AdvancedUsage

• Advanced Usage
  • Session
    • Creating a Session with Default Configuration
    • Creating a Session with Background Configuration
    • Creating a Session with Ephemeral Configuration
    • Modifying the Session Configuration
  • Session Delegate
    • Override Closures
    • Subclassing
  • Request
  • Routing Requests
    • URLConvertible
      • Type-Safe Routing
    • URLRequestConvertible
      • API Parameter Abstraction
      • CRUD and Authorization
  • Adapting and Retrying Requests
    • RequestAdapter
    • RequestRetrier
  • Custom Response Serialization
    • Response Mapping
    • Handling Errors
    • Creating a Custom Response Serializer
    • Generic Response Object Serialization
  • Security
    • ServerTrustPolicy
    • Server Trust Policy Manager
      • Subclassing Server Trust Policy Manager
    • Validating the Host
    • Validating the Certificate Chain
    • App Transport Security
    • Using Self-Signed Certificates with Local Networking
  • Network Reachability

Advanced Usage

Alamofire is built on URLSession and the Foundation URL Loading System. To make the most of this framework, it is recommended that you be familiar with the concepts and capabilities of the underlying networking stack.

Recommended Reading

• URL Loading System Programming Guide
• URLSession Class Reference
• URLCache Class Reference
• URLAuthenticationChallenge Class Reference

Session

Top-level convenience methods like Alamofire.request use a default instance of Alamofire.Session, which is configured with the default URLSessionConfiguration.

As such, the following two statements are equivalent:

Alamofire.request("https://httpbin.org/get")

let session = Alamofire.Session.default
session.request("https://httpbin.org/get")

Applications can create sessions for background and ephemeral sessions, as well as new sessions that customize the default session configuration, such as for default headers (httpAdditionalHeaders) or timeout interval (timeoutIntervalForRequest).

Creating a Session with Default Configuration

let configuration = URLSessionConfiguration.default
let session = Alamofire.Session(configuration: configuration)

Creating a Session with Background Configuration

let configuration = URLSessionConfiguration.background(withIdentifier: "com.example.app.background")
let session = Alamofire.Session(configuration: configuration)

Creating a Session with Ephemeral Configuration

let configuration = URLSessionConfiguration.ephemeral
let session = Alamofire.Session(configuration: configuration)

Modifying the Session Configuration

var defaultHeaders = Alamofire.Session.defaultHTTPHeaders
defaultHeaders["DNT"] = "1 (Do Not Track Enabled)"

let configuration = URLSessionConfiguration.default
configuration.httpAdditionalHeaders = defaultHeaders

let session = Alamofire.Session(configuration: configuration)

This is not recommended for Authorization or Content-Type headers. Instead, use the headers parameter in the top-level Alamofire.request APIs, URLRequestConvertible and ParameterEncoding, respectively.

Session Delegate

By default, an Alamofire Session instance creates a SessionDelegate object to handle all the various types of delegate callbacks that are generated by the underlying URLSession. The implementations of each delegate method handle the most common use cases for these types of calls abstracting the complexity away from the top-level APIs. However, advanced users may find the need to override the default functionality for various reasons.

Override Closures

The first way to customize the SessionDelegate behavior is through the use of the override closures. Each closure gives you the ability to override the implementation of the matching SessionDelegate API, yet still use the default implementation for all other APIs. This makes it easy to customize subsets of the delegate functionality. Here are a few examples of some of the override closures available:

/// Overrides default behavior for URLSessionDelegate method `urlSession(_:didReceive:completionHandler:)`.
open var sessionDidReceiveChallenge: ((URLSession, URLAuthenticationChallenge) -> (URLSession.AuthChallengeDisposition, URLCredential?))?

/// Overrides default behavior for URLSessionDelegate method `urlSessionDidFinishEvents(forBackgroundURLSession:)`.
open var sessionDidFinishEventsForBackgroundURLSession: ((URLSession) -> Void)?

/// Overrides default behavior for URLSessionTaskDelegate method `urlSession(_:task:willPerformHTTPRedirection:newRequest:completionHandler:)`.
open var taskWillPerformHTTPRedirection: ((URLSession, URLSessionTask, HTTPURLResponse, URLRequest) -> URLRequest?)?

