File add API server in Vapor 4

A easy file add server written in Swift

For this easy file add tutorial we’ll solely use the Vapor Swift bundle as a dependency. 📦

// swift-tools-version:5.3
import PackageDescription

let bundle = Bundle(
    identify: "myProject",
    platforms: [
       .macOS(.v10_15)
    ],
    dependencies: [
        .package(url: "https://github.com/vapor/vapor", from: "4.35.0"),
    ],
    targets: [
        .target(
            name: "App",
            dependencies: [
                .product(name: "Vapor", package: "vapor"),
            ],
            swiftSettings: [
                .unsafeFlags(["-cross-module-optimization"], .when(configuration: .launch))
            ]
        ),
        .goal(identify: "Run", dependencies: [.target(name: "App")]),
        .testTarget(identify: "AppTests", dependencies: [
            .target(name: "App"),
            .product(name: "XCTVapor", package: "vapor"),
        ])
    ]
)

You possibly can setup the mission with the required information utilizing the Vapor toolbox, alternatively you may create the whole lot by hand utilizing the Swift Bundle Supervisor, lengthy story quick, we simply want a starter Vapor mission with out further dependencies. Now in the event you open the Bundle.swift file utilizing Xcode, we will setup our routes by altering the configure.swift file.

import Vapor

public func configure(_ app: Utility) throws {

    /// allow file middleware
    app.middleware.use(FileMiddleware(publicDirectory: app.listing.publicDirectory))

    /// set max physique measurement
    app.routes.defaultMaxBodySize = "10mb"

    /// setup the add handler
    app.publish("add") { req -> EventLoopFuture in
        let key = attempt req.question.get(String.self, at: "key")
        let path = req.utility.listing.publicDirectory + key
        return req.physique.acquire()
            .unwrap(or: Abort(.noContent))
            .flatMap { req.fileio.writeFile($0, at: path) }
            .map { key }
    }
}

First we use the FileMiddleware, this can enable us to server information utilizing the Public listing inside our mission folder. In case you don’t have a listing named Public, please create one, because the file add server will want that. Don’t neglect to offer correct file system permissions if needed, in any other case we received’t be capable to write our knowledge contained in the listing. 📁

The following factor that we set is the default most physique measurement. This property can restrict the quantity of knowledge that our server can settle for, you don’t actually wish to use this methodology for giant information as a result of uploaded information will probably be saved within the system reminiscence earlier than we write them to the disk.

If you wish to add giant information to the server it is best to think about streaming the file as an alternative of gathering the file knowledge from the HTTP physique. The streaming setup would require a bit extra work, nevertheless it’s not that sophisticated, if you’re all in favour of that answer, it is best to learn the Information API and the physique streaming part utilizing official Vapor docs website.

This time we simply desire a useless easy file add API endpoint, that collects the incoming knowledge utilizing the HTTP physique right into a byte buffer object, then we merely write this buffer utilizing the fileio to the disk, utilizing the given key from the URL question parameters. If the whole lot was executed with out errors, we will return the important thing for the uploaded file.

File add duties utilizing the URLSession API
The Basis frameworks provides us a pleasant API layer for frequent networking duties. We will use the URLSession uploadTask methodology to ship a brand new URLRequest with a knowledge object to a given server, however IMHO this API is sort of unusual, as a result of the URLRequest object already has a httpBody property, however you need to explicitly go a “from: Knowledge?” argument once you assemble the duty. However why? 🤔

import Basis

extension URLSession {

    func uploadTask(with request: URLRequest, completionHandler: @escaping (Knowledge?, URLResponse?, Error?) -> Void) -> URLSessionUploadTask {
        uploadTask(with: request, from: request.httpBody, completionHandler: completionHandler)
    }
}

Anyway, I made just a little extension methodology, so after I create the URLRequest I can set the httpBody property of it and safely go it earlier than the completion block and use the contents because the from parameter. Very unusual API design alternative from Apple… 🤐

We will put this little snippet right into a easy executable Swift bundle (or after all we will create a whole utility) to check our add server. In our case I’ll place the whole lot right into a foremost.swift file.

import Basis
import Dispatch

extension URLSession {

    func uploadTask(with request: URLRequest, completionHandler: @escaping (Knowledge?, URLResponse?, Error?) -> Void) -> URLSessionUploadTask {
        uploadTask(with: request, from: request.httpBody, completionHandler: completionHandler)
    }
}


let fileData = attempt Knowledge(contentsOf: URL(fileURLWithPath: "/Customers/[user]]/[file].png"))
var request = URLRequest(url: URL(string: "http://localhost:8080/add?key=(UUID().uuidString).png")!)
request.httpMethod = "POST"
request.httpBody = fileData

let process = URLSession.shared.uploadTask(with: request) { knowledge, response, error in
    guard error == nil else {
        fatalError(error!.localizedDescription)
    }
    guard let response = response as? HTTPURLResponse else {
        fatalError("Invalid response")
    }
    guard response.statusCode == 200 else {
        fatalError("HTTP standing error: (response.statusCode)")
    }
    guard let knowledge = knowledge, let outcome = String(knowledge: knowledge, encoding: .utf8) else {
        fatalError("Invalid or lacking HTTP knowledge")
    }
    print(outcome)
    exit(0)
}

process.resume()
dispatchMain()

The above instance makes use of the Dispatch framework to attend till the asynchronous file add finishes. It is best to change the situation (and the extension) of the file if needed earlier than you run this script. Since we outlined the add route as a POST endpoint, we have now to set the httpMethod property to match this, additionally we retailer the file knowledge within the httpBody variable earlier than we create our process. The add URL ought to comprise a key, that the server can use as a reputation for the file. You possibly can add extra properties after all or use header values to test if the person has correct authorization to carry out the add operation. Then we name the add process extension methodology on the shared URLSession property. The good factor about uploadTask is you can run them on the background if wanted, that is fairly helpful if it involves iOS improvement. 📱

Contained in the completion handler we have now to test for a couple of issues. To start with if there was an error, the add will need to have failed, so we name the fatalError methodology to interrupt execution. If the response was not a legitimate HTTP response, or the standing code was not okay (200) we additionally cease. Lastly we wish to retrieve the important thing from the response physique so we test the info object and convert it to a UTF8 string if doable. Now we will use the important thing mixed with the area of the server to entry the uploaded file, this time I simply printed out the outcome, however hey, that is only a demo, in an actual world utility you may wish to return a JSON response with further knowledge. 😅

Vanilla JavaScript file uploader

Yet one more factor… you should utilize Leaf and a few Vanilla JavaScript to add information utilizing the newly created add endpoint. Truly it’s very easy to implement a brand new endpoint and render a Leaf template that does the magic. You’ll want some primary HTML and some traces of JS code to submit the contents of the file as an array buffer. It is a primary instance.

As you may see it’s a normal XHR request mixed with the FileReader JavaScript API. We use the FileReader to transform our enter to a binary knowledge, this fashion our server can write it to the file system within the anticipated format. Most often individuals are utilizing a multipart-encoded type to entry information on the server, however when you need to work with an API you can even switch uncooked file knowledge. If you wish to be taught extra about XHR requests and AJAX calls, it is best to learn my earlier article.

I even have a publish about completely different file add strategies utilizing normal HTML kinds and a Vapor 4 server as a backend. I hope you’ll discover the fitting answer that you just want to your utility. 👍