The best way to obtain recordsdata with URLSession utilizing Mix Publishers and Subscribers?

A easy picture downloader

Downloading a useful resource from an URL looks like a trivial job, however is it actually that simple? Effectively, it relies upon. If you must obtain and parse a JSON file which is only a few KB, then you may go along with the classical manner or you need to use the brand new dataTaskPublisher technique on the URLSession object from the Mix framework.

Dangerous practices ⚠️

There are some fast & soiled approaches that you need to use to get some smaller information from the web. The issue with these strategies is that you must deal rather a lot with threads and queues. Thankfully utilizing the Dispatch framework helps rather a lot, so you may flip your blocking features into non-blocking ones. 🚧

let url = URL(string: "https://jsonplaceholder.typicode.com/todos/1")!


do {
    
    let content material = strive String(contentsOf: url)
    print(content material)

    
    let information = strive Knowledge(contentsOf: url)
}
catch {
    print(error.localizedDescription)
}



DispatchQueue.world().async { [weak self] in
    
    do {
        let content material = strive String(contentsOf: url)
        DispatchQueue.foremost.async {
            
            print(content material)
        }
    }
    catch {
        print(error.localizedDescription)
    }
}

Apple made an necessary notice on their official Knowledge documentation, that you must NOT use these strategies for downloading non-file URLs, however nonetheless individuals are instructing / utilizing these unhealthy practices, however why? 😥

Do not use this synchronous technique to request network-based URLs.

My recommendation right here: at all times use the URLSession to carry out community associated data-transfers. Creating an information job is easy, it is an asynchronous operation by default, the callback runs on a background thread, so nothing might be blocked by default. Trendy networking APIs are actual good on iOS, in 99% of the circumstances you will not want Alamofire anymore for these form of duties. Say no to dependencies! 🚫

URLSession.shared.dataTask(with: url) { information, response, error in
    
    DispatchQueue.foremost.async {
        
    }
}.resume()

It is also price to say if it’s worthwhile to use a special HTTP technique (aside from GET), ship particular headers (credentials, settle for insurance policies, and so forth.) or present further information within the physique, it’s worthwhile to assemble an URLRequest object first. You’ll be able to solely ship these customized requests utilizing the URLSession APIs.

on Apple platforms you aren’t allowed to make use of the unsecure HTTP protocol anymore. If you wish to attain a URL with out the safe layer (HTTPS) you must disable App Transport Safety.

The issue with information duties

What about massive recordsdata, equivalent to pictures? Let me present you a number of tutorials earlier than we dive in:

With all due respect, I believe all of those hyperlinks above are actually unhealthy examples of loading distant pictures. Positive they do the job, they’re additionally very simple to implement, however perhaps we should always cowl the entire story… 🤐

For small interactions with distant servers, you need to use the URLSessionDataTask class to obtain response information into reminiscence (versus utilizing the URLSessionDownloadTask class, which shops the information on to the file system). An information job is right for makes use of like calling an internet service endpoint.

What’s distinction between URLSessionDataTask vs URLSessionDownloadTask?

If we learn the docs fastidiously, it turns into clear that information job is NOT the suitable candidate for downloading massive belongings. That class is designed to request solely smaller objects, for the reason that underlying information goes to be loaded into reminiscence. However the obtain job saves the content material of the response on the disk (as a substitute of reminiscence) and you’ll obtain a neighborhood file URL as a substitute of a Knowledge object. Seems that shifting from information duties to obtain duties may have a HUGE impression in your reminiscence consumption. I’ve some numbers. 📈

I downloaded the following picture file (6000x4000px 💾 13,1MB) utilizing each strategies. I made a model new storyboard based mostly Swift 5.1 venture. The essential RAM utilization was ~52MB, after I fetched the picture utilizing the URLSessionDataTask class, the reminiscence utilization jumped to ~82MB. Turning the information job right into a obtain job solely elevated the bottom reminiscence dimension by ~4MB (to a complete ~56MB), which is a big enchancment.

let url = URL(string: "https://pictures.unsplash.com/photo-1554773228-1f38662139db")!



