core/homeassistant/components/bayesian/binary_sensor.py

419 lines
13 KiB
Python

"""Use Bayesian Inference to trigger a binary sensor."""
from collections import OrderedDict
import logging
import voluptuous as vol
from homeassistant.components.binary_sensor import PLATFORM_SCHEMA, BinarySensorEntity
from homeassistant.const import (
CONF_ABOVE,
CONF_BELOW,
CONF_DEVICE_CLASS,
CONF_ENTITY_ID,
CONF_NAME,
CONF_PLATFORM,
CONF_STATE,
CONF_VALUE_TEMPLATE,
STATE_UNKNOWN,
)
from homeassistant.core import callback
from homeassistant.exceptions import ConditionError, TemplateError
from homeassistant.helpers import condition
import homeassistant.helpers.config_validation as cv
from homeassistant.helpers.event import (
TrackTemplate,
async_track_state_change_event,
async_track_template_result,
)
from homeassistant.helpers.reload import async_setup_reload_service
from homeassistant.helpers.template import result_as_boolean
from . import DOMAIN, PLATFORMS
ATTR_OBSERVATIONS = "observations"
ATTR_OCCURRED_OBSERVATION_ENTITIES = "occurred_observation_entities"
ATTR_PROBABILITY = "probability"
ATTR_PROBABILITY_THRESHOLD = "probability_threshold"
CONF_OBSERVATIONS = "observations"
CONF_PRIOR = "prior"
CONF_TEMPLATE = "template"
CONF_PROBABILITY_THRESHOLD = "probability_threshold"
CONF_P_GIVEN_F = "prob_given_false"
CONF_P_GIVEN_T = "prob_given_true"
CONF_TO_STATE = "to_state"
DEFAULT_NAME = "Bayesian Binary Sensor"
DEFAULT_PROBABILITY_THRESHOLD = 0.5
_LOGGER = logging.getLogger(__name__)
NUMERIC_STATE_SCHEMA = vol.Schema(
{
CONF_PLATFORM: "numeric_state",
vol.Required(CONF_ENTITY_ID): cv.entity_id,
vol.Optional(CONF_ABOVE): vol.Coerce(float),
vol.Optional(CONF_BELOW): vol.Coerce(float),
vol.Required(CONF_P_GIVEN_T): vol.Coerce(float),
vol.Optional(CONF_P_GIVEN_F): vol.Coerce(float),
},
required=True,
)
STATE_SCHEMA = vol.Schema(
{
CONF_PLATFORM: CONF_STATE,
vol.Required(CONF_ENTITY_ID): cv.entity_id,
vol.Required(CONF_TO_STATE): cv.string,
vol.Required(CONF_P_GIVEN_T): vol.Coerce(float),
vol.Optional(CONF_P_GIVEN_F): vol.Coerce(float),
},
required=True,
)
TEMPLATE_SCHEMA = vol.Schema(
{
CONF_PLATFORM: CONF_TEMPLATE,
vol.Required(CONF_VALUE_TEMPLATE): cv.template,
vol.Required(CONF_P_GIVEN_T): vol.Coerce(float),
vol.Optional(CONF_P_GIVEN_F): vol.Coerce(float),
},
required=True,
)
PLATFORM_SCHEMA = PLATFORM_SCHEMA.extend(
{
vol.Optional(CONF_NAME, default=DEFAULT_NAME): cv.string,
vol.Optional(CONF_DEVICE_CLASS): cv.string,
vol.Required(CONF_OBSERVATIONS): vol.Schema(
vol.All(
cv.ensure_list,
[vol.Any(NUMERIC_STATE_SCHEMA, STATE_SCHEMA, TEMPLATE_SCHEMA)],
)
),
vol.Required(CONF_PRIOR): vol.Coerce(float),
vol.Optional(
CONF_PROBABILITY_THRESHOLD, default=DEFAULT_PROBABILITY_THRESHOLD
): vol.Coerce(float),
}
)
def update_probability(prior, prob_given_true, prob_given_false):
"""Update probability using Bayes' rule."""
numerator = prob_given_true * prior
denominator = numerator + prob_given_false * (1 - prior)
return numerator / denominator
async def async_setup_platform(hass, config, async_add_entities, discovery_info=None):
"""Set up the Bayesian Binary sensor."""
await async_setup_reload_service(hass, DOMAIN, PLATFORMS)
name = config[CONF_NAME]
observations = config[CONF_OBSERVATIONS]
prior = config[CONF_PRIOR]
probability_threshold = config[CONF_PROBABILITY_THRESHOLD]
device_class = config.get(CONF_DEVICE_CLASS)
async_add_entities(
[
BayesianBinarySensor(
name, prior, observations, probability_threshold, device_class
)
]
)
class BayesianBinarySensor(BinarySensorEntity):
"""Representation of a Bayesian sensor."""
def __init__(self, name, prior, observations, probability_threshold, device_class):
"""Initialize the Bayesian sensor."""
self._name = name
self._observations = observations
self._probability_threshold = probability_threshold
self._device_class = device_class
self._deviation = False
self._callbacks = []
self.prior = prior
self.probability = prior
self.current_observations = OrderedDict({})
self.observations_by_entity = self._build_observations_by_entity()
self.observations_by_template = self._build_observations_by_template()
self.observation_handlers = {
"numeric_state": self._process_numeric_state,
"state": self._process_state,
}
async def async_added_to_hass(self):
"""
Call when entity about to be added.
All relevant update logic for instance attributes occurs within this closure.
Other methods in this class are designed to avoid directly modifying instance
attributes, by instead focusing on returning relevant data back to this method.
The goal of this method is to ensure that `self.current_observations` and `self.probability`
are set on a best-effort basis when this entity is register with hass.
