default arguments are stored as plain text in ~/.config/touchkio/Arguments.json. This file also includes the MQTT user password, which is somewhat obfuscated/encrypted, but in a way that it could be easily reverse engineered. Implementing stronger security measures would complicate the setup process and could discourage some users from configuring the application properly. When using the kiosk application without initializing the default arguments, you will need to provide them with every command. This means that the password may be stored as plain text in various files, such as touchkio.service, ~/.bash_history, etc.

To resolve the issue where the first touch on a turned-off screen triggers a click event (potentially activating Home Assistant actions), I implemented a workaround. When the screen turns off, I remove focus from the kiosk window. This way, the first click only turns the screen on and focuses the window, allowing subsequent clicks to work as expected.

Additionally, to address the problem that scrolling on the Raspberry Pi only works with the web browser scrollbar on the right, I configured the Electron app to simulate a touch device using Emulation.setEmitTouchEventsForMouse. This adjustment provides a user experience similar to that of a proper mobile device.

Electron apps are known to be resource intensive due to their architecture and the inclusion of a full web browser environment. If you just run the kiosk application without other heavy loads, everything should run smoothly.

Issues

  • Currently, only the Raspberry Pi 5 in combination with the new touch display 2 supports the MQTT integration.
    • Other variations, such as Raspberry Pi 4 or Raspberry Pi 5 with touch display 1, could likely be added, but this will require additional testing and contributions via pull requests.
  • You can use Raspberry Pi's build-in screen blanking functionality, however, if the screen is turned on through Home Assistant after being automatically turned off, it will remain on indefinitely.
    • It's recommended to either use the built-in screen blanking feature or implement a Home Assistant automation (e.g., presence detection) to manage the screen status.
  • Hyperlinks that redirect the browser away from the main window are intentionally disabled.
    • This decision was made to maintain a clean design, as navigation buttons to move back and forth between different sites were not included.
  • On the Raspberry Pi terminal you may see some ERROR:gbm_wrapper.cc messages.
    • This appears to be a known issue that currently lacks a fix, but the webview still works.

Credits

Inspired by the one and only Raspberry Pi Master Lord, Jeff Geerling.

License

MIT

About

Home Assistant Touch Panel Kiosk application for a Linux device (e.g. Raspberry Pi) with Touch Display.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • JavaScript 88.6%
  • HTML 7.6%
  • Shell 3.8%