HomeiOS DevelopmentIterating over net socket messages with async / await in Swift –...

Iterating over net socket messages with async / await in Swift – Donny Wals


Printed on: January 24, 2023

In iOS 13, we gained the flexibility to simply ship and obtain information utilizing net sockets by means of URLSession. With async/await, we gained the flexibility to fetch information from servers utilizing the await key phrase and we are able to iterate over asynchronous sequences utilizing async for loops.

We are able to even learn information from a URL one line at a time by calling the strains property on URL:

let url = URL(string: "https://donnywals.com")!

for strive await line in url.strains {
    // use line
}

Whereas that is actually cool and permits us to construct apps that ingest information in actual time if the server helps streaming our bodies, we can’t use the strains property to arrange an online socket connection and hear for incoming messages and doubtlessly ship messages over the identical connection too.

On this put up, you’ll study every little thing you might want to learn about constructing your personal mechanism to conveniently iterate over messages from an online socket asynchronously. We are going to leverage some current performance from URLSessionWebSocketTask and AsyncThrowingStream to construct our personal AsyncSequence that conveniently wraps our URLSessionWebSocketTask.

Notice that the ensuing code has solely had comparatively restricted testing performed so I can’t assure that the offered resolution can be 100% appropriate for every little thing you throw at it. Should you discover any points with the ultimate code, be happy to contact me. Bonus factors when you’re capable of present some concepts for a possible repair.

Utilizing an online socket with out async / await

Earlier than we get began, let’s shortly evaluation how one can use an online socket with out async/await. The code particulars are outlined in this put up. You’ll want to learn it if you wish to study extra about utilizing net sockets in your apps.


let url = URL(string: "ws://127.0.0.1:8080")!
let socketConnection = URLSession.shared.webSocketTask(with: url)
socketConnection.resume()

func setReceiveHandler() {
    socketConnection.obtain { end in
        defer { self.setReceiveHandler() }

        do {
            let message = strive outcome.get()
            change message {
            case let .string(string):
                print(string)
            case let .information(information):
                print(information)
            @unknown default:
                print("unkown message obtained")
            }
        } catch {
            // deal with the error
            print(error)
        }
    }
}

setReceiveHandler()

Discover how, to obtain messages from the socket, I have to name obtain with a completion handler. This methodology solely permits me to obtain a single incoming message, so I have to re-set my handler after receiving a message to robotically start listening for the following message.

This can be a nice instance of a state of affairs the place an async for loop equivalent to for strive await message in socketConnection would make plenty of sense. Sadly, this isn’t attainable out of the field. Nonetheless, URLSessionWebSocketTask supplies some type of help for async / await so we’re not solely out of luck.

A fundamental implementation of net sockets with async / await

Whereas URLSessionWebSocketTask doesn’t expose an AsyncSequence that emits incoming messages out of the field, it does include an async model of the obtain methodology you noticed earlier.

This permits us to rewrite the instance above as an async methodology as follows:

func setReceiveHandler() async {
    do {
        let message = strive await socketConnection.obtain()

        change message {
        case let .string(string):
          print(string)
        case let .information(information):
          print(information)
        @unknown default:
          print("unkown message obtained")
        }
    } catch {
        print(error)
    }

    await setReceiveHandler()
}

This code works simply wonderful, besides we don’t actually have a method to cease the recursion right here. The code you noticed earlier really has the very same subject; there’s no situation to cease listening for net socket messages even when the net socket connection has already been closed.

We might enhance our code by solely recursing if:

  1. We didn’t encounter any errors
  2. The socket connection remains to be lively

This may look a bit as follows:

func setReceiveHandler() async {
    guard socketConnection.closeCode == .invalid else {
        return
    }

    do {
        let message = strive await socketConnection.obtain()

        change message {
        case let .string(string):
          print(string)
        case let .information(information):
          print(information)
        @unknown default:
          print("unkown message obtained")
        }

        await setReceiveHandler()
    } catch {
        print(error)
    }
}

An open net socket’s closed code is at all times stated to invalid to sign that the connection has not (but) been closed. We are able to leverage this to verify that our connection remains to be lively earlier than ready for the following message to be obtained.