URLSession.shared.dataTask(with: url) { [weak self] information, response, error in
    guard let information = information else {
        return
    }
    DispatchQueue.foremost.async {
        self?.imageView.picture = UIImage(information: information)
    }
}.resume()



URLSession.shared.downloadTask(with: url) { [weak self] url, response, error in
    guard
        let cache = FileManager.default.urls(for: .cachesDirectory, in: .userDomainMask).first,
        let url = url
    else {
        return
    }

    do {
        let file = cache.appendingPathComponent("(UUID().uuidString).jpg")
        strive FileManager.default.moveItem(atPath: url.path,
                                         toPath: file.path)
        DispatchQueue.foremost.async {
            self?.imageView.picture = UIImage(contentsOfFile: file.path)
        }
    }
    catch {
        print(error.localizedDescription)
    }
}.resume()

After I rendered the picture utilizing an UIImageView the reminiscence footprint was ~118MB (whole: ~170MB) for the information job, and ~93MB (whole: ~145MB) for the obtain job. Here is a fast abstract:

• Knowledge job: ~30MB
• Knowledge job with rendering: ~118MB
• Obtain job: ~4MB
• Obtain job with rendering: ~93MB

I hope you get my level. Please remember that the Basis networking layer comes with 4 sorts of session duties. You must at all times use the suitable one that matches the job. We will say that the distinction between URLSessionDataTask vs URLSessionDownloadTask is: quite a lot of reminiscence (on this case about 25MB of RAM).

You need to use Kingfisher or SDWebImage to obtain & manipulate distant pictures..

You would possibly say that that is an edge case since many of the pictures (even HD ones) are most a number of hundred kilobytes. Nonetheless, my takeaway right here is that we will do higher, and we should always at all times accomplish that if potential. 🤓

Downloading pictures utilizing Mix

WWDC19, Apple introduced the Mix framework, which brings us a number of new extensions for some Basis objects. Trendy occasions require trendy APIs, proper? If you’re already acquainted with the brand new SDK that is good, but when you do not know what the heck is that this declarative purposeful reactive insanity, you must learn my complete tutorial in regards to the Mix framework.

The primary model of Mix shipped with a pleasant dataTaskPublisher extension technique for the URLSession class. Wait, the place are the others? No obtain job writer? What ought to we do now? 🤔

The best way to write a customized Writer?

SwiftLee has a pleasant tutorial about Mix that may assist you numerous with UIControl occasions. One other nice learn (even higher than the primary one) by Donny Wals is about understanding Publishers and Subscribers. It is a actually well-written article, you must positively test this one, I extremely suggest it. 🤘🏻

Now let’s begin creating our personal DownloadTaskPublisher. In the event you command + click on on the dataTaskPublisher technique in Xcode, you may see the corresponding interface. There may be additionally a DataTaskPublisher struct, proper under. Based mostly on that template we will create our personal extension. There are two variants of the identical information job technique, we’ll replicate this habits. The opposite factor we’d like is a DownloadTaskPublisher struct, I will present you the Swift code first, then we’ll talk about the implementation particulars.

extension URLSession {

    public func downloadTaskPublisher(for url: URL) -> URLSession.DownloadTaskPublisher {
        self.downloadTaskPublisher(for: .init(url: url))
    }

    public func downloadTaskPublisher(for request: URLRequest) -> URLSession.DownloadTaskPublisher {
        .init(request: request, session: self)
    }

    public struct DownloadTaskPublisher: Writer {

        public typealias Output = (url: URL, response: URLResponse)
        public typealias Failure = URLError

        public let request: URLRequest
        public let session: URLSession

        public init(request: URLRequest, session: URLSession) {
            self.request = request
            self.session = session
        }

        public func obtain<S>(subscriber: S) the place S: Subscriber,
            DownloadTaskPublisher.Failure == S.Failure,
            DownloadTaskPublisher.Output == S.Enter
        {
            let subscription = DownloadTaskSubscription(subscriber: subscriber, session: self.session, request: self.request)
            subscriber.obtain(subscription: subscription)
        }
    }
}

A Writer can ship an Output or a Failure message to an connected subscriber. You need to create a brand new typealias for every sort, since they each are generic constraints outlined on the protocol degree. Subsequent, we’ll retailer the session and the request objects for later use. The final a part of the protocol conformance is that you must implement the obtain<S>(subscriber: S) generic technique. This technique is answerable for attaching a brand new subscriber by way of a subscription object. Ummm… what? 🤨

A writer/subscriber relationship in Mix is solidified in a 3rd object, the subscription. When a subscriber is created and subscribes to a writer, the writer will create a subscription object and it passes a reference to the subscription to the subscriber. The subscriber will then request plenty of values from the subscription in an effort to start receiving these values.

A Writer and a Subscriber is related by way of a Subscription. The Writer solely creates the Subscription and passes it to the subscriber. The Subscription accommodates the logic that’ll fetch new information for the Subscriber. The Subscriber receives the Subscription, the values and the completion (success or failure).

• The Subscriber subscribes to a Writer
• The Writer creates a Subscription
• The Writer provides this Subscription to the Subscriber
• The Subscriber calls for some values from the Subscription
• The Subscription tries to gather the values (success or failure)
• The Subscription sends the values to the Subscriber based mostly on the demand coverage
• The Subscription sends a Failure completion to the Subscriber if an error occurs
• The Subscription sends completion if no extra values can be found

The best way to make a customized Subscription?

