Related
I want to implement the following:
throwingFunction()??.doStuff()
/* if throwingFunction throws an error:
print the error
else
returns an object with the doStuff() Method
*/
throwingFunction()??
/*
if an error is thrown,
prints the error.
else
execute the function without errors.
*/
I'm not sure where to look in the source code for examples on how do, try, catch were implemented. The Swift error docs explain how to use error handle methods that are already implemented. To be clear, I want to implement custom error handling with the above syntax.
Something like:
precedencegroup Chaining {
associativity: left
}
infix operator ?? : Chaining
extension Result {
// ERROR: Unary operator implementation must have a 'prefix' or 'postfix' modifier
static func ??(value: Result<Success, Failure>) -> Success? {
switch value {
case .success(let win):
return win
case .failure(let fail):
print(fail.localizedDescription)
return nil
}
}
}
You can define a postfix operator which takes a throwing closure as (left) operand. ?? is already defined as an infix operator, therefore you have to choose a different name:
postfix operator <?>
postfix func <?><T>(expression: () throws -> T) -> T? {
do {
return try expression()
} catch {
print(error)
return nil
}
}
Now you can call
let result = throwingFunc<?>
or chain it with
let result = (throwingFunc<?>)?.doStuff()
Previous answer:
?? is already defined as an infix operator. For a postfix operator you have to choose a different name, for example:
postfix operator <?>
extension Result {
static postfix func <?>(value: Result) -> Success? {
switch value {
case .success(let win):
return win
case .failure(let fail):
print(fail.localizedDescription)
return nil
}
}
}
Now you can call
let res = Result(catching: throwingFunc)<?>
or chain it with
let res = (Result(catching: throwingFunc)<?>)?.doStuff()
There is little chance that you can do this, without actually forking apple/swift and creating your own version of the compiler... Here are my attempts:
First, I noticed that the second part of the desired result, ?.doStuff() looks exactly like a optional chaining expression. I thought I could make a postfix ? operator that returned an optional. But it turns out, I can't declare an ? operator at all:
postfix operator ? // error
So I used a visually similar character - ‽ - instead. The type of a throwing function is () throws -> Void and I used #autoclosure so that the {} can be omitted:
typealias ThrowingFunction<T> = () throws -> T
postfix operator ‽
postfix func ‽<T>(lhs: #autoclosure ThrowingFunction<T>) -> T? {
switch Result(catching: lhs) {
case .failure(let error):
print(error.localizedDescription)
return nil
case .success(let t):
return t
}
}
// Usage:
func f() throws -> Int {
throw URLError(URLError.badURL)
}
// You have to use it like this :(
(try f()‽)
(try f()‽)?.description
The try could be omitted if the function you are calling takes no arguments:
f‽
(f‽)?.description
To make functions of other arity work without try, you need to create an implementation of ‽ for each arity, which sucks.
But the brackets must be there because of how Swift parses operators :(
Then I tried to make the approach you attempted, with key paths:
func ??<T, U>(lhs: #autoclosure ThrowingFunction<T>, rhs: KeyPath<T, U>) -> U? {
switch Result(catching: lhs) {
case .failure(let error):
print(error.localizedDescription)
return nil
case .success(let t):
return t[keyPath: rhs]
}
}
func f() throws -> Int {
throw URLError(URLError.badServerResponse)
}
This seems to be even worse, because you gotta use it like this:
try f() ?? \.description
You can't omit the try,
f ?? \.description // type inferencer freaks out, thinks T is ThrowingFunction<Int>
nor reduce the spaces on either side of ?? (See here for why):
try f()??\.description
Plus there is this backlash that is an integral part of the keypath syntax, and you can only use it for key paths, not methods. :(
Summary
You can't do this because:
You can't overload ?
You can't put a ? right after anything because it will be parsed as optional chaining
You must write try, unless you cater for every arity.
postfix operator *
#discardableResult
postfix func *<Preferred>(expression: ErrorAlt<Preferred>) -> Preferred? {
switch expression {
case .preferred(let pref):
return pref
case .error(let err):
print(err.localizedDescription)
return nil
case .initializersWereNil:
print("initializersWereNil")
return nil
}
}
Here is an example usage.
enum TestMeError: Error {
case first
}
extension Int {
func printWin() {
print("we did it!")