/// Overrides default behavior for URLSessionDataDelegate method `urlSession(_:dataTask:willCacheResponse:completionHandler:)`.
open var dataTaskWillCacheResponse: ((URLSession, URLSessionDataTask, CachedURLResponse) -> CachedURLResponse?)?

The following is a short example of how to use the taskWillPerformHTTPRedirection to avoid following redirects to any apple.com domains.

let session = Alamofire.Session(configuration: URLSessionConfiguration.default)
let delegate: Alamofire.SessionDelegate = session.delegate

delegate.taskWillPerformHTTPRedirection = { session, task, response, request in
    var finalRequest = request

    if
        let originalRequest = task.originalRequest,
        let urlString = originalRequest.url?.urlString,
        urlString.contains("apple.com")
    {
        finalRequest = originalRequest
    }

    return finalRequest
}

Subclassing

Another way to override the default implementation of the SessionDelegate is to subclass it. Subclassing allows you completely customize the behavior of the API or to create a proxy for the API and still use the default implementation. Creating a proxy allows you to log events, emit notifications, provide pre and post hook implementations, etc. Here's a quick example of subclassing the SessionDelegate and logging a message when a redirect occurs.

class LoggingSessionDelegate: SessionDelegate {
    override func urlSession(
        _ session: URLSession,
        task: URLSessionTask,
        willPerformHTTPRedirection response: HTTPURLResponse,
        newRequest request: URLRequest,
        completionHandler: @escaping (URLRequest?) -> Void)
    {
        print("URLSession will perform HTTP redirection to request: \(request)")

        super.urlSession(
            session,
            task: task,
            willPerformHTTPRedirection: response,
            newRequest: request,
            completionHandler: completionHandler
        )
    }
}

Generally speaking, either the default implementation or the override closures should provide the necessary functionality required. Subclassing should only be used as a last resort.

It is important to keep in mind that the subdelegates are initialized and destroyed in the default implementation. Be careful when subclassing to not introduce memory leaks.

Request

The result of a request, download, upload or stream methods are a DataRequest, DownloadRequest, UploadRequest and StreamRequest which all inherit from Request. All Request instances are always created by an owning session, and never initialized directly.

Each subclass has specialized methods such as authenticate, validate, responseJSON and uploadProgress that each return the caller instance in order to facilitate method chaining.

Requests can be suspended, resumed and cancelled:

• suspend(): Suspends the underlying task and dispatch queue.
• resume(): Resumes the underlying task and dispatch queue. If the owning session does not have startRequestsImmediately set to true, the request must call resume() in order to start.
• cancel(): Cancels the underlying task, producing an error that is passed to any registered response handlers.

Routing Requests

As apps grow in size, it's important to adopt common patterns as you build out your network stack. An important part of that design is how to route your requests. The Alamofire URLConvertible and URLRequestConvertible protocols along with the Router design pattern are here to help.

URLConvertible

Types adopting the URLConvertible protocol can be used to construct URLs, which are then used to construct URL requests internally. String, URL, and URLComponents conform to URLConvertible by default, allowing any of them to be passed as url parameters to the request, upload, and download methods:

let urlString = "https://httpbin.org/post"
Alamofire.request(urlString, method: .post)

let url = URL(string: urlString)!
Alamofire.request(url, method: .post)

let urlComponents = URLComponents(url: url, resolvingAgainstBaseURL: true)!
Alamofire.request(urlComponents, method: .post)

Applications interacting with web applications in a significant manner are encouraged to have custom types conform to URLConvertible as a convenient way to map domain-specific models to server resources.

Type-Safe Routing

extension User: URLConvertible {
    static let baseURLString = "https://example.com"

    func asURL() throws -> URL {
    	let urlString = User.baseURLString + "/users/\(username)/"
        return try urlString.asURL()
    }
}

let user = User(username: "mattt")
Alamofire.request(user) // https://example.com/users/mattt

URLRequestConvertible

Types adopting the URLRequestConvertible protocol can be used to construct URL requests. URLRequest conforms to URLRequestConvertible by default, allowing it to be passed into request, upload, and download methods directly (this is the recommended way to specify custom HTTP body for individual requests):

let url = URL(string: "https://httpbin.org/post")!
var urlRequest = URLRequest(url: url)
urlRequest.httpMethod = "POST"

let parameters = ["foo": "bar"]

do {
    urlRequest.httpBody = try JSONSerialization.data(withJSONObject: parameters, options: [])
} catch {
    // No-op
}

urlRequest.setValue("application/json", forHTTPHeaderField: "Content-Type")

Alamofire.request(urlRequest)

Applications interacting with web applications in a significant manner are encouraged to have custom types conform to URLRequestConvertible as a way to ensure consistency of requested endpoints. Such an approach can be used to abstract away server-side inconsistencies and provide type-safe routing, as well as manage authentication credentials and other state.

