Skip to content

home-assistant/HAKit

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

HAKit

Documentation codecov License Apache 2.0

This library allows you to connect to the Home Assistant WebSocket API to issue commands and subscribe to events as well as issue REST API requests. Future plans include offering authentication support.

API Reference

You can view the full set of documentation. The most important set of available methods are on the HAConnection protocol. This protocol acts as the main entrypoint to talking to a Home Assistant instance.

Creating and connecting

Creating a connection needs two pieces of information: the server URL and an access token. Both are retrieved when a connection attempt is made. Since Home Assistant instances' connectivity may differ based on the current network, these values are re-queried on each connection attempt.

You can get a connection instance using the exposed initializer:

let connection = HAKit.connection(configuration: .init(
  connectionInfo: {
    // Connection is required to be returned synchronously.
    // In a real implementation, handle both URL/connection info without crashing.
    try! .init(url: URL(string: "http://homeassistant.local:8123")!)
  },
  fetchAuthToken: { completion in
    // Access tokens are retrieved asynchronously, but be aware that Home Assistant
    // has a timeout of 10 seconds for sending your access token.
    completion(.success("API_Token_Here"))
  }
))

You may further configure other attributes of this connection, such as callbackQueue (where your handlers are invoked), as well as triggering manual connection attempts. See the protocol for more information.

Once you invoke .connect() (or it is invoked automatically) and until you invoke .disconnect() the connection will try to stay connected by attempting to reconnect when network status changes and after a retry period following disconnection.

Logging

You can enable logging of the library by configuring HAGlobal:

HAGlobal.log = { text in
  // just straight to the console
  print($0)
  // or something like XCGLogger
  log.info("WebSocket: \(text)")
}

This will include things like commands being sent and the lifecycle of the connections.

Sending and subscribing

There are two types of requests: those with an immediate result and those which fire events until cancelled. The actions for these are "sending" and "subscribing." For example, you might send a service call but subscribe to a template's rendering.

Requests issued will continue to retry across reconnects until executed once, and subscriptions will automatically re-register when necessary until cancelled. Each send or subscribe returns an HACancellable token which you can cancel and each subscription handler includes a token as well.

Retrieving the current user for example, like other calls in HATypedRequest and HATypedSubscription, has helper methods to offer strongly-typed values. For example, you could write it one of two ways:

// with the CurrentUser convenience helper
connection.send(.currentUser()) { result in
  switch result {
  case let .success(user):
    // e.g. user.id or user.name are available on the result object
  case let .failure(error):
    // an error occurred with the request
  }
}

// or issued directly, and getting the raw response
connection.send(.init(type: .currentUser, data: [:])) { result in
  switch result {
  case let .success(data):
    // data is an `HAData` which wraps possible responses and provides decoding
  case let .failure(error):
    // an error occurred with the request
  }
}

// you can also issue REST calls
Current.apiConnection.send(.init(
  // this will POST to `/api/template` with the data as the body
  type: .rest(.post, "template"),
  data: ["template": "{{ now() }}"]
)) { result in
  // same result type as sending a WebSocket request
}

Similarly, subscribing to events can be done both using the convenience helper or directly.

// with the RenderTemplate convenience helper
connection.subscribe(
  to: .renderTemplate("{{ states('sun.sun') }} {{ states.device_tracker | count }}"),
  initiated: { result in
    // when the initiated method is provided, this is the result of the subscription
  },
  handler: { [textView] cancelToken, response in
    // the overall response is parsed for type, but native template rendering means
    // the rendered type will be a Dictionary, Array, etc.
    textView.text = String(describing: response.result)
    // you could call `cancelToken.cancel()` to end the subscription here if desired
  }
)

// or issued directly, and getting the raw response
connection.subscribe(
  to: .init(type: .renderTemplate, data: [
    "template": "{{ states('sun.sun') }} {{ states.device_tracker | count }}"
  ]),
  initiated: { result in
    // when the initiated method is provided, this is the result of the subscription
  },
  handler: { [textView] cancelToken, data in
    // data is an `HAData` which wraps possible responses and provides decoding
    // the decode methods infer which type you're asking for and attempt to convert
    textView.text = try? data.decode("result")
    // you could call `cancelToken.cancel()` to end the subscription here if desired
  }
)  

You can also invoke any request or subscription, even those without convenience accessors around their name. The event names and request types conform to ExpressibleByStringLiteral or you can initialize them with a raw value.

