Display comma separated data on Thingsboard in time series charts - csv

Question

I get comma spearated temperature data form a device, where the last entry is the last, the first one is from 1 hour before. The provided data is temperatures from the past hour by minute.
I get json data like this:
temperature 19.2,23.4,18.3 ...... 23.0, 18.2
How to show it in Thingsboard in a time series chart with proper timing?
Thanks!

Answer 1

You might use Rule Engine to convert your JSON in the format you need.
Supposing you are able to send device data to the platform via MQTT API or HTTP API with the following payload:
{
"temp": [22, 3, 45]
}
then with the Script Transformation Node you can convert the payload (the msg field of the POST_TELEMETRY event) to a format like the one below, which can be stored on database and directly shown from timeseries widgets:
[{
"ts": 1618296229874,
"values": {
"temp": 45
}
}, {
"ts": 1618296169874,
"values": {
"temp": 3
}
}, {
"ts": 1618296109874,
"values": {
"temp": 22
}
}]
Also you might need an upstream Switch Node if you have to distinguish between different kinds of telemetry formats. The resulting full working rule chain will be something like this:
Let's say your particular device is provisioned in Thingsboard with device type MULTIVALUE THERMOSTAT, you can configure the Switch node's function as follows:
function nextRelation(metadata, msg) {
return ['other'];
}
if(metadata.deviceType === 'MULTIVALUE THERMOSTAT') {
return ['multivalue'];
}
return nextRelation(metadata, msg);
This is the Script Transformation Node's function:
var tempArray = msg.temp;
var lastTs = Date.now();
var tsCounter = 0;
var MS_IN_ONE_MINUTE = 60000;
var newMsg = [];
for (var i = tempArray.length - 1; i >= 0; i--) {
let ts = {};
ts.ts = lastTs - tsCounter;
tsCounter += MS_IN_ONE_MINUTE;
let values = {};
values.temp = tempArray[i];
ts.values = values;
newMsg.push(ts);
}
return {msg: newMsg, metadata: metadata, msgType: msgType};
The Transformation function is just a starting point. You can improve it or make it more more accurate for you real needs.
In this example I assumed the input payload doesn't contain the base hour so I got it dinamically with Date.now(). So starting from the last telemetry, for all previous ones I calculated the corresponding timestamps going backward in time.