This question already has answers here:
Alamofire Response Serialization Failed
(2 answers)
Closed 6 months ago.
I have an API where I PUT stuff to. I need to make sure to wait until I get an http 200 response from the server, but I don't know how to await that using Alamofire because my response itself if empty. So it's just an http 200 with no content.
I only can find async functions that e.g. serialize a String or Data or a Decodable, but they don't work if my response is empty.
Is there a way to await something like that in Alamofire?
I know that your question is about async/await from Alamofire, but is good to know that the http status codes 204 and 205 are exactly for this. Which means that if you have access to the server code you could send the empty responses with the http status code 204 and 205 instead of 200 and then Alamofire would not generate any errors. But assuming you don't have access to the server code and you need to parse an empty response as correct then you could use the following code:
func testRequestWithAlamofire() {
let dataResponseSerializer = DataResponseSerializer(emptyResponseCodes: [200, 204, 205]) // Default is [204, 205] so add 200 too :P
AF.request("http://www.mocky.io/v2/5aa696133100001335e716e0", method: .put).response(responseSerializer: dataResponseSerializer) { response in
switch response.result {
case .failure(let error):
print(error)
case .success(let value):
print(value)
}
}
}
And for a real and complete example of how async/await from Alamofire or any other async context look this code:
// This function get report from API and save to a local JSON to be readed by the app
func updateReport() {
Task {
guard let session = self.sessionRepository.getSession(WithUser: Defaults.lastLoggedUsername!) else { return }
guard let company = session.profile?.companies.first else { return }
self.apiManager.configure(WithToken: session.accessToken)
do {
let dateA = Date().dateAtStartOf(.year)
//let dateB = Date().dateAtEndOf(.month)
let dateB = Date() // Just now
let report = try await self.apiManager.report(CompanyId: company._id, DateA: dateA, DateB: dateB, ChartPeriodicity: .month)
self.currentReport = report
// Save data to disk to be read later
self.reportManager.saveReportToDisk(report: report!, withProfileId: session.profile!._id)
} catch {
print("Error getting report: \(error)")
}
}
}
// Get personal report from a given date range
func report(CompanyId companyId: String, DateA dateA: Date, DateB dateB: Date, ChartPeriodicity chartPeriodicity: ChartPeriodicity) async throws -> CDReport? {
try await withCheckedThrowingContinuation { continuation in
self.contappApi.request(.report(companyId: companyId, dateA: dateA, dateB: dateB, chartPeriodicity: chartPeriodicity)) { result in
switch result {
case let .success(response):
// Check status code
guard response.statusCode == 200 else {
continuation.resume(throwing: ContappNetworkError.unexpected(code: response.statusCode))
return
}
// Decode data
do {
//let report = try JSONDecoder().decode(CDReport.self, from: response.data)
let report = try CDReport(data: response.data)
continuation.resume(returning: report)
} catch {
continuation.resume(throwing: ContappNetworkError.cantDecodeDataFromNetwork)
}
case .failure(_):
continuation.resume(throwing: ContappNetworkError.networkError)
}
}
}
}
Alamofire already supports this, you just need to choose a form. Your biggest issue will be accepting a 200 with no data, as that's technically invalid since only 204 or 205 are supposed to be empty.
All Alamofire responses require some sort of payload type, but Alamofire provides an Empty type to fill this role for Decodable. So the simplest way is to use the
await AF.request(...)
.serializingDecodable(Empty.self, emptyResponseCodes: [200])
.response
Note, if you already have an Empty type or are importing Combine in the same file as this code, you may need to disambiguate by using Alamofire.Empty.
If Alamofire does not provide a method for your purpose, then you will have wrap the old Alamofire methods that uses closures as below:
func myRequest() async throws {
try await withUnsafeThrowingContinuation { continuation in
myAlamofireRequest {
continuation.resume()
}
}
}
Related
Ive got an API endpoint that i have got to take some data from but it needs an access token.
This access token is fetched from another endpoint of this api.
The access token expires every 2:30 hrs.
The way I am handling this is that every 2:20 hrs i have a a timer that fetches a new token. I know this is a bad practice since the user might turn off the internet during that fetching etc.
I am using an architectural pattern that splits my main app into 3 seperate layers.
A domain layer which contains all my models, use cases and repositories
A presentation layer which contains all my views and viewmodels.
And a Data layer that contains all my repository implementations , network constants, url builders and my API client where the request is made with alamofire.
My ApiClient is this :
public enum ApiClient {
static func requestCodable<T: Codable>(_ urlConvertible: URLRequestConvertible) -> AnyPublisher<DataResponse<T, NetworkErrorResponse>, Never> {
return AF.request(urlConvertible)
.validate()
.publishDecodable(type: T.self, emptyResponseCodes: [200])
.map { response in
response.mapError { error in
let backendError = response.data.flatMap { try? JSONDecoder().decode(BackendError.self, from: $0) }
return NetworkErrorResponse(initialError: error, backendError: backendError)
}
}
.receive(on: DispatchQueue.main)
.eraseToAnyPublisher()
}
}
So , lets assume that i am making a call to an endpoint that needs the access token, and it fails because 3 hours have gone by ...