That is significantly better already as a result of we respect closed sockets and failures a lot nicer now, however we might enhance the readability of this code a tiny bit by leveraging a whereas loop as a substitute of recursively calling the setReceiveHandler perform:

func setReceiveHandler() async {
    var isActive = true

    whereas isActive && socketConnection.closeCode == .invalid {
        do {
            let message = strive await socketConnection.obtain()

            change message {
            case let .string(string):
              print(string)
            case let .information(information):
              print(information)
            @unknown default:
              print("unkown message obtained")
            }
        } catch {
            print(error)
            isActive = false
        }
    }
}

To me, this model of the code is barely simpler to learn however that may not be the case for you. It’s functionally equal so you may select to make use of whichever possibility fits you finest.

Whereas this code works, I’m not fairly pleased with the place we’ve landed proper now. There’s plenty of logic on this perform and I would like to separate dealing with the incoming values from the calls to socketConnection.obtain() by some means. Ideally, I ought to be capable of write the next:

do {
    for strive await message in socketConnection {
        change message {
        case let .string(string):
            print(string)
        case let .information(information):
            print(information)
        @unknown default:
            print("unkown message obtained")
      }
} catch {
    // deal with error
}

That is a lot, a lot nicer from a call-site perspective and it might permit us to place the ugly bits elsewhere.

To do that, we are able to leverage the facility of AsyncStream which permits us to construct a customized async sequence of values.

Utilizing AsyncStream to emit net socket messages

Given our finish aim, there are a number of methods for us to get the place we need to be. The best method can be to jot down a perform in an extension on URLSessionWebSocketTask that might encapsulate the whereas loop you noticed earlier. This implementation would look as follows:

typealias WebSocketStream = AsyncThrowingStream<URLSessionWebSocketTask.Message, Error>

public extension URLSessionWebSocketTask {    
    var stream: WebSocketStream {
        return WebSocketStream { continuation in
            Process {
                var isAlive = true

                whereas isAlive && closeCode == .invalid {
                    do {
                        let worth = strive await obtain()
                        continuation.yield(worth)
                    } catch {
                        continuation.end(throwing: error)
                        isAlive = false
                    }
                }
            }
        }
    }
}

To make the code a little bit bit simpler to learn, I’ve outlined a typealias for my AsyncThrowingStream so we don’t have to have a look at the identical lengthy kind signature in every single place.

The code above creates an occasion of AsyncThrowingStream that asynchronously awaits new values from the net socket so long as the net socket is taken into account lively and hasn’t been closed. To emit incoming messages and potential errors, the continuation’s yield and end strategies are used. These strategies will both emit a brand new worth (yield) or finish the stream of values with an error (end).

This code works nice in lots of conditions, however there’s one subject. If we determine to shut the net socket connection from the app’s facet by calling cancel(with:purpose:) on our socketConnection, our WebSocketStream doesn’t finish. As an alternative, it will likely be caught ready for messages, and the decision website can be caught too.

Process {
    strive await Process.sleep(for: .seconds(5))
    strive await socketConnection.cancel(with: .goingAway, purpose: nil)
}

Process {    
    do {
        for strive await message in socketConnection.stream2 {
            // deal with incoming messages
        }
    } catch {
        // deal with error
    }

    print("this might by no means be printed")
}

If every little thing works as anticipated, our net socket connection will shut after 5 seconds. At that time, our for loop ought to finish and our print assertion ought to execute, because the asynchronous stream is not lively. Sadly, this isn’t the case, so we have to discover a higher technique to mannequin our stream.

URLSessionWebSocketTask doesn’t present a method for us to detect cancellation. So, I’ve discovered that it’s best to make use of an object that wraps the URLSessionWebSocketTask, and to cancel the duty by means of that object. This permits us to each finish the async stream we’re offering to callers and shut the net socket reference to one methodology name.

