Missing State Restoration in TimedPresetManager

[TTLucian]

TL;DR
Timed presets (like Boost) currently remain active indefinitely after a restart because the manager does not implement a restoration hook. Adding a restore_state method allows the manager to re-populate its internal timer from existing entity attributes and resume the countdown.

While BaseThermostat.get_my_previous_state (line 475) correctly iterates through all feature managers to call manager.restore_state(old_state), this method is currently missing from FeatureTimedPresetManager.py.

Consequently, the _timed_preset_end_time and the async_track_point_in_time task are lost during a Home Assistant restart. Since the timer is never rescheduled, the thermostat remains "stuck" in the temporary preset.

By implementing the restore_state method in FeatureTimedPresetManager.py, we can re-populate the manager's state using the data already persisted in the entity attributes.

@callback
    def restore_state(self, last_state: Any):
        """Implement the restoration hook to re-populate timed presets after restart."""
        
        # 1. Retrieve the persistence dictionary from attributes
        # Matches the key used in add_custom_attributes
        manager_attr = last_state.attributes.get("timed_preset_manager")
        if not manager_attr or not manager_attr.get("is_active"):
            return

        end_time_str = manager_attr.get("end_time")
        preset_str = manager_attr.get("preset")

        if not end_time_str or not preset_str:
            return

        try:
            # 2. Re-parse the end_time and preset mode
            end_time = datetime.fromisoformat(end_time_str)
            now = self._vtherm.now
            
            # 3. Re-populate internal manager variables
            self._is_timed_preset_active = True
            self._timed_preset = VThermPreset(preset_str)
            self._timed_preset_end_time = end_time

            # 4. Reschedule the expiration task if time remains
            if end_time > now:
                _LOGGER.info("%s - Resuming timed preset %s. Reverting at %s", self, preset_str, end_time)
                self._cancel_timer = async_track_point_in_time(
                    self._hass,
                    self._async_timed_preset_expired,
                    self._timed_preset_end_time,
                )
            else:
                _LOGGER.info("%s - Timed preset expired during downtime. Cleanup will follow.", self)
                # The existing safety check in refresh_state() will handle the cleanup 
                # and revert to requested_state during the startup cycle.
        except (ValueError, TypeError) as err:
            _LOGGER.error("%s - Failed to restore timed preset state: %s", self, err)

This gives us:

• Architecture Compliance: Plugs directly into the existing for manager in self._managers: manager.restore_state(old_state) loop in base_thermostat.py.
• Resiliency: Ensures that "Boost" or "Temporary" modes don't run forever if a server update or power flicker occurs.
• Safety: Uses the manager's existing refresh_state() to gracefully return the thermostat to its normal schedule if the time expired while the system was offline.
• Minimal Footprint: Reuses the data already being generated by add_custom_attributes.

[jmcollin78]

Hello, yes you are right. I turn this into an issue. Thank you for your report.