Keep storing statistics for sensors which change device class (#79155)

homeassistant/components/sensor/recorder.py

@@ -156,10 +156,16 @@ def _normalize_states(
     entity_id: str,
 ) -> tuple[str | None, str | None, list[tuple[float, State]]]:
     """Normalize units."""
+    old_metadata = old_metadatas[entity_id][1] if entity_id in old_metadatas else None
     state_unit: str | None = None
 
-    if device_class not in UNIT_CONVERTERS:
+    if device_class not in UNIT_CONVERTERS or (
-        # We're not normalizing this device class, return the state as they are
+        old_metadata
+        and old_metadata["unit_of_measurement"]
+        != UNIT_CONVERTERS[device_class].NORMALIZED_UNIT
+    ):
+        # We're either not normalizing this device class or this entity is not stored
+        # normalized, return the states as they are
         fstates = []
         for state in entity_history:
             try:
@@ -176,10 +182,10 @@ def _normalize_states(
                 if entity_id not in hass.data[WARN_UNSTABLE_UNIT]:
                     hass.data[WARN_UNSTABLE_UNIT].add(entity_id)
                     extra = ""
-                    if old_metadata := old_metadatas.get(entity_id):
+                    if old_metadata:
                         extra = (
                             " and matches the unit of already compiled statistics "
-                            f"({old_metadata[1]['unit_of_measurement']})"
+                            f"({old_metadata['unit_of_measurement']})"
                         )
                     _LOGGER.warning(
                         "The unit of %s is changing, got multiple %s, generation of long term "

tests/components/sensor/test_recorder.py

@@ -2142,9 +2142,9 @@ def test_compile_hourly_statistics_changing_units_3(
 
 
 @pytest.mark.parametrize(
-    "device_class, state_unit, statistic_unit, unit_class, mean, min, max",
+    "device_class, state_unit, statistic_unit, unit_class, mean1, mean2, min, max",
     [
-        ("power", "kW", "W", None, 13.050847, -10, 30),
+        ("power", "kW", "W", None, 13.050847, 13.333333, -10, 30),
     ],
 )
 def test_compile_hourly_statistics_changing_device_class_1(
@@ -2154,7 +2154,8 @@ def test_compile_hourly_statistics_changing_device_class_1(
     state_unit,
     statistic_unit,
     unit_class,
-    mean,
+    mean1,
+    mean2,
     min,
     max,
 ):
@@ -2194,7 +2195,7 @@ def test_compile_hourly_statistics_changing_device_class_1(
                 "statistic_id": "sensor.test1",
                 "start": process_timestamp_to_utc_isoformat(zero),
                 "end": process_timestamp_to_utc_isoformat(zero + timedelta(minutes=5)),
-                "mean": approx(mean),
+                "mean": approx(mean1),
                 "min": approx(min),
                 "max": approx(max),
                 "last_reset": None,
@@ -2204,7 +2205,7 @@ def test_compile_hourly_statistics_changing_device_class_1(
         ]
     }
 
-    # Update device class and record additional states
+    # Update device class and record additional states in the original UoM
     attributes["device_class"] = device_class
     four, _states = record_states(
         hass, zero + timedelta(minutes=5), "sensor.test1", attributes
@@ -2220,6 +2221,65 @@ def test_compile_hourly_statistics_changing_device_class_1(
     # Run statistics again, we get a warning, and no additional statistics is generated
     do_adhoc_statistics(hass, start=zero + timedelta(minutes=10))
     wait_recording_done(hass)
+    statistic_ids = list_statistic_ids(hass)
+    assert statistic_ids == [
+        {
+            "statistic_id": "sensor.test1",
+            "display_unit_of_measurement": state_unit,
+            "has_mean": True,
+            "has_sum": False,
+            "name": None,
+            "source": "recorder",
+            "statistics_unit_of_measurement": state_unit,
+            "unit_class": unit_class,
+        },
+    ]
+    stats = statistics_during_period(hass, zero, period="5minute")
+    assert stats == {
+        "sensor.test1": [
+            {
+                "statistic_id": "sensor.test1",
+                "start": process_timestamp_to_utc_isoformat(zero),
+                "end": process_timestamp_to_utc_isoformat(zero + timedelta(minutes=5)),
+                "mean": approx(mean1),
+                "min": approx(min),
+                "max": approx(max),
+                "last_reset": None,
+                "state": None,
+                "sum": None,
+            },
+            {
+                "statistic_id": "sensor.test1",
+                "start": process_timestamp_to_utc_isoformat(
+                    zero + timedelta(minutes=10)
+                ),
+                "end": process_timestamp_to_utc_isoformat(zero + timedelta(minutes=15)),
+                "mean": approx(mean2),
+                "min": approx(min),
+                "max": approx(max),
+                "last_reset": None,
+                "state": None,
+                "sum": None,
+            },
+        ]
+    }
+
+    # Update device class and record additional states in a different UoM
+    attributes["unit_of_measurement"] = statistic_unit
+    four, _states = record_states(
+        hass, zero + timedelta(minutes=15), "sensor.test1", attributes
+    )
+    states["sensor.test1"] += _states["sensor.test1"]
+    four, _states = record_states(
+        hass, zero + timedelta(minutes=20), "sensor.test1", attributes
+    )
+    states["sensor.test1"] += _states["sensor.test1"]
+    hist = history.get_significant_states(hass, zero, four)
+    assert dict(states) == dict(hist)
+
+    # Run statistics again, we get a warning, and no additional statistics is generated
+    do_adhoc_statistics(hass, start=zero + timedelta(minutes=20))
+    wait_recording_done(hass)
     assert (
         f"The normalized unit of sensor.test1 ({statistic_unit}) does not match the "
         f"unit of already compiled statistics ({state_unit})" in caplog.text
@@ -2244,13 +2304,26 @@ def test_compile_hourly_statistics_changing_device_class_1(
                 "statistic_id": "sensor.test1",
                 "start": process_timestamp_to_utc_isoformat(zero),
                 "end": process_timestamp_to_utc_isoformat(zero + timedelta(minutes=5)),
-                "mean": approx(mean),
+                "mean": approx(mean1),
                 "min": approx(min),
                 "max": approx(max),
                 "last_reset": None,
                 "state": None,
                 "sum": None,
-            }
+            },
+            {
+                "statistic_id": "sensor.test1",
+                "start": process_timestamp_to_utc_isoformat(
+                    zero + timedelta(minutes=10)
+                ),
+                "end": process_timestamp_to_utc_isoformat(zero + timedelta(minutes=15)),
+                "mean": approx(mean2),
+                "min": approx(min),
+                "max": approx(max),
+                "last_reset": None,
+                "state": None,
+                "sum": None,
+            },
         ]
     }
     assert "Error while processing event StatisticsTask" not in caplog.text