In addition, this method must register the state listener defined within, which
will be called any time a relevant entity changes its state.
"""
@callback
def async_threshold_sensor_state_listener(event):
"""
Handle sensor state changes.
When a state changes, we must update our list of current observations,
then calculate the new probability.
"""
new_state = event.data.get("new_state")
if new_state is None or new_state.state == STATE_UNKNOWN:
return
entity = event.data.get("entity_id")
self.current_observations.update(self._record_entity_observations(entity))
self.async_set_context(event.context)
self._recalculate_and_write_state()
self.async_on_remove(
async_track_state_change_event(
self.hass,
list(self.observations_by_entity),
async_threshold_sensor_state_listener,
)
)
@callback
def _async_template_result_changed(event, updates):
track_template_result = updates.pop()
template = track_template_result.template
result = track_template_result.result
entity = event and event.data.get("entity_id")
if isinstance(result, TemplateError):
_LOGGER.error(
"TemplateError('%s') "
"while processing template '%s' "
"in entity '%s'",
result,
template,
self.entity_id,
)
should_trigger = False
else:
should_trigger = result_as_boolean(result)
for obs in self.observations_by_template[template]:
if should_trigger:
obs_entry = {"entity_id": entity, **obs}
else:
obs_entry = None
self.current_observations[obs["id"]] = obs_entry
if event:
self.async_set_context(event.context)
self._recalculate_and_write_state()
for template in self.observations_by_template:
info = async_track_template_result(
self.hass,
[TrackTemplate(template, None)],
_async_template_result_changed,
)
self._callbacks.append(info)
self.async_on_remove(info.async_remove)
info.async_refresh()
self.current_observations.update(self._initialize_current_observations())
self.probability = self._calculate_new_probability()
self._deviation = bool(self.probability >= self._probability_threshold)
@callback
def _recalculate_and_write_state(self):
self.probability = self._calculate_new_probability()
self._deviation = bool(self.probability >= self._probability_threshold)
self.async_write_ha_state()
def _initialize_current_observations(self):
local_observations = OrderedDict({})
for entity in self.observations_by_entity:
local_observations.update(self._record_entity_observations(entity))
return local_observations
def _record_entity_observations(self, entity):
local_observations = OrderedDict({})
for entity_obs in self.observations_by_entity[entity]:
platform = entity_obs["platform"]
should_trigger = self.observation_handlers[platform](entity_obs)
if should_trigger:
obs_entry = {"entity_id": entity, **entity_obs}
else:
obs_entry = None
local_observations[entity_obs["id"]] = obs_entry
return local_observations
def _calculate_new_probability(self):
prior = self.prior
for obs in self.current_observations.values():
if obs is not None:
prior = update_probability(
prior,
obs["prob_given_true"],
obs.get("prob_given_false", 1 - obs["prob_given_true"]),
)
return prior
def _build_observations_by_entity(self):
"""
Build and return data structure of the form below.
{
"sensor.sensor1": [{"id": 0, ...}, {"id": 1, ...}],
"sensor.sensor2": [{"id": 2, ...}],
...
}
Each "observation" must be recognized uniquely, and it should be possible
for all relevant observations to be looked up via their `entity_id`.
"""
observations_by_entity = {}
for ind, obs in enumerate(self._observations):
obs["id"] = ind
if "entity_id" not in obs:
continue
entity_ids = [obs["entity_id"]]
for e_id in entity_ids:
observations_by_entity.setdefault(e_id, []).append(obs)
return observations_by_entity
def _build_observations_by_template(self):
"""
Build and return data structure of the form below.
{
"template": [{"id": 0, ...}, {"id": 1, ...}],
"template2": [{"id": 2, ...}],
...
}
Each "observation" must be recognized uniquely, and it should be possible
for all relevant observations to be looked up via their `template`.
"""
observations_by_template = {}
for ind, obs in enumerate(self._observations):
obs["id"] = ind
if "value_template" not in obs:
continue
template = obs.get(CONF_VALUE_TEMPLATE)
observations_by_template.setdefault(template, []).append(obs)
return observations_by_template
def _process_numeric_state(self, entity_observation):
"""Return True if numeric condition is met."""
entity = entity_observation["entity_id"]
try:
return condition.async_numeric_state(
self.hass,
entity,
entity_observation.get("below"),
entity_observation.get("above"),
None,
entity_observation,
)
except ConditionError:
return False
def _process_state(self, entity_observation):
"""Return True if state conditions are met."""
entity = entity_observation["entity_id"]
try:
return condition.state(
self.hass, entity, entity_observation.get("to_state")
)
except ConditionError:
return False
@property
def name(self):
"""Return the name of the sensor."""
return self._name
@property
def is_on(self):
"""Return true if sensor is on."""
return self._deviation
@property
def should_poll(self):
"""No polling needed."""
return False
@property
def device_class(self):
"""Return the sensor class of the sensor."""
return self._device_class
@property
def extra_state_attributes(self):
"""Return the state attributes of the sensor."""
attr_observations_list = [
obs.copy() for obs in self.current_observations.values() if obs is not None
]
for item in attr_observations_list:
item.pop("value_template", None)
return {
ATTR_OBSERVATIONS: attr_observations_list,
ATTR_OCCURRED_OBSERVATION_ENTITIES: list(
{
obs.get("entity_id")
for obs in self.current_observations.values()
if obs is not None and obs.get("entity_id") is not None
}
),
ATTR_PROBABILITY: round(self.probability, 2),
ATTR_PROBABILITY_THRESHOLD: self._probability_threshold,
}
async def async_update(self):
"""Get the latest data and update the states."""
if not self._callbacks:
self._recalculate_and_write_state()
return
# Force recalc of the templates. The states will
# update automatically.
for call in self._callbacks:
call.async_refresh()