API Parameter Abstraction

enum Router: URLRequestConvertible {
    case search(query: String, page: Int)

    static let baseURLString = "https://example.com"
    static let perPage = 50

    // MARK: URLRequestConvertible

    func asURLRequest() throws -> URLRequest {
        let result: (path: String, parameters: Parameters) = {
            switch self {
            case let .search(query, page) where page > 0:
                return ("/search", ["q": query, "offset": Router.perPage * page])
            case let .search(query, _):
                return ("/search", ["q": query])
            }
        }()

        let url = try Router.baseURLString.asURL()
        let urlRequest = URLRequest(url: url.appendingPathComponent(result.path))

        return try URLEncoding.default.encode(urlRequest, with: result.parameters)
    }
}

Alamofire.request(Router.search(query: "foo bar", page: 1)) // https://example.com/search?q=foo%20bar&offset=50

CRUD and Authorization

import Alamofire

enum Router: URLRequestConvertible {
    case createUser(parameters: Parameters)
    case readUser(username: String)
    case updateUser(username: String, parameters: Parameters)
    case destroyUser(username: String)

    static let baseURLString = "https://example.com"

    var method: HTTPMethod {
        switch self {
        case .createUser:
            return .post
        case .readUser:
            return .get
        case .updateUser:
            return .put
        case .destroyUser:
            return .delete
        }
    }

    var path: String {
        switch self {
        case .createUser:
            return "/users"
        case .readUser(let username):
            return "/users/\(username)"
        case .updateUser(let username, _):
            return "/users/\(username)"
        case .destroyUser(let username):
            return "/users/\(username)"
        }
    }

    // MARK: URLRequestConvertible

    func asURLRequest() throws -> URLRequest {
    	let url = try Router.baseURLString.asURL()

        var urlRequest = URLRequest(url: url.appendingPathComponent(path))
        urlRequest.httpMethod = method.rawValue

        switch self {
        case .createUser(let parameters):
            urlRequest = try URLEncoding.default.encode(urlRequest, with: parameters)
        case .updateUser(_, let parameters):
            urlRequest = try URLEncoding.default.encode(urlRequest, with: parameters)
        default:
            break
        }

        return urlRequest
    }
}

Alamofire.request(Router.readUser("mattt")) // GET https://example.com/users/mattt

Adapting and Retrying Requests

Most web services these days are behind some sort of authentication system. One of the more common ones today is OAuth. This generally involves generating an access token authorizing your application or user to call the various supported web services. While creating these initial access tokens can be laborsome, it can be even more complicated when your access token expires and you need to fetch a new one. There are many thread-safety issues that need to be considered.

The RequestAdapter and RequestRetrier protocols were created to make it much easier to create a thread-safe authentication system for a specific set of web services. Note that to use either a RequestAdapter instance or RequestRetrier instance with the Session, the instance must conform to RequestInterceptor, which is a composite of RequestAdapter and RequestRetrier. Alamofire provides default implementations in protocol extensions so you only need to change the adapt or retry function as needed.

RequestAdapter

The RequestAdapter protocol allows each Request made on a Session to be inspected and adapted before being created. One very specific way to use an adapter is to append an Authorization header to requests behind a certain type of authentication.

class AccessTokenAdapter: RequestInterceptor {
    private let accessToken: String

    init(accessToken: String) {
        self.accessToken = accessToken
    }

    // MARK: - RequestAdapter
    func adapt(_ urlRequest: URLRequest, for session: Session, completion: @escaping (Result<URLRequest, Error>) -> Void) {
        var urlRequest = urlRequest

        if let urlString = urlRequest.url?.absoluteString, urlString.hasPrefix("https://httpbin.org") {
            urlRequest.setValue("Bearer " + accessToken, forHTTPHeaderField: "Authorization")
        }

       completion(.success(urlRequest))
    }
}

let accessTokenManager = AccessTokenAdapter(accessToken: "1234")
let session = Session(interceptor: accessTokenManager)

session.request("https://httpbin.org/get")

RequestRetrier

The RequestRetrier protocol allows a Request that encountered an Error while being executed to be retried. When using both the RequestAdapter and RequestRetrier protocols together, you can create credential refresh systems for OAuth1, OAuth2, Basic Auth and even exponential backoff retry policies. The possibilities are endless. Here's an example of how you could implement a refresh flow for OAuth2 access tokens.