}
}
func testMe() -> ErrorAlt<Int> {
if true {
return .error(TestMeError.first)
} else {
return .preferred(40)
}
}
// USAGE
testMe()*
I'm trying to use Swift's new Result type in such a way that the type of the .success associated value is a generic. The rather contrived sample code below works, but is there a way to simplify the type casting so that the compiler can infer the correct type for T?
enum FetchError : Error {
case unknownKey
}
enum FetchKey {
case getWidth
case getName
}
func fetchValue<T>(_ key:FetchKey) -> Result<T, FetchError> {
switch key {
case .getName:
// Ideally I would like to just use: return .success("Johnny Appleseed")
return Result<String, FetchError>.success("Johnny Appleseed") as! Result<T, FetchError>
case .getWidth:
return Result<Double, FetchError>.success(25.0) as! Result<T, FetchError>
#unknown default:
return .failure(.unknownKey)
}
}
// This explicit type declaration is also required.
let r:Result<String, FetchError> = fetchValue(.getName)
print(r)
To your question about the type-casting, you can definitely simplify it:
case .getName:
return .success("Johnny Appleseed" as! T)
This is ok if asking for the wrong type should be considered a programming error (and so should crash), and the results will never come from external sources. If the data could ever come from an external source, then you should never crash in response to it being wrong.
In that case, we should model that kind of error:
enum FetchError : Error {
case unknownKey
case invalidType
}
Then you can have a syntax very close to what you like by adding a function to do the (possibly failing) type conversion:
func fetchValue<Value>(_ key:FetchKey) -> Result<Value, FetchError> {
func checkType(_ value: Any) -> Result<Value, FetchError> {
guard let value = value as? Value else { return .failure(.invalidType) }
return .success(value)
}
switch key {
case .getName: return checkType("Johnny Appleseed")
case .getWidth: return checkType(25.0)
#unknown default: return .failure(.unknownKey)
}
}
That said, I'd do it this way to avoid the ugliness of required type annotations:
func fetch<Value>(_: Value.Type, forKey key: FetchKey) -> Result<Value, FetchError> { ... }
let r = fetch(String.self, forKey: .getName)
This follows the pattern of Codable.
Here's the whole solution together in one place in a few different ways:
Returning Result
enum FetchError : Error {
case unknownKey
case invalidType
}
enum FetchKey {
case width
case name
}
func fetch<Value>(_: Value.Type, forKey key: FetchKey) -> Result<Value, FetchError> {
func checkType(_ value: Any) -> Result<Value, FetchError> {
guard let value = value as? Value else { return .failure(.invalidType) }
return .success(value)
}
switch key {
case .name: return checkType("Johnny Appleseed")
case .width: return checkType(25.0)
#unknown default: return .failure(.unknownKey)
}
}
With throws
I think this gets a little nicer if you throw rather than wrapping things into Result. It means you can more easily lift checkType into one place and gets very close to the syntax you said you wanted.
func fetch<Value>(_: Value.Type, forKey key: FetchKey) throws -> Value {
func checkType(value: () throws -> Any) throws -> Value {
guard let value = try value() as? Value else { throw FetchError.invalidType }
return value
}
return try checkType {
switch key {
case .name: return "Johnny Appleseed"
case .width: return 25.0
#unknown default: throw FetchError.unknownKey
}
}
}
With Optionals
This gets a little simpler with Optional if you don't really care about the errors.
func fetch<Value>(_: Value.Type, forKey key: FetchKey) -> Value? {
func _fetch() -> Any? {
switch key {
case .name: return "Johnny Appleseed"
case .width: return 25.0
#unknown default: return nil
}
}
return _fetch() as? Value
}
The function you are trying to create has a dynamically typed generic. This isn't possible as the Swift compiler needs to know the types for each method/variable at compile time.
