How to test wether a function in Swift 2.0 throws or not? How to assert that the correct ErrorType is thrown?
EDIT: Updated the code for Swift 4.1 (still valid with Swift 5.2)
Here's the latest Swift version of Fyodor Volchyok's answer who used XCTAssertThrowsError:
enum MyError: Error {
case someExpectedError
case someUnexpectedError
}
func functionThatThrows() throws {
throw MyError.someExpectedError
}
func testFunctionThatThrows() {
XCTAssertThrowsError(try functionThatThrows()) { error in
XCTAssertEqual(error as! MyError, MyError.someExpectedError)
}
}
If your Error enum has associated values, you can either have your Error enum conform to Equatable, or use the if case statement:
enum MyError: Error, Equatable {
case someExpectedError
case someUnexpectedError
case associatedValueError(value: Int)
}
func functionThatThrows() throws {
throw MyError.associatedValueError(value: 10)
}
// Equatable pattern: simplest solution if you have a simple associated value that can be tested inside 1 XCTAssertEqual
func testFunctionThatThrows() {
XCTAssertThrowsError(try functionThatThrows()) { error in
XCTAssertEqual(error as! MyError, MyError.associatedValueError(value: 10))
}
}
// if case pattern: useful if you have one or more associated values more or less complex (struct, classes...)
func testFunctionThatThrows() {
XCTAssertThrowsError(try functionThatThrows()) { error in
guard case MyError.associatedValueError(let value) = error else {
return XCTFail()
}
XCTAssertEqual(value, 10)
// if you have several values or if they require more complex tests, you can do it here
}
}
At least of Xcode 7.3 (maybe earlier) you could use built-in XCTAssertThrowsError():
XCTAssertThrowsError(try methodThatThrows())
If nothing is thrown during test you'll see something like this:
If you want to check if thrown error is of some concrete type, you could use errorHandler parameter of XCTAssertThrowsError():
enum Error: ErrorType {
case SomeExpectedError
case SomeUnexpectedError
}
func functionThatThrows() throws {
throw Error.SomeExpectedError
}
XCTAssertThrowsError(try functionThatThrows(), "some message") { (error) in
XCTAssertEqual(error as? Error, Error.SomeExpectedError)
}
Given the following functions and declarations:
enum SomeError: ErrorType {
case FifthError
case FirstError
}
func throwingFunction(x: Int) throws {
switch x {
case 1:
throw SomeError.FirstError
case 5:
throw SomeError.FifthError
default:
return
}
}
This function will throw a FifthError if 5 is given to the function and FirstError if 1 is given.
To test, that a function successfully runs the unit test could look as follows:
func testNotError() {
guard let _ = try? throwingFunction(2) else {
XCTFail("Error thrown")
return
}
}
The let _ may also be replaced by any other name, so you can further test the output.
To assert that a function throws, no matter what ErrorType the unit test could look like this:
func testError() {
if let _ = try? throwingFunction(5) {
XCTFail("No error thrown")
return
}
}
If you want to test for a specific ErrorType it's done with a do-catch-statement. This is not the best way compared to other languages.
You have to make sure that you...
return in the catch for the correct ErrorType
XCTFail() and return for all other catch
XCTFail() if no catch is executed
Given this requirements a test case could look like this:
func testFifthError() {
do {
try throwingFunction(5)
} catch SomeError.FifthError {
return
} catch {
XCTFail("Wrong error thrown")
return
}
XCTFail("No error thrown")
}
Swift 4.1 Error throwing Test for associated values
enum ParseError: Error, Equatable {
case unexpectedArgument(String)
}
func testWithNoSchemaButWithOneArgument() {
XCTAssertThrowsError(try Args(withSchema: "", andArguments: ["-x"])) { error in
XCTAssertEqual(error as? ParseError, ParseError.unexpectedArgument("Argument(s) -x unexpected."))
}
}
You can use this function:
func XCTAssertThrowsError<T, E: Error & Equatable>(
_ expression: #autoclosure () throws -> T,
error: E,
in file: StaticString = #file,
line: UInt = #line
) {
var thrownError: Error?