DISCLAIMER: This is NOT a global OAuth2 solution. It is merely an example demonstrating how one could use the RequestAdapter in conjunction with the RequestRetrier to create a thread-safe refresh system.

To reiterate, do NOT copy this sample code and drop it into a production application. This is merely an example. Each authentication system must be tailored to a particular platform and authentication type.

class OAuth2Handler: RequestInterceptor {
    private typealias RefreshCompletion = (_ succeeded: Bool, _ accessToken: String?, _ refreshToken: String?) -> Void

    private let session: Session = {
        let configuration = URLSessionConfiguration.default
        return Session(configuration: configuration)
    }()

    private let lock = NSLock()

    private var clientID: String
    private var baseURLString: String
    private var accessToken: String
    private var refreshToken: String

    private var isRefreshing = false
    private var requestsToRetry: [RequestRetryCompletion] = []

    // MARK: - Initialization

    public init(clientID: String, baseURLString: String, accessToken: String, refreshToken: String) {
        self.clientID = clientID
        self.baseURLString = baseURLString
        self.accessToken = accessToken
        self.refreshToken = refreshToken
    }

    // MARK: - RequestAdapter

    func adapt(_ urlRequest: URLRequest, for session: Session, completion: @escaping (Result<URLRequest, Error>) -> Void) {
        if let urlString = urlRequest.url?.absoluteString, urlString.hasPrefix(baseURLString) {
            var urlRequest = urlRequest
            urlRequest.setValue("Bearer " + accessToken, forHTTPHeaderField: "Authorization")
            return urlRequest
        }

        completion(.success(urlRequest))
    }

    // MARK: - RequestRetrier

    func retry(_ request: Request, for session: Session, dueTo error: Error, completion: @escaping (RetryResult) -> Void) {
        lock.lock() ; defer { lock.unlock() }

        if let response = request.task?.response as? HTTPURLResponse, response.statusCode == 401 {
            requestsToRetry.append(completion)

            if !isRefreshing {
                refreshTokens { [weak self] succeeded, accessToken, refreshToken in
                    guard let strongSelf = self else { return }

                    strongSelf.lock.lock() ; defer { strongSelf.lock.unlock() }

                    if let accessToken = accessToken, let refreshToken = refreshToken {
                        strongSelf.accessToken = accessToken
                        strongSelf.refreshToken = refreshToken
                    }

                    strongSelf.requestsToRetry.forEach { $0(.retry) }
                    strongSelf.requestsToRetry.removeAll()
                }
            }
        } else {
            completion(.doNotRetry)
        }
    }

    // MARK: - Private - Refresh Tokens

    private func refreshTokens(completion: @escaping RefreshCompletion) {
        guard !isRefreshing else { return }

        isRefreshing = true

        let urlString = "\(baseURLString)/oauth2/token"

        let parameters: [String: Any] = [
            "access_token": accessToken,
            "refresh_token": refreshToken,
            "client_id": clientID,
            "grant_type": "refresh_token"
        ]

        session.request(urlString, method: .post, parameters: parameters, encoding: JSONEncoding.default)
            .responseJSON { [weak self] response in
                guard let strongSelf = self else { return }

                if 
                    let json = response.result.value as? [String: Any], 
                    let accessToken = json["access_token"] as? String, 
                    let refreshToken = json["refresh_token"] as? String 
                {
                    completion(true, accessToken, refreshToken)
                } else {
                    completion(false, nil, nil)
                }

                strongSelf.isRefreshing = false
            }
    }
}

let baseURLString = "https://some.domain-behind-oauth2.com"

let oauthHandler = OAuth2Handler(
    clientID: "12345678",
    baseURLString: baseURLString,
    accessToken: "abcd1234",
    refreshToken: "ef56789a"
)

let session = Session(interceptor: oauth2Handler)

let urlString = "\(baseURLString)/some/endpoint"

session.request(urlString).validate().responseJSON { response in
    debugPrint(response)
}

Once the OAuth2Handler is applied as both the adapter and retrier for the Session, it will handle an invalid access token error by automatically refreshing the access token and retrying all failed requests in the same order they failed.

