A small Go daemon that reads notifications from Kodi/XBMC via the JSON-RPC socket, and performs actions based on those notifications.

I wrote this primarily because the Python callback interface can get blocked very easily by any add-on, which results in heavy delays getting callbacks executed (for example, using the service.xbmc.callbacks plugin), whereas notifications are shipped over the JSON interface immediately.

This is not an issue with service.xbmc.callbacks (good work pilluli!), but with the Kodi add-on infrastructure, and with other individual add-ons.

This daemon also aims to provide more flexibility. The targeted backends are hyperion, kodi, lifx and shell.

Due to the name change to keep in line with XBMC's new name Kodi, if you have the previous package installed as xbmc-callback-daemon, you should uninstall it before installing the new kodi-callback-daemon. The init scripts should handle your existing /etc/xbmc-callback-daemon.json config file, but you're encouraged to move this to the new location /etc/kodi-callback-daemon.json as this feature may go away in future to avoid confusion.

The Hyperion backend submits callbacks via the JSON interface. This interface is also used by the hyperion-remote command-line utility. There's no end-user documentation for this interface, so when writing callbacks, your best bet is to simply read the JSON schemas in the source tree.

The Kodi backend submits callbacks via the JSON-RPC interface. There is excellent documentation available in the Kodi wiki.

The LIFX backend submits callbacks via the LIFX LAN protocol. Callbacks are written in a local format, detailed below

The shell backend simply executes a command on the system with specified arguments.

Grab the Latest Release as a compiled binary and either install it using your package manager (Debian/Ubuntu/derivs, via the .deb package), or extract kodi-callback-daemon to somewhere on your path (eg - /usr/local/bin) on Linux/OSX/FreeBSD, or where ever on Windows.

Alternatively, you may clone this repository and build it yourself.

NB: I've not actually tested Windows/OSX/FreeBSD support at all, feel free to submit bug reports

You must configure Kodi to accept remote control request, by enabling the options:

• Allow programs on this system to control Kodi
• Allow programs on other systems to control Kodi

In Kodi settings at System Settings -> Network -> Remote Control

Linux/OSX/FreeBSD:

/path/to/bin/kodi-callback-daemon /path/to/configFile.json

Windows:

C:\Path\To\kodi-callback-daemon.exe C:\Path\To\configFile.json

The deb packages include SysV and Upstart init scripts and a systemd unit - enable and use them in the standard fashion. You will need to add your configuration file at:

/etc/kodi-callback-daemon.json

Alternatively, you may edit /etc/default/kodi-callback-daemon and set the path to your configuration file there.

You can find the SysV init script at kodi-callback-daemon.init, and the systemd unit at kodi-callback-daemon.service. Place the SysV script in /etc/init.d/, or the systemd unit at /etc/systemd/system/, and enable/start the service as you normally would.

You might alternatively start the daemon from Kodi's autostart.py. Simply edit userdata/autoexec.py in your Kodi directory (ie ~/.kodi/userdata/autoexec.py on *nix systems), and add the following:

Linux/OSX/FreeBSD:

import kodi
import subprocess

subprocess.Popen(['/path/to/bin/kodi-callback-daemon', '/path/to/configFile.json'])

Windows:

import kodi
import subprocess

subprocess.Popen(['C:\\Path\\To\\kodi-callback-daemon.exe', 'C:\\Path\\To\\configFile.json'])

Note the double-slashes necessary for escaping the Windows paths in Python strings.

Note: If you're using this method, you'll also want to make sure that the daemon is killed on Kodi exit or startup, otherwise you'll get multiple copies running and they'll fight for resources. I'm not a Python guy, so I'm open to suggestions on how to best handle this.

If you have questions on how to use the daemon, you may post them in the Kodi forum thread.

