Add unit/device_class validation and normalization to Tuya (#57913)

Co-authored-by: Martin Hjelmare <marhje52@gmail.com>

homeassistant/components/tuya/const.py

@@ -1,9 +1,68 @@
 """Constants for the Tuya integration."""
-from dataclasses import dataclass
+from __future__ import annotations
+
+from dataclasses import dataclass, field
 from enum import Enum
+from typing import Callable
 
 from tuya_iot import TuyaCloudOpenAPIEndpoint
 
+from homeassistant.const import (
+    CONCENTRATION_MICROGRAMS_PER_CUBIC_METER,
+    CONCENTRATION_MILLIGRAMS_PER_CUBIC_METER,
+    CONCENTRATION_PARTS_PER_BILLION,
+    CONCENTRATION_PARTS_PER_MILLION,
+    DEVICE_CLASS_AQI,
+    DEVICE_CLASS_BATTERY,
+    DEVICE_CLASS_CO,
+    DEVICE_CLASS_CO2,
+    DEVICE_CLASS_CURRENT,
+    DEVICE_CLASS_DATE,
+    DEVICE_CLASS_ENERGY,
+    DEVICE_CLASS_GAS,
+    DEVICE_CLASS_HUMIDITY,
+    DEVICE_CLASS_ILLUMINANCE,
+    DEVICE_CLASS_MONETARY,
+    DEVICE_CLASS_NITROGEN_DIOXIDE,
+    DEVICE_CLASS_NITROGEN_MONOXIDE,
+    DEVICE_CLASS_NITROUS_OXIDE,
+    DEVICE_CLASS_OZONE,
+    DEVICE_CLASS_PM1,
+    DEVICE_CLASS_PM10,
+    DEVICE_CLASS_PM25,
+    DEVICE_CLASS_POWER,
+    DEVICE_CLASS_POWER_FACTOR,
+    DEVICE_CLASS_PRESSURE,
+    DEVICE_CLASS_SIGNAL_STRENGTH,
+    DEVICE_CLASS_SULPHUR_DIOXIDE,
+    DEVICE_CLASS_TEMPERATURE,
+    DEVICE_CLASS_TIMESTAMP,
+    DEVICE_CLASS_VOLATILE_ORGANIC_COMPOUNDS,
+    DEVICE_CLASS_VOLTAGE,
+    ELECTRIC_CURRENT_AMPERE,
+    ELECTRIC_CURRENT_MILLIAMPERE,
+    ELECTRIC_POTENTIAL_MILLIVOLT,
+    ELECTRIC_POTENTIAL_VOLT,
+    ENERGY_KILO_WATT_HOUR,
+    ENERGY_WATT_HOUR,
+    LIGHT_LUX,
+    PERCENTAGE,
+    POWER_KILO_WATT,
+    POWER_WATT,
+    PRESSURE_BAR,
+    PRESSURE_HPA,
+    PRESSURE_INHG,
+    PRESSURE_MBAR,
+    PRESSURE_PA,
+    PRESSURE_PSI,
+    SIGNAL_STRENGTH_DECIBELS,
+    SIGNAL_STRENGTH_DECIBELS_MILLIWATT,
+    TEMP_CELSIUS,
+    TEMP_FAHRENHEIT,
+    VOLUME_CUBIC_FEET,
+    VOLUME_CUBIC_METERS,
+)
+
 DOMAIN = "tuya"
 
 CONF_AUTH_TYPE = "auth_type"
@@ -155,6 +214,212 @@ class DPCode(str, Enum):
     WORK_MODE = "work_mode"  # Working mode
 