If you needed them to execute in the same order they were created, you could sort them by their task identifiers.

The example above only checks for a 401 response code which is not nearly robust enough, but does demonstrate how one could check for an invalid access token error. In a production application, one would want to check the realm and most likely the www-authenticate header response although it depends on the OAuth2 implementation.

Another important note is that this authentication system could be shared between multiple sessions. For example, you may need to use both a default and ephemeral session configuration for the same set of web services. The example above allows the same oauthHandler instance to be shared across multiple sessions to manage the single refresh flow.

Custom Response Serialization

Alamofire provides built-in response serialization for data, strings, JSON, and property lists:

Alamofire.request(...).responseData { (resp: DataResponse<Data>) in ... }
Alamofire.request(...).responseString { (resp: DataResponse<String>) in ... }
Alamofire.request(...).responseJSON { (resp: DataResponse<Any>) in ... }
Alamofire.request(...).responsePropertyList { (resp: DataResponse<Any>) in ... }

Those responses wrap deserialized values (Data, String, Any) or errors (network, validation errors), as well as meta-data (URL request, HTTP headers, status code, metrics, ...).

You have several ways to customize all of those response elements:

• Response Mapping
• Handling Errors
• Creating a Custom Response Serializer
• Generic Response Object Serialization

Response Mapping

Response mapping is the simplest way to produce customized responses. It transforms the value of a response, while preserving eventual errors and meta-data. For example, you can turn a json response DataResponse<Any> into a response that holds an application model, such as DataResponse<User>. You perform response mapping with the DataResponse.map method:

Alamofire.request("https://example.com/users/mattt").responseJSON { (response: DataResponse<Any>) in
    let userResponse = response.map { json in
        // We assume an existing User(json: Any) initializer
        return User(json: json)
    }

    // Process userResponse, of type DataResponse<User>:
    if let user = userResponse.value {
        print("User: { username: \(user.username), name: \(user.name) }")
    }
}

When the transformation may throw an error, use flatMap instead:

Alamofire.request("https://example.com/users/mattt").responseJSON { response in
    let userResponse = response.flatMap { json in
        try User(json: json)
    }
}

Response mapping is a good fit for your custom completion handlers:

@discardableResult
func loadUser(completionHandler: @escaping (DataResponse<User>) -> Void) -> Alamofire.DataRequest {
    return Alamofire.request("https://example.com/users/mattt").responseJSON { response in
        let userResponse = response.flatMap { json in
            try User(json: json)
        }

        completionHandler(userResponse)
    }
}

loadUser { response in
    if let user = response.value {
        print("User: { username: \(user.username), name: \(user.name) }")
    }
}

When the map/flatMap closure may process a big amount of data, make sure you execute it outside of the main thread:

@discardableResult
func loadUser(completionHandler: @escaping (DataResponse<User>) -> Void) -> Alamofire.DataRequest {
    let utilityQueue = DispatchQueue.global(qos: .utility)

    return Alamofire.request("https://example.com/users/mattt").responseJSON(queue: utilityQueue) { response in
        let userResponse = response.flatMap { json in
            try User(json: json)
        }

        DispatchQueue.main.async {
            completionHandler(userResponse)
        }
    }
}

map and flatMap are also available for download responses.

Handling Errors

Before implementing custom response serializers or object serialization methods, it's important to consider how to handle any errors that may occur. There are two basic options: passing existing errors along unmodified, to be dealt with at response time; or, wrapping all errors in an Error type specific to your app.

For example, here's a simple BackendError enum which will be used in later examples:

enum BackendError: Error {
    case network(error: Error) // Capture any underlying Error from the URLSession API
    case dataSerialization(error: Error)
    case jsonSerialization(error: Error)
    case xmlSerialization(error: Error)
    case objectSerialization(reason: String)
}

Creating a Custom Response Serializer

Alamofire provides built-in response serialization for strings, JSON, and property lists, but others can be added in extensions on Alamofire.DataRequest and / or Alamofire.DownloadRequest.