Okay, time to create our subscription for our little Mix based mostly downloader, I believe that you’ll perceive the connection between these three objects if we put collectively the ultimate items of the code. 🧩

extension URLSession {

    ultimate class DownloadTaskSubscription<SubscriberType: Subscriber>: Subscription the place
        SubscriberType.Enter == (url: URL, response: URLResponse),
        SubscriberType.Failure == URLError
    {
        non-public var subscriber: SubscriberType?
        non-public weak var session: URLSession!
        non-public var request: URLRequest!
        non-public var job: URLSessionDownloadTask!

        init(subscriber: SubscriberType, session: URLSession, request: URLRequest) {
            self.subscriber = subscriber
            self.session = session
            self.request = request
        }

        func request(_ demand: Subscribers.Demand) {
            guard demand > 0 else {
                return
            }
            self.job = self.session.downloadTask(with: request) { [weak self] url, response, error in
                if let error = error as? URLError {
                    self?.subscriber?.obtain(completion: .failure(error))
                    return
                }
                guard let response = response else {
                    self?.subscriber?.obtain(completion: .failure(URLError(.badServerResponse)))
                    return
                }
                guard let url = url else {
                    self?.subscriber?.obtain(completion: .failure(URLError(.badURL)))
                    return
                }
                do {
                    let cacheDir = FileManager.default.urls(for: .cachesDirectory, in: .userDomainMask).first!
                    let fileUrl = cacheDir.appendingPathComponent((UUID().uuidString))
                    strive FileManager.default.moveItem(atPath: url.path, toPath: fileUrl.path)
                    _ = self?.subscriber?.obtain((url: fileUrl, response: response))
                    self?.subscriber?.obtain(completion: .completed)
                }
                catch {
                    self?.subscriber?.obtain(completion: .failure(URLError(.cannotCreateFile)))
                }
            }
            self.job.resume()
        }

        func cancel() {
            self.job.cancel()
        }
    }
}

A Subscriber has an Enter and a Failure sort. A subscriber can solely subscribe to a writer with the identical varieties. The Writer’s Output & Failure varieties should be an identical with the Subscription Enter and Failure varieties. This time we will not go along with an associatedType, however we’ve got to create a generic worth that has a constraint on these necessities by utilizing a the place clause. The rationale behind that is that we do not know what sort of Subscriber will subscribe to this subscription. It may be both a category A or B, who is aware of… 🤷‍♂️

We now have to cross a number of properties within the init technique, retailer them as occasion variables (watch out with courses, you must use weak if relevant). Lastly we implement the worth request technique, by respecting the demand coverage. The demand is only a quantity. It tells us what number of values can we ship again to the subscriber at most. In our case we’ll have max 1 worth, so if the demand is bigger than zero, we’re good to go. You’ll be able to ship messages to the subscriber by calling varied obtain strategies on it.

You need to manually ship the completion occasion with the .completed or the .failure(T) worth. Additionally we’ve got to maneuver the downloaded momentary file earlier than the completion block returns in any other case we’ll utterly lose it. This time I will merely transfer the file to the appliance cache listing. As a free of charge cancellation is a good way to finish battery draining operations. You simply must implement a customized cancel() technique. In our case, we will name the identical technique on the underlying URLSessionDownloadTask.

That is it. We’re prepared with the customized writer & subscription. Wanna strive them out?

The best way to create a customized Subscriber?

For instance that there are 4 sorts of subscriptions. You need to use the .sink or the .assign technique to make a brand new subscription, there’s additionally a factor known as Topic, which will be subscribed for writer occasions or you may construct your very personal Subscriber object. In the event you select this path you need to use the .subscribe technique to affiliate the writer and the subscriber. You can too subscribe a topic.

ultimate class DownloadTaskSubscriber: Subscriber {
    typealias Enter = (url: URL, response: URLResponse)
    typealias Failure = URLError

    var subscription: Subscription?

    func obtain(subscription: Subscription) {
        self.subscription = subscription
        self.subscription?.request(.limitless)
    }

    func obtain(_ enter: Enter) -> Subscribers.Demand {
        print("Subscriber worth (enter.url)")
        return .limitless
    }

    func obtain(completion: Subscribers.Completion<Failure>) {
        print("Subscriber completion (completion)")
        self.subscription?.cancel()
        self.subscription = nil
    }
}

The subscriber above will merely print out the incoming values. We now have to be extraordinarily cautious with reminiscence administration. The acquired subscription might be saved as a robust property, however when the writer sends a completion occasion we should always cancel the subscription and take away the reference.

When a worth arrives we’ve got to return a requirement. In our case it actually would not matter since we’ll solely have 1 incoming worth, however if you would like to restrict your writer, you need to use e.g. .max(1) as a requirement.