 
+@dataclass
+class UnitOfMeasurement:
+    """Describes a unit of measurement."""
+
+    unit: str
+    device_classes: set[str]
+
+    aliases: set[str] = field(default_factory=set)
+    conversion_unit: str | None = None
+    conversion_fn: Callable[[float], float] | None = None
+
+
+# A tuple of available units of measurements we can work with.
+# Tuya's devices aren't consistent in UOM use, thus this provides
+# a list of aliases for units and possible conversions we can do
+# to make them compatible with our model.
+UNITS = (
+    UnitOfMeasurement(
+        unit="",
+        aliases={" "},
+        device_classes={
+            DEVICE_CLASS_AQI,
+            DEVICE_CLASS_DATE,
+            DEVICE_CLASS_MONETARY,
+            DEVICE_CLASS_TIMESTAMP,
+        },
+    ),
+    UnitOfMeasurement(
+        unit=PERCENTAGE,
+        aliases={"pct", "percent"},
+        device_classes={
+            DEVICE_CLASS_BATTERY,
+            DEVICE_CLASS_HUMIDITY,
+            DEVICE_CLASS_POWER_FACTOR,
+        },
+    ),
+    UnitOfMeasurement(
+        unit=CONCENTRATION_PARTS_PER_MILLION,
+        device_classes={
+            DEVICE_CLASS_CO,
+            DEVICE_CLASS_CO2,
+        },
+    ),
+    UnitOfMeasurement(
+        unit=CONCENTRATION_PARTS_PER_BILLION,
+        device_classes={
+            DEVICE_CLASS_CO,
+            DEVICE_CLASS_CO2,
+        },
+        conversion_unit=CONCENTRATION_PARTS_PER_MILLION,
+        conversion_fn=lambda x: x / 1000,
+    ),
+    UnitOfMeasurement(
+        unit=ELECTRIC_CURRENT_AMPERE,
+        aliases={"a", "ampere"},
+        device_classes={DEVICE_CLASS_CURRENT},
+    ),
+    UnitOfMeasurement(
+        unit=ELECTRIC_CURRENT_MILLIAMPERE,
+        aliases={"ma", "milliampere"},
+        device_classes={DEVICE_CLASS_CURRENT},
+        conversion_unit=ELECTRIC_CURRENT_AMPERE,
+        conversion_fn=lambda x: x / 1000,
+    ),
+    UnitOfMeasurement(
+        unit=ENERGY_WATT_HOUR,
+        aliases={"wh", "watthour"},
+        device_classes={DEVICE_CLASS_ENERGY},
+    ),
+    UnitOfMeasurement(
+        unit=ENERGY_KILO_WATT_HOUR,
+        aliases={"kwh", "kilowatt-hour"},
+        device_classes={DEVICE_CLASS_ENERGY},
+    ),
+    UnitOfMeasurement(
+        unit=VOLUME_CUBIC_FEET,
+        aliases={"ft3"},
+        device_classes={DEVICE_CLASS_GAS},
+    ),
+    UnitOfMeasurement(
+        unit=VOLUME_CUBIC_METERS,
+        aliases={"m3"},
+        device_classes={DEVICE_CLASS_GAS},
+    ),
+    UnitOfMeasurement(
+        unit=LIGHT_LUX,
+        aliases={"lux"},
+        device_classes={DEVICE_CLASS_ILLUMINANCE},
+    ),
+    UnitOfMeasurement(
+        unit="lm",
+        aliases={"lum", "lumen"},
+        device_classes={DEVICE_CLASS_ILLUMINANCE},
+    ),
+    UnitOfMeasurement(
+        unit=CONCENTRATION_MICROGRAMS_PER_CUBIC_METER,
+        aliases={"ug/m3", "µg/m3", "ug/m³"},
+        device_classes={
+            DEVICE_CLASS_NITROGEN_DIOXIDE,
+            DEVICE_CLASS_NITROGEN_MONOXIDE,
+            DEVICE_CLASS_NITROUS_OXIDE,
+            DEVICE_CLASS_OZONE,
+            DEVICE_CLASS_PM1,
+            DEVICE_CLASS_PM25,
+            DEVICE_CLASS_PM10,
+            DEVICE_CLASS_SULPHUR_DIOXIDE,
+            DEVICE_CLASS_VOLATILE_ORGANIC_COMPOUNDS,
+        },
+    ),
+    UnitOfMeasurement(
+        unit=CONCENTRATION_MILLIGRAMS_PER_CUBIC_METER,
+        aliases={"mg/m3"},
+        device_classes={
+            DEVICE_CLASS_NITROGEN_DIOXIDE,
+            DEVICE_CLASS_NITROGEN_MONOXIDE,
+            DEVICE_CLASS_NITROUS_OXIDE,
+            DEVICE_CLASS_OZONE,
+            DEVICE_CLASS_PM1,
+            DEVICE_CLASS_PM25,
+            DEVICE_CLASS_PM10,
+            DEVICE_CLASS_SULPHUR_DIOXIDE,
+            DEVICE_CLASS_VOLATILE_ORGANIC_COMPOUNDS,
+        },
+        conversion_unit=CONCENTRATION_MICROGRAMS_PER_CUBIC_METER,
+        conversion_fn=lambda x: x * 1000,
+    ),
+    UnitOfMeasurement(
+        unit=POWER_WATT,
+        aliases={"watt"},
+        device_classes={DEVICE_CLASS_POWER},
+    ),
+    UnitOfMeasurement(
+        unit=POWER_KILO_WATT,
+        aliases={"kilowatt"},
+        device_classes={DEVICE_CLASS_POWER},
+    ),
+    UnitOfMeasurement(
+        unit=PRESSURE_BAR,
+        device_classes={DEVICE_CLASS_PRESSURE},
+    ),
+    UnitOfMeasurement(
+        unit=PRESSURE_MBAR,
+        aliases={"millibar"},
+        device_classes={DEVICE_CLASS_PRESSURE},
+    ),
+    UnitOfMeasurement(
+        unit=PRESSURE_HPA,
+        aliases={"hpa", "hectopascal"},
+        device_classes={DEVICE_CLASS_PRESSURE},
+    ),
+    UnitOfMeasurement(
+        unit=PRESSURE_INHG,
+        aliases={"inhg"},
+        device_classes={DEVICE_CLASS_PRESSURE},
+    ),
+    UnitOfMeasurement(
+        unit=PRESSURE_PSI,
+        device_classes={DEVICE_CLASS_PRESSURE},
+    ),
+    UnitOfMeasurement(
+        unit=PRESSURE_PA,
+        device_classes={DEVICE_CLASS_PRESSURE},
+    ),
+    UnitOfMeasurement(
+        unit=SIGNAL_STRENGTH_DECIBELS,
+        aliases={"db"},
+        device_classes={DEVICE_CLASS_SIGNAL_STRENGTH},
+    ),
+    UnitOfMeasurement(
+        unit=SIGNAL_STRENGTH_DECIBELS_MILLIWATT,
+        aliases={"dbm"},
+        device_classes={DEVICE_CLASS_SIGNAL_STRENGTH},
+    ),
+    UnitOfMeasurement(
+        unit=TEMP_CELSIUS,
+        aliases={"°c", "c", "celsius"},
+        device_classes={DEVICE_CLASS_TEMPERATURE},
+    ),
+    UnitOfMeasurement(
+        unit=TEMP_FAHRENHEIT,
+        aliases={"°f", "f", "fahrenheit"},
+        device_classes={DEVICE_CLASS_TEMPERATURE},
+    ),
+    UnitOfMeasurement(
+        unit=ELECTRIC_POTENTIAL_VOLT,
+        aliases={"volt"},
+        device_classes={DEVICE_CLASS_VOLTAGE},
+    ),
+    UnitOfMeasurement(
+        unit=ELECTRIC_POTENTIAL_MILLIVOLT,
+        aliases={"mv", "millivolt"},
+        device_classes={DEVICE_CLASS_VOLTAGE},
+        conversion_unit=ELECTRIC_POTENTIAL_VOLT,
+        conversion_fn=lambda x: x / 1000,
+    ),
+)
+
+
+DEVICE_CLASS_UNITS: dict[str, dict[str, UnitOfMeasurement]] = {}
+for uom in UNITS:
+    for device_class in uom.device_classes:
+        DEVICE_CLASS_UNITS.setdefault(device_class, {})[uom.unit] = uom
+        for unit_alias in uom.aliases:
+            DEVICE_CLASS_UNITS[device_class][unit_alias] = uom
+
+
 @dataclass
 class Country:
     """Describe a supported country."""