Right here’s what that object appears like:

class SocketStream: AsyncSequence {
    typealias AsyncIterator = WebSocketStream.Iterator
    typealias Aspect = URLSessionWebSocketTask.Message

    non-public var continuation: WebSocketStream.Continuation?
    non-public let process: URLSessionWebSocketTask

    non-public lazy var stream: WebSocketStream = {
        return WebSocketStream { continuation in
            self.continuation = continuation

            Process {
                var isAlive = true

                whereas isAlive && process.closeCode == .invalid {
                    do {
                        let worth = strive await process.obtain()
                        continuation.yield(worth)
                    } catch {
                        continuation.end(throwing: error)
                        isAlive = false
                    }
                }
            }
        }
    }()

    init(process: URLSessionWebSocketTask) {
        self.process = process
        process.resume()
    }

    deinit {
        continuation?.end()
    }

    func makeAsyncIterator() -> AsyncIterator {
        return stream.makeAsyncIterator()
    }

    func cancel() async throws {
        process.cancel(with: .goingAway, purpose: nil)
        continuation?.end()
    }
}

There’s a bunch of code right here, nevertheless it’s not too dangerous. The primary few strains are all about establishing some kind aliases and properties for comfort. The lazy var stream is actually the very same code that you simply’ve already within the URLSessionWebSocketTask extension from earlier than.

When our SocketStream‘s deinit is known as we be sure that we finish our stream. There’s additionally a cancel methodology that closes the socket connection in addition to the stream. As a result of SocketStream conforms to AsyncSequence we should present an Iterator object that’s used after we attempt to iterate over our SocketStreams. We merely ask our inner stream object to make an iterator and use that as our return worth.

Utilizing the code above appears as follows:

let url = URL(string: "ws://127.0.0.1:8080")!
let socketConnection = URLSession.shared.webSocketTask(with: url)
let stream = SocketStream(process: socketConnection)

Process {  
    do {
        for strive await message in stream {
            // deal with incoming messages
        }
    } catch {
        // deal with error
    }

    print("this can be printed as soon as the stream ends")
}

To cancel our stream after 5 seconds similar to earlier than, you may run the next process in parallel with our iterating process:

Process {
    strive await Process.sleep(for: .seconds(5))
    strive await stream.cancel()
}

Process {
    // iterate...
}

Whereas that is fairly cool, we do have a little bit of a difficulty right here due to the next little bit of code:

non-public lazy var stream: WebSocketStream = {
    return WebSocketStream { continuation in
        self.continuation = continuation

        Process {
            var isAlive = true

            whereas isAlive && process.closeCode == .invalid {
                do {
                    let worth = strive await process.obtain()
                    continuation.yield(worth)
                } catch {
                    continuation.end(throwing: error)
                    isAlive = false
                }
            }
        }
    }
}()

The duty that we run our whereas loop in received’t finish until we finish our stream from inside our catch block. If we manually shut the net socket connection utilizing the cancel methodology we write earlier, the decision to obtain() won’t ever obtain an error nor a worth which signifies that it will likely be caught perpetually.

Probably the most dependable technique to repair that is to return to the callback primarily based model of obtain to drive your async stream:

non-public lazy var stream: WebSocketStream = {
    return WebSocketStream { continuation in
        self.continuation = continuation
        waitForNextValue()
    }
}()

non-public func waitForNextValue() {
    guard process.closeCode == .invalid else {
        continuation?.end()
        return
    }

    process.obtain(completionHandler: { [weak self] end in
        guard let continuation = self?.continuation else {
            return
        }

        do {
            let message = strive outcome.get()
            continuation.yield(message)
            self?.waitForNextValue()
        } catch {
            continuation.end(throwing: error)
        }
    })
}

With this method we don’t have any lingering duties, and our name website is as clear and concise as ever; we’ve solely modified a few of our inner logic.

In Abstract

Swift Concurrency supplies many helpful options for writing higher code, and Apple shortly adopted async / await for current APIs. Nonetheless, some APIs that might be helpful are lacking, equivalent to iterating over net socket messages.

On this put up, you discovered how one can use async streams to create an async sequence that emits net socket messages. You first noticed a completely async / await model that was neat, however had reminiscence and process lifecycle points. Then, you noticed a model that mixes a callback-based method with the async stream.

The result’s a simple technique to iterate over incoming net socket messages with async / await. When you have any questions, feedback, or enhancements for this put up, please do not hesitate to achieve out to me on Twitter.

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments