Home

Awesome

Platypush logo

Build Status Issues Github stars Github forks Last Commit Ask on Lemmy Join chat on Matrix

pip version CodeFactor Contributions License Sponsor Blog Documentation Wiki Join chat on IRC PayPal

<!-- toc --> <!-- tocstop -->

Introduction

Platypush is a general-purpose and extensible platform for automation across multiple services and devices with hundreds of supported integrations.

It enables users to create their own self-hosted pieces of automation based on events (if this happens then do that) and it provides a comprehensive and customizable user interface that collects everything you need to visualize and control under one roof.

It borrows concepts from IFTTT, Tasker and Home Assistant to provide an environment where the user can easily connect things together. It focuses on an automation-as-code and API-first approach, offering power users great flexibility in customizing their routines.

It's built with compatibility and flexibility in mind, and it can easily run on any device that can run a Python interpreter - from a Raspberry Pi, to an old smartphone, to a beefy server.

What it can do

You can use Platypush to do things like:

The full list of available integrations is available at docs.platypush.tech, which also contains a more in-depth wiki on the features supported by the platform.

The wiki is also mirrored on git.platypush.tech.

The blog regularly publishes content with step-by-step tutorials and recipes.

Core concepts

The foundations of Platypush rest on a few simple building blocks that offer great versatility to build arbitrarily complex automation routines:

A few examples

The bulk of the configuration of Platypush lives under the config.yaml file. An extensive config.yaml example is provided in the repo. All the sections are optional - the only one enabled by default is the HTTP server, backend.http, but that is optional too.

Let's take an example where we want to control the following entities:

We'll need the following plugins enabled in the config.yaml:

The documentation pages of these plugins already provide some comprehensive configuration snippets that you can use.

The most basic configuration would be something like this:

# Enable it if you want the enable the HTTP API and the Web interface
backend.http:

light.hue:
  # IP/hostname of the Hue bridge
  bridge: 192.168.1.10
  # Default groups that should be targeted by actions if none is specified
  # (default: all lights/groups)
  groups:
    - Living Room

# Check the plugin documentation on how to get the credentials
assistant.google:

music.mopidy:  # Or music.mpd
  # IP/hostname of the MPD/Mopidy server
  host: 192.168.1.2

Now that we have our integrations configured, let's build some automation routines.

Turn on the lights when I say so

In this case we will have to create a hook that listens to a SpeechRecognizedEvent triggered by the assistant - for example, when we say "OK, Google" followed by "turn on the lights".

We can declare the hook in YAML format directly in the config.yaml, or in one of the files included in it through the include: directive:

event.hook.turn_lights_on_voice_command:
  if:
    type: platypush.message.event.assistant.SpeechRecognizedEvent
    # Note that a minimal regex-like syntax is supported here.
    # This condition matches both a phrase that contains
    # "turn on the lights" and one that contains "turn on lights"
    phrase: "turn on (the)? lights"
  then:
    - action: light.hue.on
      args:
      groups:
        - Living Room

Or we can declare the hook in a Python script - you just have to create a .py file (e.g. lights.py) under a scripts directory located under the same folder as your config.yaml:

from platypush import run, when
from platypush.events.assistant import SpeechRecognizedEvent

@when(SpeechRecognizedEvent, phrase="turn on (the)? lights")
def lights_on_voice_command():  # Also accepts an optional `event` argument
  run('light.hue.on', groups=['Living Room'])

Or, using the get_plugin API:

from platypush import get_plugin, when
from platypush.events.assistant import SpeechRecognizedEvent

@when(SpeechRecognizedEvent, phrase="turn on (the)? lights")
def lights_on_voice_command():
  get_plugin('light.hue').on(groups=['Living Room'])

Play the music when I say so

The approach is similar for a "play the music" voice command. YAML:

event.hook.play_music_voice_command:
  if:
    type: platypush.message.event.assistant.SpeechRecognizedEvent
    phrase: "play (the)? music"
  then:
    - action: music.mopidy.play

Python:

from platypush import run, when
from platypush.events.assistant import SpeechRecognizedEvent

@when(SpeechRecognizedEvent, phrase="play (the)? music")
def lights_on_voice_command():
  run('music.mopidy.play')

Turn on the lights when the sun goes down

This example requires the sun plugin configured:

sun:
  latitude: LAT
  longitude: LONG

You can then simply subscribe to SunsetEvent. YAML:

event.hook.sunset_lights_on:
  if:
    type: platypush.message.event.sun.SunsetEvent
  then:
    - action: light.hue.on

Python:

from platypush import run, when
from platypush.events.sun import SunsetEvent

@when(SunsetEvent)
def sunset_lights_on():
  run('light.hue.on')

Event matching and token extraction through hook templates

You can also operate token extraction from event arguments if the values are strings.