Decoding

Many methods will deliver results as HAData when not using convenience wrappers. This is delivered in place of an underlying dictionary or array response and it features convenience methods to decode keys from the dictionaries as particular types. This works similarly to Decodable but not identically -- many Home Assistant calls will return results that must be preserved as Any and Swift's JSON coding does not handle this well.

See HADataDecodable for the available methods.

Cached Results

HACache allows you to cache the result of requests and subscribe to events to update them.

The library includes a few built-in caches which you can access via HACachesContainer via connection.caches. See the documentation for which are available.

Populate

Populating the cache is done using HACachePopulateInfo which contains:

  1. The request to send
  2. The transform converting the request's result to the cached value

For example, if you wanted to issue the request .getStates() and keep track of all entityId:

let populate = HACachePopulateInfo<Set<String>>(
  request: .getStates(),
  transform: { info in
    return Set(info.incoming.map(\.entityId))
  }
)

Subscribe

Updating the value of the cache is done using one or more HACacheSubscribeInfo, each of which contains:

  1. The event to subscribe to
  2. How to convert the event to what the cache should do: reissue the populate request, update the value to a new version, and ignore the event entirely.

For example, if you wanted to watch for state changes to update entityIds:

let subscribe = HACacheSubscribeInfo<Set<String>>(
  subscription: .stateChanged(),
  transform: { info in
    var entityIds = info.current
    if info.incoming.newState == nil {
      entityIds.remove(info.incoming.entityId)
    } else {
      entityIds.insert(info.incoming.entityId)
    }
    return .replace(entityIds)
  }
)

Putting it together

Putting these two together, we can create a cache that maintains the current known entityIds:

let entityIds = HACache(connection: connection, populate: populate, subscribe: subscribe)

You can now subscribe to changes in the cache:

entityIds.subscribe { token, value in
  print("current entity ids are: \(value)")
}

Caches will defer doing their populate until the connection is connected and it has at least one subscriber. You can control whether it disconnects from subscriptions when it has no subscribers via the shouldResetWithoutSubscribers property.

Mocks

This library contains optional additions for writing tests. See the source for more information.

PromiseKit

This library contains optional additions for use with PromiseKit. See the source for more information.

Installation

Swift Package Manager

To install the library, either add it as a dependency in a Package.swift like:

.package(url: "https://github.com/home-assistant/HAKit", from: Version(0, 4, 2))

To add it to an Xcode project, you can do this by adding the URL to File > Swift Packages > Add Package Dependency. You will find a few targets available:

  • HAKit, the library itself
  • HAKit+PromiseKit, which includes the PromiseKit additions
  • HAKit+Mocks, which includes mocks to make writing tests easier

CocoaPods

Add the following line to your Podfile:

pod "HAKit", "~> 0.4.4"
# We are working from a fork of Starscream due to a necessary fix, please specify in your podfile
pod 'Starscream', git: 'https://github.com/bgoncal/starscream', branch: 'ha-URLSession-fix'
# pod "HAKit/PromiseKit" # optional, for PromiseKit support
# pod "HAKit/Mocks" # optional, for tests

Contributing

See CONTRIBUTING.md more information on how to build and modify this library.

License

This library is available under the Apache 2.0 license. It also has an underlying dependency on Starscream for WebSocket connectivity on older versions of iOS. Starscream is also available under the Apache 2.0 license.