How can I say to my api client to fetch a new access token and then retry the endpoint that failed ?
Thanks for any help in advance.
Since I'm not aware of your implementation I'll try to give you what I think is a possible solution:
public enum ApiClient {
static func requestCodable<T: Codable>(_ urlConvertible: URLRequestConvertible, isRetry: Bool = false) -> AnyPublisher<DataResponse<T, NetworkErrorResponse>, Never> {
return AF
.request(urlConvertible)
.validate()
.publishDecodable(type: T.self, emptyResponseCodes: [200])
.map { response in
response.mapError { error in
let backendError = response.data.flatMap { try? JSONDecoder().decode(BackendError.self, from: $0) }
return NetworkErrorResponse(initialError: error, backendError: backendError)
}
}
.flatMap { result -> AnyPublisher<DataResponse<T, NetworkErrorResponse>, Never> in
if !isRetry && result == "401 error code" {
return requestAccessToken()
.flatMap({
// probably build a new urlConvertible with the new token
self.requestCodable(urlConvertible, isRetry: true)
})
} else {
return Just(result).eraseToAnyPublisher()
}
}
.receive(on: DispatchQueue.main)
.eraseToAnyPublisher()
}
}
You could add a flatMap after your map and verify if you got the 401 error. If so, you would need to get the token, then use flatMap to re-request the requestCodable. You might notice there's a isRetry param, it is there to avoid loops.
I am currently writing an Alamofire HTTP request and am running into an issue where my view is not loading - likely because there is no data. The confusing part is that this was working yesterday. In the request I was able to do print(data) and the result was 506 bytes which, if my calculation is correct, is about the correct size given the JSON payload returned from the endpoint below.
#State var recipes = [Recipe]()
AF.request("http://localhost:3000/recipes").responseJSON { response in
guard let data = response.data else { return }
if let response = try? JSONDecoder().decode([Recipe].self, from: data) {
DispatchQueue.main.async {
self.recipes = response
}
return
}
}
I can confirm that the endpoint that is being hit returns the following data...
[
{
"name":"Manhattan",
"image":"https://images.unsplash.com/photo-1536935338788-846bb9981813?ixid=MnwxMjA3fDB8MHxwaG90by1wYWdlfHx8fGVufDB8fHx8&ixlib=rb-1.2.1&auto=format&fit=crop&w=2486&q=80",
"spirit":"Bourbon",
"ice":"Crushed",
"glass":"Coupe",
"yield":"3.75",
"description":"This is a good drink. You should make it.",
"ingredients":[
{
"bottle":"High West Son Of Bourye",
"amount":"2.5"
},
{
"bottle":"Cocchi Vermouth Di Torino",
"amount":"0.75"
},
{
"bottle":"Simple Syrup",
"amount":"0.083"
}
]
}
]
I also have my Recipe and Ingredient model here which should be able to decode based on the above JSON.
struct Recipe: Decodable, Identifiable {
var id = UUID()
var name: String
var image: String
var spirit: String
var ice: String
var glass: String
var yield: String
var description: String
var ingredients: [Ingredient]
}
struct Ingredient: Decodable, Identifiable {
var id = UUID()
var bottle: String
var amount: String
}
Is anybody able to spot an issue? I was trying to put a debugging print in the DispatchQueue but it is not printing which, to me, sounds like an error. However I am new to Swift/XCode/iOS and am not sure the best debugging practices for this.
If you can't debug yourself, NEVER USE try?. With more experience, I'd say that we tend to not use try?, but sometimes we do. But when we write try?, we are able to find an possible issue, ie debug if needed.
Let's do a proper try then, with a do/catch:
do {
let response = try JSONDecoder().decode([Recipe].self, from: data
DispatchQueue.main.async {
self.recipes = response
}
} catch {
print("Oops, there was en error while decoding: \(error)") // and not error.localizedDescription as it's more for users than developpers, so you'll skip all the useful informations
}
And read the output.
Going further?
Don't believe what's the API is supposed to return.
I've seen plenty and plenty of questions where the returned values was an error message, a XML Error message, a JSON Error message, an HTML Error message, and a JSON value missing, or of bad type, etc. And that, your JSONDecoder wasn't expecting it...
Reasons could be various, from bad/missing parameters, bad/missing APIKey, server down, bad/missing header, etc.
But, then, print the returned value.
print(String(data: data, encoding: .utf8) ?? "No data found")
So print it directly when you get it, or at least in the catch:
} catch {
print("Oops, there was en error while decoding: \(error)") // and not error.localizedDescription as it's more for users than developpers, so you'll skip all the useful informations
print("While getting response stringified: \(String(data: data, encoding: .utf8) ?? "No data found")")
}
If you don't understand the error message output, it's okay, there is no shame about it. But your first job is to get that error message. You can share it on SO if you don't understand it, you might get help with that. But currently, we can't guess what's wrong with your code.