For example, you can use advanced pattern matching and token extraction to create voice assistant hooks that will match a template with parametrized field which will be passed as arguments to your event hook:

from platypush import run, when
from platypush.events.assistant import SpeechRecognizedEvent

@when(SpeechRecognizedEvent, phrase='play ${title} by ${artist}')
def on_music_play_command(event, title, artist):
  results = run(
    'music.mpd.search',
    filter={
      'artist': artist,
      'title': title,
    }
  )

  if results:
    run('music.mpd.play', results[0]['file'])

Complex hook conditions

Your event hooks can include more complex filters too. Structured filters against partial event arguments are also possible, and relational operators are supported as well. For example:

from platypush import when
from platypush.events.sensor import SensorDataChangeEvent

@when(SensorDataChangeEvent, data=1):
def hook_1(event):
    """
    Triggered when event.data == 1
    """

@when(SensorDataChangeEvent, data={'state': 1}):
def hook_2(event):
    """
    Triggered when event.data['state'] == 1
    """

@when(SensorDataChangeEvent, data={
  'temperature': {'$gt': 25},
  'humidity': {'$le': 15}
}):
def hook_3(event):
    """
    Triggered when event.data['temperature'] > 25 and
    event.data['humidity'] <= 15.
    """

The supported relational fields are the same supported by ElasticSearch - $gt for greater than, $lt for lesser than, $ge for greater or equal, $ne for not equal, etc.

Turn off the lights at 1 AM

We can use a cron for this case. YAML:

cron.lights_off_night:
  # Run this every day at 1 AM
  cron_expression: '0 1 * * *'
  actions:
      - action: light.hue.off

Python:

from platypush import cron, run

@cron('0 1 * * *')
def lights_off_night():
  run('light.hue.off')

Greet me with lights and music when I come home

Let's create an at_home procedure for this purpose. We can also use a text-to-speech plugin like the tts plugin (it requires no configuration as it relies on the Google Translate frontend API, but other, more sophisticated plugins are also available) to have a warm voice to welcome us home. YAML:

# Make sure that the sound plugin is also enabled, for audio processing
sound:

procedure.at_home:
  - action: tts.say
    args:
      text: "Welcome home!"

  # Get luminosity data from a sensor - e.g. LTR559
  - action: gpio.sensor.ltr559.get_data

  # If it's lower than a certain threshold, turn on the lights.
  # Note that we can directly access attributes returned by the
  # previous request(s) as local context variables within the
  # procedure/hook/cron. In this case, `light` is an attribute returned
  # on the response of the previous command.

  # Otherwise, you can also use the special `output` variable to get only
  # the response of the latest action, e.g. `output['light']`

  # Also note the use of the special `if ${}` construct. It accepts
  # a snippet of Python code and it can access variables within the
  # current context.
  - if ${light is not None and light < 110}:
      - action: light.hue.on

  - action: music.mopidy.play
    args:
      resource: "uri:to:my:favourite:playlist"

Python:

from platypush import procedure, run

@procedure("at_home")
def at_home_proc():
  run('tts.say', text='Welcome home!')

  luminosity = run('gpio.sensor.ltr559.get_data').get('light', 0)
  if luminosity < 110:
    run('light.hue.on')

  run('music.mopidy.play', resource='uri:to:my:favourite:playlist')

You can then call the procedure from a hook or another script:

from platypush import run

run('procedure.at_home')

Or, from YAML:

procedure.some_other_procedure:
  - action: procedure.at_home

Or using the API (see next section).

HTTP API

Actions and procedures can also be called using the JSON-RPC API exposed by Platypush.

