So, I had an itch. A variety of data coming in from a variety of sources, all over different protocols and methods, with different names, heirarchies, and paths. For a while I’d been pushing the data into MySQL, but this seemed a bit of a hack, and it still didn’t really cure the problem as it was more of a historical record than an attempt at mapping between the source and the destination, as it were.
I’d been fiddling with MQTT for a while, and had started publishing data over it via my personal broker. And then I read Roberts post over at http://blog.hekkers.net/2012/09/18/mqtt-about-dumb-sensors-topics-and-clean-code/ about republishing it in a logical way.
So simple! So, I wrote a republisher. It’s very simple, and it scratches my itch (and hopefully Roberts too). Data comes in via whatever naming scheme or topic you like, and it gets republished in whatever fashion you want. The map is maintained in map.csv, and items are separated by commas. The MQTT spec doesn’t reserve any characters for the topic name, so I’ve used a comma out of CSV convenience.
The code is over at https://github.com/kylegordon/mqtt-republisher