Use the following guide to understand how various Automation DSL nodes may be used to construct an automation.
All automation DSL is placed within a single automation
node. The
automation
node forms the boundary between the outer Kotlin language
context and the embedded DSL context.
Sequential flow
The sequential flow is the default type of automation flow.
Here's a very basic Automation DSL template that uses a sequential flow consisting of a starter, a condition, and an action:
import com.google.home.automation.action
import com.google.home.automation.automation
import com.google.home.automation.condition
import com.google.home.automation.sequential
import com.google.home.automation.starter
...
automation {
sequential {
starter<_>(...)
condition {...}
action {...}
}
}
This can be refined by adding additional nodes.
Starter
Starter nodes define the initial circumstances that activate an automation. For example, a change in state or value. An automation must have at least one starter, otherwise it will fail validation. In order to add more than one starter to an automation, you must use a select node.
Starter based on trait attribute
When declaring a starter node that's based on a trait attribute, specify:
- the device
- the device type to which the trait belongs
- the trait
starter<_>(thermostat, TemperatureSensorDevice, TemperatureMeasurement)
The device type parameter is required because it lets you specify which device
type within a device the automation addresses. For example, a device might be
composed of a
FanDevice
and a
HeatingCoolingUnitDevice
,
both of which contain the
OnOff
trait. By specifying the device type, there is no ambiguity about which part of
the device triggers the automation.
Starter based on event
When declaring a starter node that's based on an event, specify:
- the device
- the device type to which the trait belongs
- the event
starter<_>(doorBell, GoogleDoorbellDevice, DoorbellPressed)
Starter based on a structure and event, with parameters
Some events can have parameters, so these parameters also need to be included in the starter.
For example, this starter uses the
Time
trait's ScheduledTimeEvent
to activate the automation at 7:00am:
val earlyMorning = starter<_>(structure, Time.ScheduledTimeEvent) {
parameter(Time.ScheduledTimeEvent.clockTime(
LocalTime.of(7, 0, 0, 0)))
}
Manual starter
A manual starter is a special type of starter that lets the user manually run the automation.
When declaring a manual starter:
- Don't specify a trait or device type.
- Provide a UI element that calls
Automation.execute()
.
When placing a manual starter in a select
flow along with another starter, the
manual starter overrides the other starter:
select {
manualStarter()
starter<_>(thermostat, TemperatureSensorDevice, TemperatureMeasurement)
}
Note that any condition
nodes following a manual starter will be evaluated,
and could block execution of the automation, depending on the condition
expression.
One way to structure your automation so that condition
nodes don't block an
automation that was activated with a manual starter is to put the other starter
in a separate sequential flow along with its condition
:
automation_graph {
sequential {
select {
sequential {
starter<_>(...)
condition {...}
}
sequential {
manualStarter()
}
}
action {...}
}
}
Reference the value of an attribute
To use the value of an attribute in an expression, use the following syntax.
With a stateReader
:
val time = stateReader<_>(structure, Structure, Time)
val currTime = time.currentTime
With a starter
:
val starterNode = starter<_>(device1, LaundryWasherDevice, OnOff)
condition() {
expression = starterNode.onOff equals true
}
Condition nodes and expressions
A condition node represents a decision point that determines whether the
automation continues or not. An automation can have multiple condition
nodes.
If any condition
node's expression evaluates to false
, execution of the
entire automation ends.
Within a condition
node, you can combine multiple condition criteria using
various operators, as long as the expression evaluates to a single
boolean value. If the resulting value is true
, the condition is met and the
automation continues execution of the next node. If it's false
, the automation
stops executing at that point.
Expressions are formed similarly to expressions in Kotlin, and may contain primitive values such as numbers, characters, strings, and booleans, as well as Enum values. Grouping sub expressions with parentheses lets you control the order in which they're evaluated.
Here's an example of a condition
which combines multiple subexpressions into a
single expression:
condition() {
val expr1 = starterNode.lockState equals DlLockState.Unlocked
val expr2 = stateReaderNode.lockState equals true
val expr3 = occupancySensingDevice.occupied notEquals 0
val expr4 = timeStateReaderNode
.currentTime
.between(
timeStateReaderNode.sunsetTime,
timeStateReaderNode.sunriseTime)
expression = (expr1 and expr2) or (expr3 and expr4)
}
You may reference the value of a trait accessed through a starter:
val starterNode = starter<_>(device, OnOff)
condition() { expression = starterNode.onOff equals true }
stateReader
The other way to reference trait attribute values in a condition
node is with
a stateReader
node.
To do this, first capture the trait attribute value in a stateReader
node. A
stateReader
takes the structure
and the trait as arguments:
import com.google.home.automation.stateReader
...
val filterMonitoringState = stateReader<_>(structure, ActivatedCarbonFilterMonitoring)
Then reference the stateReader
in the condition
node:
condition() {
expression =
filterMonitoringState.changeIndication
.equals(ChangeIndicationEnum.Warning)
}
Using comparison and
logical operators, multiple
stateReaders
may be used in a condition
node:
val armState = stateReader<_>(doorLock, DoorLockDevice, ArmDisarm )
val doorLockState = stateReader<_>(doorLock, DoorLockDevice, DoorLock)
condition() {
expression =
(armState.armState equals true)
and
(doorLockState.lockState equals true)
}
Condition duration
In addition to a boolean expression in a condition, you can specify a timeframe during which the expression must be true in order to run the automation. For example, you can define a condition that fires only if a light has been on for ten minutes.
condition {
expression(lightStateReader.onOff == true)
forDuration(Duration.ofMinutes(10))
}
The duration can range from one to 30 minutes.
Action nodes
The action node is where the work of the automation takes place.
In this example, the action invokes the
AssistantBroadcast
trait's
broadcast()
command:
action(device, SpeakerDevice) {
command(AssistantBroadcast.broadcast("Intruder detected!"))
}
Import statements
When developing automations, it isn't always obvious how to import various elements of the Home APIs into your code.
Trait attributes are imported from the trait's Companion
object:
import com.google.home.matter.standard.OnOff.Companion.onOff
Data structures that are defined by a trait are imported from the trait class whose name ends in "-Trait":
import com.google.home.matter.standard.MediaPlaybackTrait.PlaybackStateEnum
Trait commands are imported from the trait's Companion
object:
import com.google.home.matter.standard.Thermostat.Companion.setTemperatureSetpointHold