Allow multiple observations of same entity (#9391)

* Allow multiple observations of same entity

Why:

* There may be different probabilities for multiple states of the same
entity.

This change addresses the need by:

* Keeping a list of observations for each entity to check on each state
change of the given entity.
* Adding a numeric id to each observation so that they can be
effectively added and removed from `self.current_obs`.
* Adding a test to confirm functionality.

* fix overzealous indenting
This commit is contained in:
Jeff McGehee authored and Martin Hjelmare committed 2017-09-12 18:52:09 +02:00
1 parent c9fc3fae6e
commit 29b62f814f
2 files changed
+81 -10

No files matched your search

@@ -102,7 +102,13 @@ class BayesianBinarySensor(BinarySensorDevice):
self.current_obs = OrderedDict({})
self.entity_obs = {obs['entity_id']: obs for obs in self._observations}
to_observe = set(obs['entity_id'] for obs in self._observations)
self.entity_obs = dict.fromkeys(to_observe, [])
for ind, obs in enumerate(self._observations):
obs["id"] = ind
self.entity_obs[obs['entity_id']].append(obs)
self.watchers = {
'numeric_state': self._process_numeric_state,
@@ -120,16 +126,17 @@ class BayesianBinarySensor(BinarySensorDevice):
if new_state.state == STATE_UNKNOWN:
return
entity_obs = self.entity_obs[entity]
platform = entity_obs['platform']
entity_obs_list = self.entity_obs[entity]
self.watchers[platform](entity_obs)
for entity_obs in entity_obs_list:
platform = entity_obs['platform']
self.watchers[platform](entity_obs)
prior = self.prior
for obs in self.current_obs.values():
prior = update_probability(prior, obs['prob_true'],
obs['prob_false'])
self.probability = prior
self.hass.async_add_job(self.async_update_ha_state, True)
@@ -140,20 +147,20 @@ class BayesianBinarySensor(BinarySensorDevice):
def _update_current_obs(self, entity_observation, should_trigger):
"""Update current observation."""
entity = entity_observation['entity_id']
obs_id = entity_observation['id']
if should_trigger:
prob_true = entity_observation['prob_given_true']
prob_false = entity_observation.get(
'prob_given_false', 1 - prob_true)
self.current_obs[entity] = {
self.current_obs[obs_id] = {
'prob_true': prob_true,
'prob_false': prob_false
}
else:
self.current_obs.pop(entity, None)
self.current_obs.pop(obs_id, None)
def _process_numeric_state(self, entity_observation):
"""Add entity to current_obs if numeric state conditions are met."""