The configuration file is written in JSON (I know, JSON is awful for configuration, but since we're passing JSON messages everywhere, it makes the most sense here), and has the following top-level members:

• kodi object defines the Kodi connection parameters (required)
• hyperion object defines the hyperion connection parameters (optional, but required if you're using the Hyperion backend)
• lifx object defines the lifx connection parameters (optional, but required if you're using the LIFX backend)
• debug boolean enables debug logging (optional)
• callbacks object (required, or nothing will be done!).

See the config.example.json for my Hyperion/LIFX setup, which uses most of the available features.

Specify your Kodi/XBMC IP address and port for the JSON interface in the kodi property:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  }
}

If you're using the Hyperion backend, specify your Hyperion address and port for the JSON interface in the hyperion property:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "hyperion": {
    "address": "127.0.0.1",
    "port": 19444
  }
}

If you're using the LIFX backend, you just need to declare it as an object.

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "lifx": {}
}

You can optionally include a timeout for operations, if you're finding it unreliable (LIFX lights can sometimes be slow). The timeout is a string with the format <number><unit>, ie, for 5 seconds:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "lifx": {
    "timeout": "5s"
  }
}

You can enable debug logging by setting the debug property to true:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "hyperion": {
    "address": "127.0.0.1",
    "port": 19444
  },
  "lifx": {},
  "debug": true
}

The callbacks object is keyed by the Kodi notification method, with each method containing an array of callback objects. There is one exception, which is the Startup method - any callbacks attached to this method will be executed when the daemon starts up. Each callback in the array will be executed sequentially. The callback must contain a backend property, the value of which is one of ["hyperion", "kodi", "lifx", "shell"]. All other properties are backend-specific.

Callbacks using the kodi backend contain the backend property, a method string property containing the Kodi RPC method to call, and a params object property containing the parameters to for the RPC call. An example might look like:

{
  "backend": "kodi",
  "method": "GUI.ShowNotification",
  "params": {
    "title": "Callback Daemon",
    "message": "Hello from the callback daemon!",
    "displaytime": 15000
  }
}

Full example, mixing hyperion, kodi and shell callbacks:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "hyperion": {
    "address": "127.0.0.1",
    "port": 19444
  },
  "callbacks": {
    "Startup": [
      {
        "backend": "hyperion",
        "command": "effect",
        "priority": 86,
        "effect": {
          "name": "Rainbow swirl"
        }
      },
      {
        "backend": "shell",
        "background": true,
        "command": "/bin/echo",
        "arguments": [
          "-e",
          "The 'arguments' property is optional\n"
        ]
      },
      {
        "backend": "kodi",
        "method": "GUI.ShowNotification",
        "params": {
          "title": "Callback Daemon",
          "message": "Hello from the callback daemon!",
          "displaytime": 15000
        }
      }
    ],
    "Player.OnStop": [
      {
        "backend": "hyperion",
        "command": "effect",
        "priority": 86,
        "effect": {
          "name": "Rainbow swirl"
        }
      }
    ]
  }
}

For details on the available RPC methods, see the Kodi wiki page. Unfortunately, because the Kodi JSON-RPC implementation is non-conformant, you will get errors logged when using RPC methods that return strings instead of conformant JSON-RPC objects, but the calls will execute fine.

Callbacks using the hyperion backend contain the backend property, and all other properties are sent verbatim as the request to Hyperion. For example, to execute the Rainbow swirl effect, the callback would look something like this:

{
  "backend": "hyperion",
  "command": "effect",
  "effect": {
    "name": "Rainbow swirl"
  }
}

(see the Hyperion JSON schemas for details on the required fields).