Your configuration requires the backend.http section enabled if you want to use the HTTP API - default listen port: 8008.

After ensuring that the HTTP backend is enabled, head to http://localhost:8008 and register a new user.

Platypush local user registration page

From the Web UI, head to Settings → Tokens, insert your password again and click Generate JWT token.

User token generation UI

Alternatively, you can retrieve a token via HTTP request:

❯ curl -XPOST -H 'Content-Type: application/json' -d '
{
  "username": "$YOUR_USER",
  "password": "$YOUR_PASSWORD"
}' http://localhost:8008/auth

You can then send requests to Platypush using a simple RPC API:

❯ curl -XPOST \
    -d '{"type":"request", "action":"procedure.at_home"}' \
    -H "Authorization: Bearer $YOUR_TOKEN" \
    -H "Content-Type: application/json" \
    http://localhost:8008/execute
❮
{
  "id": "724754df98968247a284557ce32f74bb",
  "type": "response",
  "target": "http",
  "origin": "myhost",
  "_timestamp": 1716575901.046127,
  "response": {
    "output": {
      "success": true
    },
    "errors": []
  }
}

If your procedure returned something, then that will be returned on the API response too, so downstream consumers can use it.

The POST /execute endpoint accepts a payload in the format:

{
  "type": "request",  // Constant
  "action": "<plugin-name>.<action-name>",  // Or procedure.<name>
  "args": {
    "arg1": "arg2",
    // ...
  }
}

In our procedure.at_home example, you can for instance create an automation snippet paired with your phone that runs the routine whenever you arrive home (or your phone does):

  1. Install an app like Tasker to create automation tasks on your Android device.

  2. Install a plugin like AutoLocation to create automation tasks based on your phone's location.

  3. Create a profile that triggers whenever you enter your home location (and/or exit it).

Tasker screenshot showing an At Home/Outside Home pair of profiles

  1. Leverage the HTTP Request Tasker action to send a request to your Platypush API to trigger the routine.

The Execute tab

The Web interface also provides an Execute tab under the menu sidebar. You can use this tab to dynamically discover the actions exposed by various plugins (and also your own procedures):

Screenshot of the Execute tab showing the autocomplete discovery of the actions

Screenshot of the Execute tab showing the automatically generated documentation for a given action and its parameters

Screenshot of the Execute tab showing the output of an action being run

Websocket API

Events

You can subscribe to events generated by the application over the /ws/events Websocket endpoint, and send events to this endpoint too.

This is useful if you want to synchronize Platypush events with another client, or send custom events outside of those native to the application and build custom automation hooks on them.

Sending events:

❯ wscat -H "Authorization: Bearer $YOUR_TOKEN" \
    -c "ws://localhost:8008/ws/events" \
    -w 1 \
    -x '
{
  "type": "event",
  "args": {
    "type": "platypush.message.event.custom.CustomEvent",
    "subtype": "foo",
    "args": {
      "bar": "baz"
    }
  }
}'

Receiving events:

❯ wscat -H "Authorization: Bearer $YOUR_TOKEN" -c "ws://localhost:8008/ws/events"

Actions

You can also send requests to the /ws/requests Websocket endpoint, and get responses asynchronously on the same channel:

❯ wscat -H "Authorization: Bearer $YOUR_TOKEN" \
    -c "ws://localhost:8008/ws/requests" \
    -w 1 \
    -x '{"type": "requests", "action": "procedure.foo.bar"}'

Web hooks

You can use Platypush to expose your custom routines as dynamic Web hooks that can be called by any client.

All you need is to register a listener for a WebhookEvent

from platypush import run, when
from platypush.events.http.hook import WebhookEvent

hook_token = "abcdefabcdef"

# Expose the hook under the /hook/at_home endpoint
@when(WebhookEvent, hook="at_home")
def at_home_webhook(event: WebhookEvent):
    # Unlike the calls to /execute, custom web hooks are unauthenticated.
    # If you want authentication, you'll need to implement your custom logic by
    # parsing the event headers
    if event.headers.get("X-Token") != hook_token:
        # Tuple with <response, http-code, [response-headers]>
        event.send_response(("Unauthorized", 401))
        return

    run('procedure.at_home')

    # Return anything back to the client
    return {'status': 'ok'}

Then you can invoke your custom logic over HTTP:

❯ curl -H 'X-Token: abcdefabcdef' 'http://localhost:8008/hook/at_home'

Entities

Entities are another building block of Platypush. Many integrations will store their state or connected devices in the form of entities - e.g. the sensors detected by the Z-Wave/Zigbee/Bluetooth integration, or the lights connected to a Hue bridge, or your cloud nodes, or your custom Arduino/ESP machinery, and so on.