For example, here's how a response handler using Ono might be implemented:

extension DataRequest {
    static func xmlResponseSerializer() -> DataResponseSerializer<ONOXMLDocument> {
        return DataResponseSerializer { request, response, data, error in
            // Pass through any underlying URLSession error to the .network case.
            guard error == nil else { return .failure(BackendError.network(error: error!)) }

            // Use Alamofire's existing data serializer to extract the data, passing the error as nil, as it has
            // already been handled.
            let result = Request.serializeResponseData(response: response, data: data, error: nil)

            guard case let .success(validData) = result else {
                return .failure(BackendError.dataSerialization(error: result.error! as! AFError))
            }

            do {
                let xml = try ONOXMLDocument(data: validData)
                return .success(xml)
            } catch {
                return .failure(BackendError.xmlSerialization(error: error))
            }
        }
    }

    @discardableResult
    func responseXMLDocument(
        queue: DispatchQueue? = nil,
        completionHandler: @escaping (DataResponse<ONOXMLDocument>) -> Void)
        -> Self
    {
        return response(
            queue: queue,
            responseSerializer: DataRequest.xmlResponseSerializer(),
            completionHandler: completionHandler
        )
    }
}

Generic Response Object Serialization

Generics can be used to provide automatic, type-safe response object serialization.

protocol ResponseObjectSerializable {
    init?(response: HTTPURLResponse, representation: Any)
}

extension DataRequest {
    func responseObject<T: ResponseObjectSerializable>(
        queue: DispatchQueue? = nil,
        completionHandler: @escaping (DataResponse<T>) -> Void)
        -> Self
    {
        let responseSerializer = DataResponseSerializer<T> { request, response, data, error in
            guard error == nil else { return .failure(BackendError.network(error: error!)) }

            let jsonResponseSerializer = DataRequest.jsonResponseSerializer(options: .allowFragments)
            let result = jsonResponseSerializer.serializeResponse(request, response, data, nil)

            guard case let .success(jsonObject) = result else {
                return .failure(BackendError.jsonSerialization(error: result.error!))
            }

            guard let response = response, let responseObject = T(response: response, representation: jsonObject) else {
                return .failure(BackendError.objectSerialization(reason: "JSON could not be serialized: \(jsonObject)"))
            }

            return .success(responseObject)
        }

        return response(queue: queue, responseSerializer: responseSerializer, completionHandler: completionHandler)
    }
}

struct User: ResponseObjectSerializable, CustomStringConvertible {
    let username: String
    let name: String

    var description: String {
        return "User: { username: \(username), name: \(name) }"
    }

    init?(response: HTTPURLResponse, representation: Any) {
        guard
            let username = response.url?.lastPathComponent,
            let representation = representation as? [String: Any],
            let name = representation["name"] as? String
        else { return nil }

        self.username = username
        self.name = name
    }
}

Alamofire.request("https://example.com/users/mattt").responseObject { (response: DataResponse<User>) in
    debugPrint(response)

    if let user = response.result.value {
        print("User: { username: \(user.username), name: \(user.name) }")
    }
}

The same approach can also be used to handle endpoints that return a representation of a collection of objects:

protocol ResponseCollectionSerializable {
    static func collection(from response: HTTPURLResponse, withRepresentation representation: Any) -> [Self]
}

extension ResponseCollectionSerializable where Self: ResponseObjectSerializable {
    static func collection(from response: HTTPURLResponse, withRepresentation representation: Any) -> [Self] {
        var collection: [Self] = []

        if let representation = representation as? [[String: Any]] {
            for itemRepresentation in representation {
                if let item = Self(response: response, representation: itemRepresentation) {
                    collection.append(item)
                }
            }
        }

        return collection
    }
}

extension DataRequest {
    @discardableResult
    func responseCollection<T: ResponseCollectionSerializable>(
        queue: DispatchQueue? = nil,
        completionHandler: @escaping (DataResponse<[T]>) -> Void) -> Self
    {
        let responseSerializer = DataResponseSerializer<[T]> { request, response, data, error in
            guard error == nil else { return .failure(BackendError.network(error: error!)) }

            let jsonSerializer = DataRequest.jsonResponseSerializer(options: .allowFragments)
            let result = jsonSerializer.serializeResponse(request, response, data, nil)