homeassistant/components/tuya/sensor.py

@@ -31,7 +31,13 @@ from homeassistant.helpers.typing import StateType
 
 from . import HomeAssistantTuyaData
 from .base import EnumTypeData, IntegerTypeData, TuyaEntity
-from .const import DOMAIN, TUYA_DISCOVERY_NEW, DPCode
+from .const import (
+    DEVICE_CLASS_UNITS,
+    DOMAIN,
+    TUYA_DISCOVERY_NEW,
+    DPCode,
+    UnitOfMeasurement,
+)
 
 # All descriptions can be found here. Mostly the Integer data types in the
 # default status set of each category (that don't have a set instruction)
@@ -209,6 +215,7 @@ class TuyaSensorEntity(TuyaEntity, SensorEntity):
 
     _status_range: TuyaDeviceStatusRange | None = None
     _type_data: IntegerTypeData | EnumTypeData | None = None
+    _uom: UnitOfMeasurement | None = None
 
     def __init__(
         self,
@@ -235,6 +242,37 @@ class TuyaSensorEntity(TuyaEntity, SensorEntity):
             elif self._status_range.type == "Enum":
                 self._type_data = EnumTypeData.from_json(self._status_range.values)
 
+        # Logic to ensure the set device class and API received Unit Of Measurement
+        # match Home Assistants requirements.
+        if (
+            self.device_class is not None
+            and description.native_unit_of_measurement is None
+        ):
+            # We cannot have a device class, if the UOM isn't set or the
+            # device class cannot be found in the validation mapping.
+            if (
+                self.unit_of_measurement is None
+                or self.device_class not in DEVICE_CLASS_UNITS
+            ):
+                self._attr_device_class = None
+                return
+
+            uoms = DEVICE_CLASS_UNITS[self.device_class]
+            self._uom = uoms.get(self.unit_of_measurement) or uoms.get(
+                self.unit_of_measurement.lower()
+            )
+
+            # Unknown unit of measurement, device class should not be used.
+            if self._uom is None:
+                self._attr_device_class = None
+                return
+
+            # Found unit of measurement, use the standardized Unit
+            # Use the target conversion unit (if set)
+            self._attr_native_unit_of_measurement = (
+                self._uom.conversion_unit or self._uom.unit
+            )
+
     @property
     def native_value(self) -> StateType:
         """Return the value reported by the sensor."""
@@ -253,7 +291,10 @@ class TuyaSensorEntity(TuyaEntity, SensorEntity):
 
         # Scale integer/float value
         if isinstance(self._type_data, IntegerTypeData):
-            return self._type_data.scale_value(value)
+            scaled_value = self._type_data.scale_value(value)
+            if self._uom and self._uom.conversion_fn is not None:
+                return self._uom.conversion_fn(scaled_value)
+            return scaled_value
 
         # Unexpected enum value
         if (