Fetch

Fetch is a resource based network abstraction based on Alamofire

• Define API using typed resources
• Decode HTTP response body using Decodable
• Encode HTTP request body using Encodable
• Root keys: decode multiple wrapper container
• Logging
• Stubbing
  • Simple response or error
  • Random stub from array
  • Cycle through stub array
• Caching
  • Multiple caching policies
  • In-memory cache
  • Disk cache

First setup the APIClient using a Config. It's just one line. Call it in the application(_:didFinishLaunchingWithOptions:) function.

APIClient.shared.setup(with: Config(baseURL: URL(string: "https://api.github.com")!))

Lets create a struct named Organization with a few properties. The model will be parsed from the network response.

struct Organization: Decodable, Equatable {
    let id: Int
    let name: String
    let location: String
}

A Resource contains all necessary information to create a network request and parse the response.

let resource = Resource<Organization>(
    method: .get,
    path: "/orgs/allaboutapps")

Sending the request and parsing the response into the typed model Organisation:

resource.request { (result) in
    switch result {
    case .success(let networkResponse):
        print("Status code:", networkResponse.urlResponse.statusCode)
        print("Model:", networkResponse.model)
    case .failure(let apiError):
        print("Error:", apiError)
    }
}

Per default the configuration uses the JSONDecoder and JSONEncoder provided by the standard library but is not limited to it, both of these types have been extended to conform to ResourceDecoderProtocol and ResourceEncoderProtocol that allows you to define your own custom decoder/encoder. The Resource provides decoding and encoding closures that use the decoder and encoder defined in the configuration. If you want to implement a different behaviour for a resource you can provide a closure during the creation of a resource.

Payload unwrapping

Sometimes there is content that is packed in an envelop and makes parsing difficult. In this case you can define so called "root keys". Root keys define a path to the content in the envelop you want to parse. This means that only the content defined with the root keys will be parsed.

Example

This is a response that should be parsed.

{
  "data": {
    "people": [
        {
          "name": "Alex"
        },
        {
          "name": "Jeff"
        },
        {
          "name": "Tom"
        },
        {
          "name": "Xavier"
        }
    ]
  }
}

We only want the people which is an array of Person. Instead of defining a structure that models the hierachy we define "root keys" on a resource to only get the array.

struct Person: Decodable {
    let name: String
}

let resource = Resource<[Person]>(
    path: "/people",
    rootKeys: ["data", "people"]
)
resource.request { result in
...
}

Fetch gives you a versatile set of possibilities to perform stubbing. To perform stubbing shouldStub on APIClients Config have to be enabled and a stub have to be registered for a resource.

Simulate a successful network request with a json response

let stub = StubResponse(statusCode: 200, fileName: "success.json", delay: 2.0)
        
let resource = Resource<Person>(path: "/test")
    
APIClient.shared.stubProvider.register(stub: stub, for: resource)
    

The above stub will return a 200 status code with the content from the success json file loaded from your app's bundle and will be delayed by two seconds.

Simulate an unauthorized error

let stub = StubResponse(statusCode: 401, fileName: "unauthorized.json", delay: 2.0)
        
let resource = Resource<Person>(path: "/unauthorized")
    
APIClient.shared.stubProvider.register(stub: stub, for: resource)

Stubbing is not limited to json only, you can also provide raw data or provide an instance which conforms to the Encodable protocol.

Stubbing with Encodable

struct Person: Encodable {
    let name: String
    let age: Int
}
    
let peter = Person(name: "Peter", age: 18)

let stub = StubResponse(statusCode: 200, encodable: peter, delay: 2.0)
        
let resource = Resource<Person>(path: "/peter")
    
APIClient.shared.stubProvider.register(stub: stub, for: resource)

Alternating stubbing

let successStub = StubResponse(statusCode: 200, fileName: "success.json", delay: 0.0)
let failureStub = StubResponse(statusCode: 404, fileName: "notFound.json", delay: 0.0)

let alternatingStub = AlternatingStub(stubs: [successStub, failureStub])
       
let resource = Resource<Person>(path: "/peter")
   
APIClient.shared.stubProvider.register(stub: alternatingStub, for: resource)

Every time the resource is executed it will iterate over the given stubs and always return a different stub than before.