Suppose you have
func whatToFetch() -> FetchKey {
// Randomly returns one of the FetchKey cases
}
let r = fetchValue(whatToFetch())
There is no way for the compiler to know the type of r before the call is made. Your example code works because your force-casts are perfectly lined up. But in a correct implementation you should not have to specify what type T is inside your generic function
In your case, it seems generics are not really needed and you could do away with protocols:
enum FetchKey {
case someStringValueKey
case someDoubleValueKey
case someModelValueKey
}
protocol FetchResult {}
extension String: FetchResult {}
extension Double: FetchResult {}
extension SomeModel: FetchResult {}
func fetch(_ key: FetchKey) -> Result<FetchResult,Error> {
switch key {
case .someStringValueKey:
return .success("")
case .someDoubleValueKey:
return .success(1.0)
case .someModelValueKey:
return .success(SomeModel())
}
}
let wtf = whatToFetch()
let r = fetch(wtf)
switch r {
case .success(let value):
switch wtf {
case .someStringValueKey:
guard let stringValue = value as? String else { return }
case .someDoubleValueKey:
guard let doubleValue = value as? Double else { return }
case .someModelValueKey:
guard let someModelValue = value as? SomeModel else { return }
}
case .failure(let error):
print("Better do more than just print on production code")
}
Trying to use an unqualified generic fetchValue<T> to do what you want probably isn’t going to work. The reason is that in a generic function, the T is specified by the caller, not the function. You’re essentially saying, “Ask fetchValue for any T you want, and it will give you a result.”
There’s missing link in the way you’ve set up your types. The red flag that’s your primary clue to this is the use of as!. That’s a sign that you are making assumptions about type relationships that you’re not telling the compiler about.
What does that mean? Well, note that with your code, let r: Result<URLConnection, FetchError> = fetchValue(.getName) also compiles — but then crashes at runtime!
One solution is to have an umbrella type that gathers all the possible result types, as Emil’s solution does. In this approach, you erase the type of the result, and ask callers to dynamically extract it. Callers have to deal with the possibility that the result might be of any type, not necessarily the one they expected. This is a common pattern when dealing with dynamically structured data such as JSON documents.
A second approach is to do what Rob suggests: let the caller specify the type, but translate an incorrect type into an error.
A third approach to solving this is to find a way to associate keys with result types in the type system, i.e. to tell the compiler getName → String, getWidth → Double. Unfortunately, AFAIK, there’s no way to do that with individual enum cases, so you’ll need to encode the keys as something other than an enum.
Here’s one way to do it:
enum FetchError : Error {
case unknownKey
}
protocol FetchKey {
associatedtype ValueType
func get() -> ValueType
}
struct GetNameKey: FetchKey {
func get() -> String { // This line establishes the getName → String mapping
return "Johnny Appleseed"
}
}
struct GetWidthKey: FetchKey {
func get() -> Double { // This line establishes the getWidth → Double mapping
return 25.0
}
}
// This function signature means: “If you give fetchValue a key, it
// will give you a result _with the appropriate type_ for that key”
//
func fetchValue<K: FetchKey>(_ key: K) -> Result<K.ValueType, FetchError> {
// The return type here is Result<K.ValueType, FetchError>, but
// now Swift has enough into to infer it!
return Result.success(key.get())
}
// This now works without type inference:
let r0 = fetchValue(GetNameKey())
print(r0)
// And this now (correctly) fails to compile:
// let r1: Result<Double, FetchError> = fetchValue(GetNameKey())
Which approach should you use?
Does each key always return a single, consistent type that you know beforehand (e.g. name is always a string)?
Yes: Use my approach above
No: If a fetch succeeds, but the type that came back isn’t the type the caller asked for, how do you want it reported?
As a Result.failure, and the caller can’t see the value: Use Rob’s approach
As a Result.success, and the caller has to figure out what type the value is: Use Emil’s approach
I just found another way to make a great use of protocols and protocol extensions in Swift by extending the Optional protocol to add a function so I can provide default values.