            guard case let .success(jsonObject) = result else {
                return .failure(BackendError.jsonSerialization(error: result.error!))
            }

            guard let response = response else {
                let reason = "Response collection could not be serialized due to nil response."
                return .failure(BackendError.objectSerialization(reason: reason))
            }

            return .success(T.collection(from: response, withRepresentation: jsonObject))
        }

        return response(responseSerializer: responseSerializer, completionHandler: completionHandler)
    }
}

struct User: ResponseObjectSerializable, ResponseCollectionSerializable, CustomStringConvertible {
    let username: String
    let name: String

    var description: String {
        return "User: { username: \(username), name: \(name) }"
    }

    init?(response: HTTPURLResponse, representation: Any) {
        guard
            let username = response.url?.lastPathComponent,
            let representation = representation as? [String: Any],
            let name = representation["name"] as? String
        else { return nil }

        self.username = username
        self.name = name
    }
}

Alamofire.request("https://example.com/users").responseCollection { (response: DataResponse<[User]>) in
    debugPrint(response)

    if let users = response.result.value {
        users.forEach { print("- \($0)") }
    }
}

Security

Using a secure HTTPS connection when communicating with servers and web services is an important step in securing sensitive data. By default, Alamofire will evaluate the certificate chain provided by the server using Apple's built in validation provided by the Security framework. While this guarantees the certificate chain is valid, it does not prevent man-in-the-middle (MITM) attacks or other potential vulnerabilities. In order to mitigate MITM attacks, applications dealing with sensitive customer data or financial information should use certificate or public key pinning provided by the ServerTrustPolicy.

ServerTrustPolicy

The ServerTrustPolicy enumeration evaluates the server trust generally provided by an URLAuthenticationChallenge when connecting to a server over a secure HTTPS connection.

let serverTrustPolicy = ServerTrustPolicy.pinCertificates(
    certificates: ServerTrustPolicy.certificates(),
    validateCertificateChain: true,
    validateHost: true
)

There are many different cases of server trust evaluation giving you complete control over the validation process:

• performDefaultEvaluation: Uses the default server trust evaluation while allowing you to control whether to validate the host provided by the challenge.
• pinCertificates: Uses the pinned certificates to validate the server trust. The server trust is considered valid if one of the pinned certificates match one of the server certificates.
• pinPublicKeys: Uses the pinned public keys to validate the server trust. The server trust is considered valid if one of the pinned public keys match one of the server certificate public keys.
• disableEvaluation: Disables all evaluation which in turn will always consider any server trust as valid.
• customEvaluation: Uses the associated closure to evaluate the validity of the server trust thus giving you complete control over the validation process. Use with caution.

Server Trust Policy Manager

The ServerTrustPolicyManager is responsible for storing an internal mapping of server trust policies to a particular host. This allows Alamofire to evaluate each host against a different server trust policy.

let serverTrustPolicies: [String: ServerTrustPolicy] = [
    "test.example.com": .pinCertificates(
        certificates: ServerTrustPolicy.certificates(),
        validateCertificateChain: true,
        validateHost: true
    ),
    "insecure.expired-apis.com": .disableEvaluation
]

let session = Session(
    serverTrustPolicyManager: ServerTrustPolicyManager(policies: serverTrustPolicies)
)

Make sure to keep a reference to the new Session instance, otherwise your requests will all get cancelled when your session is deallocated.

These server trust policies will result in the following behavior:

• test.example.com will always use certificate pinning with certificate chain and host validation enabled thus requiring the following criteria to be met to allow the TLS handshake to succeed:
  • Certificate chain MUST be valid.
  • Certificate chain MUST include one of the pinned certificates.
  • Challenge host MUST match the host in the certificate chain's leaf certificate.
• insecure.expired-apis.com will never evaluate the certificate chain and will always allow the TLS handshake to succeed.
• All other hosts will use the default evaluation provided by Apple.

Subclassing Server Trust Policy Manager

If you find yourself needing more flexible server trust policy matching behavior (i.e. wildcarded domains), then subclass the ServerTrustPolicyManager and override the serverTrustPolicyForHost method with your own custom implementation.

class CustomServerTrustPolicyManager: ServerTrustPolicyManager {
    override func serverTrustPolicy(forHost host: String) -> ServerTrustPolicy? {
        var policy: ServerTrustPolicy?

