Restrict recorder query to include max age (#18231)

* Update query to include maxAge

Updated the query from recorded to include MaxAge if set; reducing the amount of records retrieved that would otherwise be purged anyway for the sensor.

* Added newline in docstring

Added newline in docstring.

* Add test + small fix

Added test to ensure query works correctly
Query should be greater then or equal instead of greater then. Fixed.

* Fixed lint issue

Fixed lint issue.

homeassistant/components/sensor/statistics.py

@@ -247,6 +247,9 @@ class StatisticsSensor(Entity):
         The query will get the list of states in DESCENDING order so that we
         can limit the result to self._sample_size. Afterwards reverse the
         list so that we get it in the right order again.
+
+        If MaxAge is provided then query will restrict to entries younger then
+        current datetime - MaxAge.
         """
         from homeassistant.components.recorder.models import States
         _LOGGER.debug("%s: initializing values from the database",
@@ -254,7 +257,17 @@ class StatisticsSensor(Entity):
 
         with session_scope(hass=self._hass) as session:
             query = session.query(States)\
-                .filter(States.entity_id == self._entity_id.lower())\
+                .filter(States.entity_id == self._entity_id.lower())
+
+            if self._max_age is not None:
+                records_older_then = dt_util.utcnow() - self._max_age
+                _LOGGER.debug("%s: retrieve records not older then %s",
+                              self.entity_id, records_older_then)
+                query = query.filter(States.last_updated >= records_older_then)
+            else:
+                _LOGGER.debug("%s: retrieving all records.", self.entity_id)
+
+            query = query\
                 .order_by(States.last_updated.desc())\
                 .limit(self._sampling_size)
             states = execute(query)

tests/components/sensor/test_statistics.py

@@ -3,6 +3,7 @@ import unittest
 import statistics
 
 from homeassistant.setup import setup_component
+from homeassistant.components.sensor.statistics import StatisticsSensor
 from homeassistant.const import (
     ATTR_UNIT_OF_MEASUREMENT, TEMP_CELSIUS, STATE_UNKNOWN)
 from homeassistant.util import dt as dt_util
@@ -233,3 +234,61 @@ class TestStatisticsSensor(unittest.TestCase):
         # check if the result is as in test_sensor_source()
         state = self.hass.states.get('sensor.test_mean')
         assert str(self.mean) == state.state
+
+    def test_initialize_from_database_with_maxage(self):
+        """Test initializing the statistics from the database."""
+        mock_data = {
+            'return_time': datetime(2017, 8, 2, 12, 23, 42,
+                                    tzinfo=dt_util.UTC),
+        }
+
+        def mock_now():
+            return mock_data['return_time']
+
+        # Testing correct retrieval from recorder, thus we do not
+        # want purging to occur within the class itself.
+        def mock_purge(self):
+            return
+
+        # Set maximum age to 3 hours.
+        max_age = 3
+        # Determine what our minimum age should be based on test values.
+        expected_min_age = mock_data['return_time'] + \
+            timedelta(hours=len(self.values) - max_age)
+
+        # enable the recorder
+        init_recorder_component(self.hass)
+
+        with patch('homeassistant.components.sensor.statistics.dt_util.utcnow',
+                   new=mock_now), \
+                patch.object(StatisticsSensor, '_purge_old', mock_purge):
+            # store some values
+            for value in self.values:
+                self.hass.states.set('sensor.test_monitored', value,
+                                     {ATTR_UNIT_OF_MEASUREMENT: TEMP_CELSIUS})
+                self.hass.block_till_done()
+                # insert the next value 1 hour later
+                mock_data['return_time'] += timedelta(hours=1)
+
+            # wait for the recorder to really store the data
+            self.hass.data[recorder.DATA_INSTANCE].block_till_done()
+            # only now create the statistics component, so that it must read
+            # the data from the database
+            assert setup_component(self.hass, 'sensor', {
+                'sensor': {
+                    'platform': 'statistics',
+                    'name': 'test',
+                    'entity_id': 'sensor.test_monitored',
+                    'sampling_size': 100,
+                    'max_age': {'hours': max_age}
+                }
+            })
+
+            # check if the result is as in test_sensor_source()
+            state = self.hass.states.get('sensor.test_mean')
+
+        assert expected_min_age == state.attributes.get('min_age')
+        # The max_age timestamp should be 1 hour before what we have right
+        # now in mock_data['return_time'].
+        assert mock_data['return_time'] == state.attributes.get('max_age') +\
+            timedelta(hours=1)