I wrote a blog post about this here: https://janthielemann.de/random-stuff/providing-default-values-optional-string-empty-optional-string-swift-3-1/
The gist of the post is that I needed a clean and easy way to provide default values for optional String which are nil or empty. To do this, I created a Emptyable protocol end extended the Optional protocol like so:
protocol Emptyable {
var isEmpty: Bool { get }
}
extension Optional where Wrapped: Emptyable {
func orWhenNilOrEmpty<T: Emptyable>(_ defaultValue: T) -> T {
switch(self) {
case .none:
return defaultValue
case .some(let value) where value.isEmpty:
return defaultValue
case .some(let value):
return value as! T
}
}
}
extension String: Emptyable {}
Now the question is: Is there a way I can get rid of the Emptyable protocol and instead have a conditional check whether or not a property or function is implemented by the generic type so that I automatically get orWhenNilOrEmpty() for each and every type which has isEmpty?
UPDATE
As suggested by Paulo, the T generic is actually not needed and I created a operator for even quicker access and more convenient usage (at least I think so. Feel free to correct me, I'm always happy to learn new things and improve myself).
I call it the "not empty nil coalescing" operator (who can come up with a better names? I feel like I suck at naming things :/ ). Hopefully some day it helps somebody:
protocol Emptyable {
var isEmpty: Bool { get }
}
infix operator ???: NilCoalescingPrecedence
extension Optional where Wrapped: Emptyable {
func orWhenNilOrEmpty(_ defaultValue: Wrapped) -> Wrapped {
switch(self) {
case .none:
return defaultValue
case .some(let value) where value.isEmpty:
return defaultValue
case .some(let value):
return value
}
}
static func ???(left: Wrapped?, right: Wrapped) -> Wrapped {
return left.orWhenNilOrEmpty(right)
}
}
extension String: Emptyable {}
extension Array: Emptyable {}
extension MyStruct: Emptyable {
let text: String
let number: Int
var isEmpty: Bool { return text.isEmpty && number == 0 }
init(text: String, number: Int) {
self.text = text
self.number = number
}
}
let mandatoryNotEmptyString = optionalOrEmptyString ??? "Default Value"
let mandatoryNotEmptyStruct = optionalOrEmptyStruct ??? MyStruct(name: "Hello World", number: 1)
No, you cannot query if an object or value has a certain property as a constraint on an extension without using a protocol. That would require reflection in a way that is currently not implemented in Swift. Also, an isEmpty property could have different meanings for different types, so testing for the existence of a method or property instead of a protocol could lead to unexpected behaviour.
You could just write
if let unwrappedString = optionalString, !unwrappedString.isEmpty {
// Do stuff
} else {
// Use default value
}
No protocol or extension required and very readable.
In Swift 4, which is coming out this fall, String will conform to BidirectionalCollection, which inherits from Collection. The Collection protocol provides an isEmpty property, so your extension could be
extension Optional where Wrapped: Collection {
// ...
}
But even then you should consider to set empty strings to nil when storing them in the first place, because you now have two states (nil and empty) which seem to represent the exact same thing.
Consider this code:
enum Type {
case Foo(Int)
case Bar(Int)
var isBar: Bool {
if case .Bar = self {
return true
} else {
return false
}
}
}
That's gross. I would like to write something like this instead:
enum Type {
case Foo(Int)
case Bar(Int)
var isBar: Bool {
return case .Bar = self
}
}
But such a construct does not seem to exist in Swift, or I cannot find it.
Since there's data associated with each case, I don't think it's possible to implement the ~= operator (or any other helper) in a way that's equivalent to the above expression. And in any case, if case statements exist for free for all enums, and don't need to be manually implemented.
Thus my question: is there any more concise/declarative/clean/idiomatic way to implement isBar than what I have above? Or, more directly, is there any way to express if case statements as Swift expressions?
UPDATE 2:
Another workaround... Create a var that returns an Int ONLY based on the case, then use a static (or instance, I thought static looked cleaner) method to test equivalence of just the case. It won't clash with Equatable, you don't have to overload an operator (unless you want to replace the static method with one), and you also wouldn't have to create separate var isFoo, var isBar, etc.