XCTAssertThrowsError(
try expression(),
file: file,
line: line) {
thrownError = $0
}
XCTAssertTrue(
thrownError is E,
"Unexpected error type: \(type(of: thrownError))",
file: file,
line: line
)
XCTAssertEqual(
thrownError as? E,
error,
file: file,
line: line
)
}
Example:
XCTAssertThrowsError(try funcThatThrowsSpecificError(), error: SpecificErrorEnum.someError)
Related
In order to track server errors from a Restfull Api on an SwiftUI IOS App I am feeding the error message from NSData to the "reason" parameter of next NetworkErrors enum:
enum NetworkError: Error {
case domainError(reason:String)
case decodingError(reason:String)
case encodingError(reason:String)
}
The error reason is fed when decoding the NSURLSession response:
static func postRequest<T:Decodable, U:Codable>(_ endpoint:String, _ input:U?, completion: #escaping (Result<T,NetworkError>) -> Void) {
...
do {
let retval = try JSONDecoder().decode(T.self, from: data)
completion(.success(retval))
} catch let DecodingError.dataCorrupted(context) {
let responseData = String(data: data, encoding: String.Encoding.utf8)
completion(.failure(.decodingError(reason: responseData ?? "Data corrupted from response")))
} catch {
...
}
...
}
The error reason should be available on next code, but I'm only able to print the localizedDescription:
Button(action:{
self.postRequest(endpoint, self.value){ (result: Result<Bool,NetworkError>) in
switch result {
case .success:
print("value saved successfully")
case .failure(let error):
print("failure to save value")
print(error.localizedDescription)
}
}
}){
Image(systemName:"icloud.and.arrow.up")
}
In the failure case, we know that error is a NetworkError, so now disassemble that error with another switch:
switch error {
case .domainError(let reason): // do something
case .decodingError(let reason): // do something
case .encodingError(let reason): // do something
}
with RxSwift 3.6.1 I made this extension to ObservableType to get a new token after an error request:
public extension ObservableType where E == Response {
public func retryWithToken() -> Observable<E> {
return retryWhen { error -> Observable<Response> in
return error.flatMap({ (error) -> Observable<Response> in
if let myApiError: MyApiError = error as? MyApiError {
if (myApiError == MyApiError.tokenError) {
return Session.shared.myProvider.request(.generateToken)
} else {
return Observable.error(myApiError)
}
}
return Observable.error(error)
})
}
}
}
and then I can use it:
Session.shared.myProvider.rx
.request(.mySampleRequest)
.filterSuccessfulStatusCodes()
.retryWithToken()
.subscribe { event in
....
}.disposed(by: self.disposeBag)
but with RxSwift 4.0.0 now the sequence expect a
PrimitiveSequence<SingleTrait, Response>
someone can explain to me how to do the same with RxSwift 4.0.0? I try with an extension to PrimitiveSequence but I've some compilation errors.
I believe that has nothing to do with RxSwift but is a Moya change. MoyaProvider.rx.request returns Single which is a typealias for PrimitiveSequence which is not an ObservableType.
You declare your function upon the ObservableType.
So just do asObservable() before retryWithToken()
I've defined a protocol with a protocol extension to simplify working with NSError.
protocol NSErrorConvertible: RawRepresentable {
var domain: String { get }
var localizedDescription: String { get }
}
extension NSErrorConvertible where RawValue == Int {
func generateError(parameters parameters: [String] = []) -> NSError {
let error = self.generateError(format: self.localizedDescription, parameters: parameters)
return error
}
func generateError(format format: String, parameters: [String] = []) -> NSError {
let description = String(format: format, arguments: parameters) // <-- BREAK
let error = NSError(domain: self.domain, code: self.rawValue, localizedDescription: description)
return error
}
}
extension NSError {
convenience init(domain: String, code: Int, localizedDescription: String) {
let userInfo = [NSLocalizedDescriptionKey : localizedDescription]
self.init(domain: domain, code: code, userInfo: userInfo)
}
}
Here's how the protocol is used:
enum DefaultEngineErrors: Int, NSErrorConvertible {
case ImagesNotSupported
case FooDoesNotHaveABar
case NilBar
var domain: String { return "DefaultEngineErrors" }
var localizedDescription: String {
switch self {
case .ImagesNotSupported: return "%s: Images are not supported."
case .FooDoesNotHaveABar: return "%s: Foo does not have a bar."
case .NilBar: return "%s: The bar is nil."