        // Implement your custom domain matching behavior...

        return policy
    }
}

Validating the Host

The .performDefaultEvaluation, .pinCertificates and .pinPublicKeys server trust policies all take a validateHost parameter. Setting the value to true will cause the server trust evaluation to verify that hostname in the certificate matches the hostname of the challenge. If they do not match, evaluation will fail. A validateHost value of false will still evaluate the full certificate chain, but will not validate the hostname of the leaf certificate.

It is recommended that validateHost always be set to true in production environments.

Validating the Certificate Chain

Pinning certificates and public keys both have the option of validating the certificate chain using the validateCertificateChain parameter. By setting this value to true, the full certificate chain will be evaluated in addition to performing a byte equality check against the pinned certificates or public keys. A value of false will skip the certificate chain validation, but will still perform the byte equality check.

There are several cases where it may make sense to disable certificate chain validation. The most common use cases for disabling validation are self-signed and expired certificates. The evaluation would always fail in both of these cases, but the byte equality check will still ensure you are receiving the certificate you expect from the server.

It is recommended that validateCertificateChain always be set to true in production environments.

App Transport Security

With the addition of App Transport Security (ATS) in iOS 9, it is possible that using a custom ServerTrustPolicyManager with several ServerTrustPolicy objects will have no effect. If you continuously see CFNetwork SSLHandshake failed (-9806) errors, you have probably run into this problem. Apple's ATS system overrides the entire challenge system unless you configure the ATS settings in your app's plist to disable enough of it to allow your app to evaluate the server trust.

If you run into this problem (high probability with self-signed certificates), you can work around this issue by adding the following to your Info.plist.

<dict>
    <key>NSAppTransportSecurity</key>
    <dict>
        <key>NSExceptionDomains</key>
        <dict>
            <key>example.com</key>
            <dict>
                <key>NSExceptionAllowsInsecureHTTPLoads</key>
                <true/>
                <key>NSExceptionRequiresForwardSecrecy</key>
                <false/>
                <key>NSIncludesSubdomains</key>
                <true/>
                <!-- Optional: Specify minimum TLS version -->
                <key>NSTemporaryExceptionMinimumTLSVersion</key>
                <string>TLSv1.2</string>
            </dict>
        </dict>
    </dict>
</dict>

Whether you need to set the NSExceptionRequiresForwardSecrecy to NO depends on whether your TLS connection is using an allowed cipher suite. In certain cases, it will need to be set to NO. The NSExceptionAllowsInsecureHTTPLoads MUST be set to YES in order to allow the SessionDelegate to receive challenge callbacks. Once the challenge callbacks are being called, the ServerTrustPolicyManager will take over the server trust evaluation. You may also need to specify the NSTemporaryExceptionMinimumTLSVersion if you're trying to connect to a host that only supports TLS versions less than 1.2.

It is recommended to always use valid certificates in production environments.

Using Self-Signed Certificates with Local Networking

If you are attempting to connect to a server running on your localhost, and you are using self-signed certificates, you will need to add the following to your Info.plist.

<dict>
    <key>NSAppTransportSecurity</key>
    <dict>
        <key>NSAllowsLocalNetworking</key>
        <true/>
    </dict>
</dict>

According to Apple documentation, setting NSAllowsLocalNetworking to YES allows loading of local resources without disabling ATS for the rest of your app.

Network Reachability

The NetworkReachabilityManager listens for reachability changes of hosts and addresses for both WWAN and WiFi network interfaces.

let manager = NetworkReachabilityManager(host: "www.apple.com")

manager?.listener = { status in
    print("Network Status Changed: \(status)")
}

manager?.startListening()

Make sure to remember to retain the manager in the above example, or no status changes will be reported. Also, do not include the scheme in the host string or reachability won't function correctly.

There are some important things to remember when using network reachability to determine what to do next.

• Do NOT use Reachability to determine if a network request should be sent.
  • You should ALWAYS send it.
• When Reachability is restored, use the event to retry failed network requests.
  • Even though the network requests may still fail, this is a good moment to retry them.
• The network reachability status can be useful for determining why a network request may have failed.
  • If a network request fails, it is more useful to tell the user that the network request failed due to being offline rather than a more technical error, such as "request timed out."