What

flower is a service that implements the InfluxDB / Flux API, and executes it as SignalFlow instead of Flux.

It talks to an InfluxDB backend. This could eventually be pluggable.

The service is configured with the following environment variables:

Variable	Usage
FLOWER_LISTEN_ADDRESS	The address to listen on, eg :8123
FLOWER_INFLUX_HOST	The InfluxDB host to talk to, eg https://influxdb.local/
FLOWER_INFLUX_TOKEN	The InfluxDB token to authenticate with, eg foo==
FLOWER_INFLUX_ORG	The InfluxDB org to authenticate to, eg local

Why

I like SignalFlow.

Querying

This implements data() and a handful of stream methods. The SignalFlow parser/evaluator is mostly accurate, lacking imports and .map().

Differences

Filtering

• partition_filter() is not implemented.
• regex_filter() is an additional filter function.

Specifying the time range

This is being tested against Grafana. Because the InfluxDB plugin in Grafana does not pass a time range, it is necessary to explicitly pass it in the query. This is done with the special window() function:

window("v.timeRangeStart", "v.timeRangeStop")

Grafana will inject the suitable time values in to the program. If this is not provided, the query will run from -1h to the current time.

Example

Queries something from Home Assistant exporting to InfluxDB.

window('v.timeRangeStart', 'v.timeRangeStop')

A = (
  data(
    'W',
    filter=filter('entity_id', 'power_plug_*1234*'),
    rollup='max',
  )
  .publish('W')
)

B = (
  A
  .mean(over='5m')
  .publish('W smoothed')
)