Here’s a fast pattern code for all of the Mix subscriber varieties written in Swift 5.1:

class ViewController: UIViewController {

    @IBOutlet weak var imageView: UIImageView!

    static let url = URL(string: "https://pictures.unsplash.com/photo-1554773228-1f38662139db")!

    static var defaultValue: (url: URL, response: URLResponse) = {
        let fallbackUrl = URL(fileURLWithPath: "fallback-image-path")
        let fallbackResponse = URLResponse(url: fallbackUrl, mimeType: "foo", expectedContentLength: 1, textEncodingName: "bar")
        return (url: fallbackUrl, response: fallbackResponse)
    }()

    @Printed var worth: (url: URL, response: URLResponse) = ViewController.defaultValue
    let topic = PassthroughSubject<(url: URL, response: URLResponse), URLError>()
    let subscriber = DownloadTaskSubscriber()

    var sinkOperation: AnyCancellable?

    var assignOperation: AnyCancellable?
    var assignSinkOperation: AnyCancellable?

    var subjectOperation: AnyCancellable?
    var subjectSinkOperation: AnyCancellable?

    override func viewDidLoad() {
        tremendous.viewDidLoad()

        self.sinkExample()
        self.assignExample()
        self.subjectExample()
        self.subscriberExample()
    }

    func sinkExample() {
        self.sinkOperation = URLSession.shared
            .downloadTaskPublisher(for: ViewController.url)
            .sink(receiveCompletion: { completion in
                print("Sink completion: (completion)")
            }) { worth in
                print("Sink worth: (worth.url)")
            }
    }

    func assignExample() {
        self.assignSinkOperation = self.$worth.sink { worth in
            print("Assign worth: (worth.url)")
        }

        self.assignOperation = URLSession.shared
            .downloadTaskPublisher(for: ViewController.url)
            .replaceError(with: ViewController.defaultValue)
            .assign(to: .worth, on: self)
    }

    func subjectExample() {
        self.subjectSinkOperation = self.topic.sink(receiveCompletion: { completion in
            print("Topic completion: (completion)")
        }) { worth in
            print("Topic worth: (worth.url)")
        }

        self.subjectOperation = URLSession.shared
            .downloadTaskPublisher(for: ViewController.url)
            .subscribe(self.topic)
    }

    func subscriberExample() {
        URLSession.shared
            .downloadTaskPublisher(for: ViewController.url)
            .subscribe(DownloadTaskSubscriber())
    }
}

That is very nice. We will obtain a file utilizing our customized Mix based mostly URLSession extension.

Remember to retailer the AnyCancellable pointer in any other case your entire Mix operation might be deallocated manner earlier than you could possibly obtain something from the chain / stream.

Placing all the pieces collectively

I promised a working picture downloader, so let me clarify the entire move. We now have a customized obtain job writer that’ll save our take away picture file domestically and returns a tuple with the file url and the response. ✅

Subsequent I will merely assume that there was a legitimate picture behind the url, and the server returned a legitimate response, so I will map the writer’s output to an UIImage object. I am additionally going to exchange any form of error with a fallback picture worth. In a real-world utility, you must at all times do some further checkings on the URLResponse object, however for the sake of simplicity I will skip that for now.

The very last thing is to replace our picture view with the returned picture. Since this can be a UI job it ought to occur on the primary thread, so we’ve got to make use of the obtain(on:) operation to change context. If you wish to be taught extra about schedulers within the Mix framework you must learn Vadim Bulavin‘s article. It is a gem. 💎

If you’re not receiving values on sure appleOS variations, that is would possibly as a result of there was a change in Mix round December, 2019. You must test these hyperlinks: link1, link2

Anyway, here is the ultimate Swift code for a potential picture obtain operation, easy & declarative. 👍

class ViewController: UIViewController {

    @IBOutlet weak var imageView: UIImageView!

    var operation: AnyCancellable?

    override func viewDidLoad() {
        tremendous.viewDidLoad()

        let url = URL(string: "https://pictures.unsplash.com/photo-1554773228-1f38662139db")!

        self.operation = URLSession.shared
            .downloadTaskPublisher(for: url)
            .map { UIImage(contentsOfFile: $0.url.path)! }
            .replaceError(with: UIImage(named: "fallback"))
            .obtain(on: DispatchQueue.foremost)
            .assign(to: .picture, on: self.imageView)
    }
}

Lastly, we will show our picture. Ouch, however wait… there’s nonetheless room for enhancements. What about caching? Plus a 6000x4000px image is sort of large for a small show, should not we resize / scale the picture first? What occurs if I wish to use the picture in an inventory, should not I cancel the obtain duties when the consumer scrolls? 😳

Possibly I will write about these points in an upcoming tutorial, however I believe that is the purpose the place I ought to finish this text. Be happy to mess around with my answer and please share your concepts & ideas with me on Twitter.