Entities provide a consistent interface to interact with your integrations regardless of their type and the plugin that handles them. For instance, all temperature sensors will expose the same interface, regardless if they are Bluetooth or Zigbee sensors, and all the media plugins will expose the same interface, regardless if they manage Chromecasts, Kodi, Plex, Jellyfin or a local VLC player.

Once you enable the HTTP backend and a few integrations that export entities and register a user, you can query the detected entities via:

curl -XPOST -H 'Content-Type: application/json' \
    -H "Authorization: Bearer $YOUR_TOKEN" \
    -d '{"type":"request", "action":"entities.get"}' \
    http://localhost:8008/execute

All the entities expose the same interface and can be manipulated through the same API. Also, when an entity is updated it always emits an EntityUpdateEvent, so you can easily create hooks that react to these events and act on multiple types of entities.

If you enabled the HTTP backend, then you can also access all the entities from the home panel of the Web UI.

Screenshot of the entities UI

Screenshot of the entities UI

Screenshot of the application main panel, showing the Bluetooth, Serial, SmartThings and System integrations

Core Installation

System package manager installation

Arch Linux

You can either install the platypush package (for the latest stable version) or the platypush-git package (for the latest git version) through your favourite AUR package manager. For example, using yay:

$ yay platypush
# Or
$ yay platypush-git

The Arch Linux packages on AUR are automatically updated upon new git commits or tags.

Debian/Ubuntu

  1. Add the Platypush APT key to your trusted keyring:
# wget -q -O \
    /etc/apt/trusted.gpg.d/platypush.asc \
    https://apt.platypush.tech/pubkey.txt
  1. Add the Platypush repository to your APT sources:
#  wget -q -O \
    /etc/apt/sources.list.d/platypush.list \
    https://apt.platypush.tech/lists/platypush-<deb_version>-<branch>.list

Where:

For example, to install the latest stable tags on Debian stable:

# wget -q -O \
    /etc/apt/sources.list.d/platypush.list \
    https://apt.platypush.tech/lists/platypush-stable-main.list
  1. Update your repos and install Platypush:
# apt update
# apt install platypush

Fedora

RPM builds targeting the latest Fedora release are automatically built on every push pipeline.

To install Platypush via RPM on Fedora:

# yum config-manager --add-repo https://rpm.platypush.tech/platypush.repo

# yum install platypush
# Or
# yum install platypush-git

pip

$ pip install platypush

Or, for the latest git version:

# Official repo
$ pip install git+https://git.platypush.tech/platypush/platypush
# Github mirror
$ pip install git+https://github.com/blacklight/platypush

Docker

$ git clone https://git.platypush.tech/platypush/platypush.git
$ cd platypush
# Copy .env.example to .env and edit docker-compose.yml if required.
# In particular, you may want /etc/platypush and /var/lib/platypush
# to point to directories on your hosts
$ docker compose up

Note that the default Dockerfile uses Alpine, but in docker-compose.yml you can also specify an alternative Dockerfile - Debian, Ubuntu and Fedora are supported.

Manual installation

$ git clone https://git.platypush.tech/platypush/platypush.git
$ cd platypush
$ pip install .

Plugins installation

All the plugins included in the main repo will be available once you have installed the core platform.

However, some plugins may require extra (optional) dependencies. You have several ways of installing those dependencies:

pip

You can install extra dependencies via pip extras:

pip install 'platypush[plugin1,plugin2,...]'

For example:

pip install 'platypush[light.hue,music.mpd,rss]'

Will install Platypush with the dependencies for the light.hue, music.mpd and rss plugins.

Web interface

Plugins can be installed from the Web interface too. Navigate to the Extensions entry in the sidebar, select the extension that you want to install, select the Install tab and click Install.