Random stubbing

The RandomStub works similar to the AlternatingStub but always returns a random stub from the array.

Conditional stubbing

Simulating behaviour based on specific conditions is something that can be realised with conditional stubbing.

Example

Simulate an endpoint that is protected by user authorization and return a success or an error based on the authorization state of your app

let conditionalStub = ClosureStub { () -> Stub in
let unauthorizedStub = StubResponse(statusCode: 401, data: Data(), delay: 2)
let okStub = StubResponse(statusCode: 200, data: Data(), delay: 2)
  return CredentialsController.shared.currentCredentials == nil ? unauthorizedStub : okStub
}

let resource = Resource(path: "/auth/secret")
    
APIClient.shared.stubProvider.register(stub: conditionalStub, for: resource)

Custom stubbing

You can create a custom stub by conforming to the Stub protocol.

struct CustomStub: Stub {
...
}

Custom StubProvider

You can create a custom StubProvider by conforming to the StubProvider protocol.

struct CustomStubProvider: StubProvider {
...
}

Init/Setup APIClient with custom stubProvider

let client = APIClient(config: Config(stubProvider: customStubProvider))
APIClient.shared.setup(with: Config(stubProvider: customStubProvider))

Or, replace default StubProvider on APIClient

APIClient.shared.setStubProvider(customStubProvider)

The following cache types are implemented:

• Memory
• Disk
• Hybrid

Setting up a cache

let cache = MemoryCache(defaultExpiration: .seconds(3600))
        
let config = Config(
    baseURL: URL(string: "https://example.com")!,
    cache: cache,
    cachePolicy: .networkOnlyUpdateCache)

let client = APIClient(config: config)

Note: To make use of caching the model you load from a resource has to conform to Cacheable.

Hybrid Cache

The hybrid cache allows you to combine two separate caches, the cache types used are not limited.

Custom cache implementations

To implement a custom cache you have to create a class/struct which conforms to the Cache protocol.

class SpecialCache: Cache {
    ...
}

Caching Policies

A Cache Policy defines the loading behaviour of a resource. You can set a policy directly on a resource when it is created, in the configuration of an APIClient or you can pass it as an argument to the fetch function of the resource.

Note: The policy defined in the resource is always preferred over the policy defined in the configuration.

Load from cache otherwise from network

This will first try to read the requested data from the cache, if the data is not available or expired the data will be loaded from the network.

let resource: Resource<X> = ...
resource.fetch(cachePolicy: .cacheFirstNetworkIfNotFoundOrExpired) { (result, finishedLoading) in 
    ...
}

Load from network and update cache

This will load the data from network and update the cache. The completion closure will only be called with the value from the network.

let resource: Resource<Person> = ...
resource.fetch(cachePolicy: .networkOnlyUpdateCache) { (result, finishedLoading) in 
...
}

Load data from cache and always from network

This will load data from the cache and load data from the network. You will get both values in the completion closure asynchronously.

let resource: Resource<Person> = ...
resource.fetch(cachePolicy: .cacheFirstNetworkAlways) { (result, finishedLoading) in 
    ...
}

For an overview of policies check out the implementation in Cache.swift

Use Xcode 11+: Go to Project > Swift Packages > + and enter git@github.com:allaboutapps/Fetch.git

Or update your Package.swift file manually:

dependencies: [
    .package(url: "git@github.com:allaboutapps/Fetch.git", from: "1.0.9"),
    ....
],
targets: [
    .target(name: "YourApp", dependencies: ["Fetch"]),
]

• iOS 11.0+
• Xcode 10.2+
• Swift 5+

• Create something awesome, make the code better, add some functionality, whatever (this is the hardest part).
• Fork it
• Create new branch to make your changes
• Commit all your changes to your branch
• Submit a pull request