It's a good idea to drop some clues in your code when looking for a failure.
If it were me I'd do something like this:
AF.request("http://localhost:3000/recipes").responseJSON { response in
guard let data = response.data else {
print("Error trying to receive data in ", #file, #function)
return
}
do {
let response = try JSONDecoder().decode([Recipe].self, from: data) {
DispatchQueue.main.async {
self.recipes = response
}
} catch {
print("Error failed to decode json data with error: \(error) in \(#file)", #function)
}
}
I have an API client that uses generic API response that conforms to Codable Protocol and uses JSONDecoder to decode the response as shown below, how do I handle having a response which doesn't return JSON ( status code 201 created)?
dataRequest.validate().responseJSON { response in
if let error = response.error {
completion(.failure(error.localizedDescription))
} else if let data = response.data {
do {
let apiResponse = try JSONDecoder().decode(T.Response.self, from: data)
completion(.success(apiResponse))
} catch {
completion(.failure(error.localizedDescription))
}
} else {
completion(.failure("Something went wrong, please try again later."))
}
}
It returns this error:
the response could not be serialized input data was nil or zero-length
In this case you can look at the statusCode property of the response (assuming that it is a HTTPURLResponse) and make your determination about whether or not there will be a body to parse. I would put it immediately after the error check.
In my app I use Moya and Alamofire (And also Moya/RxSwift and Moya-ObjectMapper) libraries for all network requests and responses.
I would like to handle the response of all types of requests in one handler, but also uniquely handle every request.
For example for any request I can get the response "Not valid Version", I would like to avoid to check in every response if this error arrived.
Is there an elegant way to handle this use case with Moya?
Apparently that is very simple, You just should create your own plugin. And add it to your Provider instance (You can add it in the init function)
For example:
struct NetworkErrorsPlugin: PluginType {
/// Called immediately before a request is sent over the network (or stubbed).
func willSendRequest(request: RequestType, target: TargetType) { }
/// Called after a response has been received, but before the MoyaProvider has invoked its completion handler.
func didReceiveResponse(result: Result<Moya.Response, Moya.Error>, target: TargetType) {
let responseJSON: AnyObject
if let response = result.value {
do {
responseJSON = try response.mapJSON()
if let response = Mapper<GeneralServerResponse>().map(responseJSON) {
switch response.status {
case .Failure(let cause):
if cause == "Not valid Version" {
print("Version Error")
}
default:
break
}
}
} catch {
print("Falure to prase json response")
}
} else {
print("Network Error = \(result.error)")
}
}
}
I suggest to use generic parametrized method.
class DefaultNetworkPerformer {
private var provider: RxMoyaProvider<GitHubApi> = RxMoyaProvider<GitHubApi>()
func performRequest<T:Mappable>(_ request: GitHubApi) -> Observable<T> {
return provider.request(request).mapObject(T.self)
}
}
DefaultNetworkPerformer will handle all requests from you Moya TargetType. In my case it was GitHubApi. Example usage of this implementation is:
var networkPerformer = DefaultNetworkPerformer()
let observable: Observable<User> = networkPerformer.performRequest(GitHubApi.user(username: "testUser"))
here you 'inform' network performer that response will contain User object.
observable.subscribe {
event in
switch event {
case .next(let user):
//if mapping will succeed here you'll get an Mapped Object. In my case it was User that conforms to Mappable protocol
break
case .error(let error):
//here you'll get MoyaError if something went wrong
break
case .completed:
break
}
}
I tried a long time to handle a JSON response with SwiftyJSON, but I don't know how to parse the response. Here's my code:
var jsonString:String = ""
Alamofire
.request(.GET, url + "/HMServer/rest/administration/version")
.responseJSON {
(request, response, data, error) -> Void in
let json = JSON(object: data!)
//here I want to do something with parsing
}
The requests I do with Alamofire and get back a JSON response. The response look like:
[message: [SERVER_VERSION: 0.1, INTERFACE_VERSION: 0.1], type: success]
I want to save all elements in strings and give them back. How can I parse the JSON response saved in the let let json? I tried to use Alamofire-SwiftJSON but the code does not work. All examples I found are too old because the SwiftyJSON code was refactored a few days ago.
THX!
I have fixed Alamofire-SwiftJSON's issue,
but you can do it by yourself in responseJSON's closure like:
Alamofire.request(.GET, url + "/HMServer/rest/administration/version")
.responseJSON { (request, response, data, error) -> Void in
if error != nil {
self.swiftyJSON = SwiftyJSON.JSON.Null(error)
} else if object != nil {
self.swiftyJSON = SwiftyJSON.JSON(object: object!)
} else {
self.swiftyJSON = SwiftyJSON.JSON.Null(nil)
}
}
Above code is not like Alamofire-SwiftJSON in the global queue, the initialization (AnyObject to SwiftyJSON) is running in the main queue.