Screenshot of the extensions installation Web view

This section also includes the Configuration tab, with a ready-to-paste configuration snippet template for that plugin, as well as a documentation page that includes all the actions supported by a given plugin and the events it triggers.

Docker

If you already have the base installation of Platypush on your machine, and you have a configuration file with a custom set of integrations, then you may opt to generate a custom Docker image from your configuration file, with all the extra dependencies configured, using the platydock command.

The following command:

❯ platydock -c /path/to/your/config.yaml -d platypush-test

Will create a Platypush Docker image for a device with ID platypush-test, with all the requirements for the additional integrations listed in config.yaml.

You can pass the --print option if you just want to print the content of the output Dockerfile instead of generating the image.

By default the image will use Alpine Linux as a base. You can use the -i/--image to specify another supported base image - ubuntu, debian or fedora.

Virtual environment

If you already have the base installation of Platypush on your machine, and you have a configuration file with a custom set of integrations, then you may opt to generate a custom virtual environment from your configuration file, with all the extra dependencies configured, using the platyvenv command.

The following command:

❯ platyvenv -c /path/to/your/config.yaml -o /path/to/your/venv

Will create a new virtual environment under /path/to/your/venv using the specified config.yaml to determine which optional dependencies should be installed.

You can then run Platypush after activating your new environment:

❯ source /path/to/your/venv/bin/activate
❯ platypush -c /path/to/your/config.yaml

Manual installation

The plugin/backend documentation reports all the dependencies required by each plugin, as well as the commands to install them on multiple platforms.

If you want to customize your installation, or if you need to install dependencies for a plugin that requires some manual steps, you can check out any plugin-specific installation steps from its documentation.

Configuration

Configuration file

You can use the default config.yaml as a template/reference.

The location of the config.yaml to be used by the application is determined in the following way:

  1. It can be passed through the command-line -c/--config argument.
  2. If not specified via -c, it will be read from the PLATYPUSH_CONFIG environment variable.
  3. If not specified, use ./config.yaml if available.
  4. If not available, and you are running Platypush within a Docker container, or as a privileged user (and usually you shouldn't), or as a systemd service created by a supported package manager, then /etc/platypush/config.yaml will be used if available.
  5. Otherwise, if you are running Platypush as a non-privileged user or in a virtual environment, $XDG_CONFIG_HOME/platypush/config.yaml will be used (defaults to ~/.config/platypush/config.yaml).

Scripts directory

By default, any custom Python scripts will be searched under <CONFDIR>/scripts, where <CONFDIR> is the path to your config.yaml.

You can override it in your config.yaml:

scripts_dir: /path/to/custom/scripts

Since everything under the scripts directory will be imported as a submodule, you can create your own libraries of scripts that can import other scripts:

# Content of scripts/music.py

from platypush import run

def music_play(plugin='music.mopidy', resource=None):
  run(f'{plugin}.play', resource)

# Content of scripts/lights.py

from platypush import run

def lights_toggle(plugin='light.hue', groups=('Living Room',)):
  run(f'{plugin}.toggle', groups=groups)

# Content of scripts/home.py

from platypush import procedure

from scripts.music import music_play
from scripts.lights import lights_toggle

@procedure
def at_home():
  music_play()
  lights_toggle()

Splitting configuration on multiple files

The config.yaml file can become very complex, especially if you embed many hooks and procedures in it in YAML format.

To make the configuration more maintainable, and also to isolate modules that you can reuse across multiple instances, you can leverage the include directive:

# All paths are relative to config.yaml, or to the location of the current file
include:
  - assistant.yaml
  - db.yaml
  - media.yaml
  - mqtt.yaml
  - sensors.yaml
  # ...

Working directory

This is where the application will store its data and integration plugins will store their data. The order of precedence is:

Database

The application stores entities, variables, users, integrations state and more on a database. The engine configuration supports the SQLAlchemy engine syntax.

Note: The application uses a local SQLite database by default, which is natively supported by SQLAlchemy. The application has also been tested against MySQL/MariaDB and Postgres, and should work fine with any modern relational database supported by SQLAlchemy. However, any backend other than SQLite may require an additional Python dependency for the SQLAlchemy driver (for example pg8000 for PostgreSQL).

