MQTT to Adafruit-IO

Adafruit recently opened up a beta for their IoT Dashboard. This service provides a simple way to aggregate IoT data and create dashboards with various graphs with them. For a good introduction, have a lookat Jeremy Morgan's blog.

I have started a while back to organize my own IoT projects and experiments around an MQTT message broker. This provides a light-weight publish/subscribe message bus for machine to machine communication. (Buzzword Bingo!)

While Adafruit.IO can be accessed through the MQTT protocol, I like to have a little bit more control. The way I use MQTT is by publishing all the raw data to specific topics, then processing it and re-publishing the aggregate data.

For example, suppose in a room I have two sensors, an ambient light sensor and a presence sensor. Both will publish their raw data through MQTT, where it's picked up by an orchestrator that combines it and sends out an on/off command to the room-light topic for that room, to turn on or off the lights.

While it's possible to do all this through the MQTT interface, it's both unnecessary and slow to do so. To publish only selected data, I mocked up a simple Python script that forwards data from specific topics on my local MQTT broker to Adafruit.IO:

import paho.mqtt.client as mqtt  
import urllib, urllib2

aiokey = '<SECRET_KEY_HERE>'

def aio_publish(feed, value):  
    data = urllib.urlencode({'value': value})
    req = urllib2.Request('' % feed, data)
    req.add_header('Content-Type', 'application/x-www-form-urlencoded')
    req.add_header('x-aio-key', aiokey)
    res = urllib2.urlopen(req)

def on_connect(client, data, rc):  

def on_message(client, data, msg):  
    parts = msg.topic.split('/')
    if len(parts) == 3 and parts[1] == 'aio':
        aio_publish(parts[2], str(msg.payload))

if __name__ == "__main__":  
    client = mqtt.Client()
    client.on_connect = on_connect
    client.on_message = on_message

    client.connect("localhost", 1883, 60)

While the code is fairly self-explanatory, in short what it does is (from near the bottom): create an MQTT client that connects to the local broker and process messages for the duration of the program.

When it's successfully connected (on_connect), it'll subscribe to all topics starting with /aio/. For each message it receives (on_message), it'll make sure the topic is in the form /aio/<feed>, then publish it to (aio_publish(feed, value))

In aio_publish the value is first encoded for transferal over HTTP, then a request is prepared to the specific URL, the correct headers are set and finally the request is executed, sending the encoded data off to Adafruit.IO

For production usage, you'd probably want to build something a little more robust, around Adafruit's own API library, but as a proof-of-concept the approach above works fine.