}
}
}
And here is how the protocol extension methods are called:
let error = DefaultEngineErrors.FooDoesNotHaveABar.generateError(parameters: ["\(foo.id)"])
When this code is run, the debugger breaks on the line that builds the description in the second generateError extension function and the following message is displayed in the debugger console:
fatal error: can't unsafeBitCast between types of different sizes
Note that if I replace the call to String(format:) with just format, everything works fine.
Q: What is wrong with what I've done here?
In addition to this, I can't examine the parameters in the protocol extension. The following message is displayed when I try:
(lldb) po format
error: <EXPR>:1:1: error: non-nominal type '$__lldb_context' (aka 'Self') cannot be extended
extension $__lldb_context {
^ ~~~~~~~~~~~~~~~
<EXPR>:15:5: error: value of type 'DefaultEngineErrors' has no member '$__lldb_wrapped_expr_54'
$__lldb_injected_self.$__lldb_wrapped_expr_54(
^~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~
Swift string Array is not compatible with CVarArgType... or [CVarArgType]
try using.
extension NSErrorConvertible where RawValue == Int {
func generateError(format format: String) -> NSError {
let error = NSError(domain: self.domain, code: self.rawValue, localizedDescription: format)
return error
}
func generateError(args : CVarArgType...) -> NSError {
let returnString = NSString(format: self.localizedDescription, arguments: getVaList(args)) as String
// let returnString = withVaList(args) {
// NSString(format: self.localizedDescription, arguments: $0)
// } as String
let error = generateError(format: returnString)
return error
}
}
enum DefaultEngineErrors: Int, NSErrorConvertible {
case ImagesNotSupported
case FooDoesNotHaveABar
case FooWithMultiPar
case NilBar
var domain: String { return "DefaultEngineErrors" }
var localizedDescription: String {
switch self {
case .ImagesNotSupported: return "%#: Images are not supported."
case .FooDoesNotHaveABar: return "%#: Foo does not have a bar."
case .FooWithMultiPar: return "%#:%# Foo does not have a bar."
case .NilBar: return "%#: The bar is nil."
}
}
}
print(DefaultEngineErrors.FooDoesNotHaveABar.generateError("test"))
print(DefaultEngineErrors.FooWithMultiPar.generateError("test1","test2"))
Check this
This should be simple and obvious but try catch in swift2 frustrating me. here is my code.
#IBAction func btnAdd_TouchDown(sender: AnyObject) {
do { try AddNewNotesToList() } catch{
if(Ex.UnknownError != "")
{
Error(Ex.UnknownError)
}
}
}
func AddNewNotesToList() throws
{
var obj: TreatmentNotesDTO = TreatmentNotesDTO()
obj.therapist_id = Int(TherapistId)! // getting error here
return
}
Error:
fatal error: unexpectedly found nil while unwrapping an Optional value
Debugger should go to catch but its breaking up. I am from c# and just started swift2. Any help
I would suggest you to use guard to catch such errors:
guard let therapistID = Int(TherapistId) else {
print(TherapistId + " cannot be converted to an Int")
return
// or mark function as throws and return an error
throw NSError(domain: TherapistId + " cannot be converted to an Int", code: 0, userInfo: nil)
}
obj.therapist_id = therapistID
Note that Int(TherapistId)! doesn't throw an error with throws. It is a fatal error which stops program execution.
I am trying to run the query to get object in background with ID but when I run the method query.getObjectInBackgroundWithId , I am getting the error message:
"Cannot invoke 'getObjectInBackgroundWithId' with an argument list of type (string, block: (PFObject!,NSError?) -> Void"
The same thing happens when I use user.signUpInBackgroundWithBlock so I'm thinking maybe Xcode updated some of the features while using 'block's and now maybe the syntax is different? Any ideas?
Here's a snippet of my code:
http://imgur.com/1CvfhbU
YES!!! Thank you!
The new sample code for getObject is:
query.getObjectInBackgroundWithId("Oaq79bhv53") {
(gameScore: PFObject?, error: NSError?) -> Void in
if error == nil && gameScore != nil {
println(gameScore)
} else {
println("error")
}
}
I figured it out!
user.signUpInBackgroundWithBlock{(succeeded: Bool, signUpError: NSError?) -> Void in
Swift 4.2:
query.getObjectInBackground(withId: "YourId") { (gameScore, error) in
if error == nil {
// Success!
print(gameScore)
} else {
// Fail!
}
}