Order of precedence for the engine:

Device ID

The device ID is a unique identifier for a Platypush instance on a network and is used to reliably dispatch messages when multiple instances use a shared backend.

The order of precedence is:

systemd service

If you installed Platypush from a system package manager then you'll also have a systemd service installed for it.

You can start/enable Platypush like any other systemd service:

# systemctl start platypush
# systemctl enable platypush

Or, if you want to run the Platypush service as a generic user:

❯ systemctl --user start platypush
❯ systemctl --user enable platypush

Otherwise, you can create your own systemd service copying the provided .service file to e.g. ~/.config/systemd/user or /etc/systemd/system.

Redis

Platypush uses Redis as a in-memory queue to deliver messages and as a pub/sub bus for inter-process communication.

If you installed Platypush through a package manager, then the Redis service will automatically be installed and started if you launch the Platypush service as a privileged user.

If you run Platypush in a container then by default it'll start its own Redis instance through the --start-redis command-line option.

You can customize the Redis configuration through the:

  1. --redis-host, --redis-port and --redis-queue command-line options.
  2. PLATYPUSH_REDIS_HOST, PLATYPUSH_REDIS_PORT and PLATYPUSH_REDIS_QUEUE environment variables.
  3. Through your config.yaml:
# See https://redis-py.readthedocs.io/en/latest/connections.html#redis.Redis
# for the full list of supported parameters
redis:
  host: redis-host
  port: 6379
  username: redis-user
  password: redis-pass

If --start-redis is set, the application can be configured to start a custom redis-server executable through the:

  1. --redis-bin command-line option.
  2. PLATYPUSH_REDIS_BIN environment variable.

Alternative drop-in implementations such as keydb-server, valkey or redict are also supported.

nginx

If you want to access your Platypush web panel outside your home network, it may be a good idea to use an nginx/Apache reverse proxy with a valid SSL certificate (e.g. managed by certbot). A sample an nginx configuration is provided in the repository.

The Web interface

Other Web panels

Besides the built-in panels that we've already seen in the other sections, Several integrations add their own feature-rich panels to the Web view, turning Platypush into a gateway to all of your services - from Zigbee sensors, to media players and services, to your music cloud, and more.

For example, the music view is available to most of the music plugins.

Screenshot of one of the music panels

Screenshot of the Snapcast panel, which can be used to synchronize your music streams across multiple devices

Another example is the camera panel, to monitor your cameras, get stand-alone feed URLs, and take photos. This becomes available in the UI if you enable at least a camera plugin.

Camera panel screenshot 1

If you enabled at least one local media plugin (like media.vlc, media.mplayer etc.) then you'll also unlock the media UI, which allows you to index, search, view and cast media files under the configured media_dirs, and it also integrates with other configured/supported backends such as YouTube, Plex and Jellyfin.

Media panel screenshot 1

Dashboards

The web service also provides means for the user to create custom dashboards that can be used to show information from multiple sources on a large screen.

Screenshot of a Platypush dashboard, showing a calendar widget, the current music state, weather, news from the RSS integration, and a carousel of custom pictures.

PWA support

Note that having the web application served over SSL is a requirement for the PWA (progressive web app) to work. The Platypush PWA allows you to install a Platypush native-like client on your mobile devices if you don't want to use the full Android app.

Two-factor authentication

Support for 2FA over OTP codes requires to enable the otp and qrcode plugins.

After installing the dependencies, you can enable it by navigating to Settings -> Users from the Web panel. Then select your user, choose Set up 2FA and proceed with the steps on screen to set up your authenticator.

Mobile app

An official Android app is provided on the F-Droid store. It allows to easily discover and manage multiple Platypush services on a network through the web interface, and it easily brings the power of Platypush to your fingertips.

Browser extension

A browser extension is available for Chrome and Firefox.

The browser extension allows you to run Platypush actions and procedures directly from your browser, associate keybindings with them, so you can run your favourite routines with a few keystrokes anywhere in your browser, and provides an advanced API to interact with the Web pages you visit - for example, you can build an action that gets the content of a page you're visiting and uses Platypush to distill it in readable format, or send the URL to another service.

Tests

To run the tests simply run pytest either from the project root folder or the tests/ folder.