And if we wanted to run this callback on Startup, and on Player.OnStop notifications, our full configuration might look like this:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "hyperion": {
    "address": "127.0.0.1",
    "port": 19444
  },
  "debug": true,
  "callbacks": {
    "Startup": [
      {
        "backend": "hyperion",
        "command": "effect",
        "priority": 86,
        "effect": {
          "name": "Rainbow swirl"
        }
      }
    ],
    "Player.OnStop": [
      {
        "backend": "hyperion",
        "command": "effect",
        "priority": 86,
        "effect": {
          "name": "Rainbow swirl"
        }
      }
    ]
  }
}

Callbacks using the lifx backend contain the backend property, a combination of the following properties:

{
  "backend": "lifx",
  "power": true,
  "powerDuration": "5s",
  "color": {
    "hue": 65535,
    "brightness": 65535,
    "saturation": 65535,
    "kelvin": 2500
  },
  "colorDuration": "5s",
  "lights": [
    "Lounge",
    "Kitchen"
  ]
}

And if we wanted to run this callback on Player.OnPlay, and on Player.OnStop notifications, our full configuration might look like this:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "lifx": {},
  "debug": true,
  "callbacks": {
    "Player.OnPlay": [
      {
        "backend": "lifx",
        "power": false,
        "powerDuration": "2s",
        "color": {
          "hue": 0,
          "brightness": 0,
          "saturation": 65535,
          "kelvin": 3000
        },
        "colorDuration": "2s",
        "lights": [
          "Lounge",
          "Kitchen"
        ]
      }
    ],
    "Player.OnStop": [
      {
        "backend": "lifx",
        "power": true,
        "powerDuration": "2s",
        "color": {
          "hue": 0,
          "brightness": 65535,
          "saturation": 0,
          "kelvin": 3000
        },
        "colorDuration": "2s",
        "lights": [
          "Lounge",
          "Kitchen"
        ]
      }
    ]
  }
}

Callbacks using the shell backend contain the backend property, a command string property containing the path to the executable to be run, an optional arguments array containing a list of arguments to be passed to the command, and an optional background property to allow forking a long-running process without waiting for it to return. An example might look like:

{
  "backend": "shell",
  "background": true,
  "command": "/bin/echo",
  "arguments": [
    "-e",
    "The 'arguments' property is optional\n"
  ]
}

Full example, mixing hyperion and shell callbacks:

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "hyperion": {
    "address": "127.0.0.1",
    "port": 19444
  },
  "callbacks": {
    "Startup": [
      {
        "backend": "hyperion",
        "command": "effect",
        "priority": 86,
        "effect": {
          "name": "Rainbow swirl"
        }
      },
      {
        "backend": "shell",
        "background": true,
        "command": "/bin/echo",
        "arguments": [
          "-e",
          "The 'arguments' property is optional\n"
        ]
      }
    ],
    "Player.OnStop": [
      {
        "backend": "hyperion",
        "command": "effect",
        "priority": 86,
        "effect": {
          "name": "Rainbow swirl"
        }
      }
    ]
  }
}

The Player.OnPlay notification has one additional, optional property available to callbacks: types. This property may contain an array of item types sent with Kodi notifications with this method. At the time of writing, these types are ["movie", "episode", "song"]. Callbacks with a types property will only execute if the played media type matches one of the listed types in the callback. Callbacks with no types property are always executed on Player.OnPlay notifications. The following example increases Hyperion saturation/value, and decreases gamma compensation for music so that visualizations produce punchy lighting effects, and conversely sets much more sedate values for video types. It also executes a clear command on channel 86 when any media is played (types is omitted).

{
  "kodi": {
    "address": "127.0.0.1",
    "port": 9090
  },
  "hyperion": {
    "address": "127.0.0.1",
    "port": 19444
  },
  "debug": true,
  "callbacks": {
    "Player.OnPlay": [
      {
        "types": ["movie", "episode"],
        "backend": "hyperion",
        "command": "transform",
        "transform": {
          "gamma": [2.2, 2.2, 2.8],
          "valueGain": 1.0,
          "saturationGain": 1.0
        }
      },
      {
        "types": ["song"],
        "backend": "hyperion",
        "command": "transform",
        "transform": {
          "gamma": [0.8, 0.8, 0.8],
          "valueGain": 2.0,
          "saturationGain": 2.0
        }
      },
      {
        "backend": "hyperion",
        "command": "clear",
        "priority": 86
      }
    ]
  }
}

See CONTRIBUTING.md for build/contribution details.

We use goop for dependency management, and goxc to produce release builds.