Import old energy readings for use in energy dashboard

CentralCommand · June 1, 2022, 4:24pm

Just wanted to note for any interested developers it seems there is a way for an integration to add data in the past to history without changing the current state of entities. It can be done by using async_add_external_statistics:

github.com

home-assistant/core/blob/74e2d5c5c312cf3ba154b5206ceb19ba884c6fb4/homeassistant/components/recorder/statistics.py#L1335-L1362

    
      
          def async_add_external_statistics(
              hass: HomeAssistant,
              metadata: StatisticMetaData,
              statistics: Iterable[StatisticData],
          ) -> None:
              """Add hourly statistics from an external source.
          
          
    This inserts an add_external_statistics job in the recorder's queue.
              """
              # The statistic_id has same limitations as an entity_id, but with a ':' as separator
              if not valid_statistic_id(metadata["statistic_id"]):
                  raise HomeAssistantError("Invalid statistic_id")
          
          
    # The source must not be empty and must be aligned with the statistic_id
              domain, _object_id = split_statistic_id(metadata["statistic_id"])
              if not metadata["source"] or metadata["source"] != domain:
                  raise HomeAssistantError("Invalid source")
          
          
    for statistic in statistics:
                  start = statistic["start"]

This file has been truncated. show original

There are currently two integrations in core making use of this if you’re looking for an example, Demo:

github.com

home-assistant/core/blob/74e2d5c5c312cf3ba154b5206ceb19ba884c6fb4/homeassistant/components/demo/init.py#L236

    
      
          metadata = {
              "source": DOMAIN,
              "statistic_id": f"{DOMAIN}:temperature_outdoor",
              "unit_of_measurement": "°C",
              "has_mean": True,
              "has_sum": False,
          }
          statistics = _generate_mean_statistics(
              yesterday_midnight, yesterday_midnight + datetime.timedelta(days=1), 15, 1
          )
          async_add_external_statistics(hass, metadata, statistics)
          
          
# Fake yesterday's energy consumption
          metadata = {
              "source": DOMAIN,
              "statistic_id": f"{DOMAIN}:energy_consumption",
              "unit_of_measurement": "kWh",
              "has_mean": False,
              "has_sum": True,
          }
          statistic_id = f"{DOMAIN}:energy_consumption"

And tibber:

github.com

home-assistant/core/blob/74e2d5c5c312cf3ba154b5206ceb19ba884c6fb4/homeassistant/components/tibber/sensor.py#L642

    
      
          else:
              unit = home.currency
          metadata = StatisticMetaData(
              has_mean=False,
              has_sum=True,
              name=f"{home.name} {sensor_type}",
              source=TIBBER_DOMAIN,
              statistic_id=statistic_id,
              unit_of_measurement=unit,
          )
          async_add_external_statistics(self.hass, metadata, statistics)

To be clear, this isn’t a solution to this FR. But if anyone is interested in building an integration to contribute to core or make available in HACS which does import of historical data I think it can be done using this API in core without the need to write a lot of custom SQL.