I know you used this example to ask a more generic question (how can I use 'if case' as an expression?) but if that's not possible, this may be a valid workaround. I apologize if this treats "the symptoms" not "the problem"
enum Something{
case Foo(Int)
case Bar(Int)
static func sameCase(a: Something, b: Something) -> Bool {
return a.caseValue == b.caseValue
}
var caseValue: Int {
switch self {
case .Foo(_):
return 0
case .Bar(_):
return 1
}
}
//if necessary
var isBar: Bool {
return Something.sameCase(self, b: Something.Bar(0))
}
}
Something.sameCase(.Bar(0), b: .Foo(0)) // false
Something.sameCase(.Bar(1), b: .Foo(2)) // false
Something.sameCase(.Foo(0), b: .Foo(0)) // true
Something.sameCase(.Bar(1), b: .Bar(2)) // true
Something.Bar(0).isBar // true
Something.Bar(5).isBar // true
Something.Foo(5).isBar // false
UPDATE 1:
Ok, so this seems to work. If you overload the == operator to ignore values and return true only when both enums are the same case, you can pass any value in your isFoo method and still determine the type.
I'm assuming you will need to customize this function to accommodate the the associated values, but it seems like a step in the right direction
enum Something {
case Foo(Int)
case Bar(Int)
var isFoo: Bool {
return self == .Foo(0) // number doesn't matter here... see below
}
}
func ==(a: Something, b: Something) -> Bool {
switch (a,b) {
case (.Bar(_), .Bar(_)):
return true
case (.Foo(_), .Foo(_)):
return true
default:
return false
}
}
let oneFoo = Something.Foo(1)
let twoFoo = Something.Foo(2)
let oneBar = Something.Bar(1)
let twoBar = Something.Bar(2)
oneFoo == twoFoo // true
oneFoo == oneFoo // true
oneFoo == oneBar // false
oneFoo == twoBar // false
OLD:
You can use self and the case name to directly check which case it is, you don't have to use the case keyword. Hopefully this will work for your situation:
enum Something{
case Foo(Int)
case Bar(Int)
var isFoo: Bool {
switch self {
case Foo:
return true
case Bar:
return false
}
}
}
So, there is a neater way, but requires a 3rd-party package: CasePaths
The idea is they work similarly to KeyPaths, and they come with a / operator to trigger it. There is also a ~= operator to check if a CasePath matches an instance.
So, you can achieve something like your original example like so:
import CasePaths
enum Type {
case Foo(Int)
case Bar(Int)
var isBar: Bool {
/Self.Bar ~= self
}
}
You can also get the value:
extension Type {
/// Returns the `Int` if this is a `Bar`, otherwise `nil`.
var barValue: Int? {
(/Self.Bar).extract(from: self)
}
}
You can do several other useful things with CasePaths as well, such as extracting the Foo values in an array of Type values:
let values: [Type] = [.Foo(1), .Bar(2), .Foo(3), .Foo(4), .Bar(5)]
let foos = values.compactMap(/Type.Foo) // [1, 3, 4]
let bars = values.compactMap(/Type.Bar) // [2, 5]
I'm sure there is somewhat of a performance cost, but it may not be an issue in your context.
I have a similar wondering, and I kept searching for some work arounds about this, and landed on this page. I came up with code like this to compromise.
fileprivate enum TypePrimitive {
case foo
case bar
}
enum Type {
case foo(Int)
case bar(Int)
fileprivate var primitiveType: TypePrimitive {
switch self {
case .foo(_): return .foo
case .bar(_): return .bar
}
}
var isFoo: Bool { self.primitiveType == .foo }
var isBar: Bool { self.primitiveType == .bar }
}
I hope Apple will provide better solution by adding some features in Swift language.
Are you looking for the ? operator ?
documentation is here under the Ternary Conditional Operator title.
I want to associate two raw values to an enum instance (imagine an enum representing error types, I want Error.Teapot to have an Int type property code with value 418, and a String property set to I'm a teapot.)
Note the difference between raw values and associated values here—I want all Teapot instances to have a code of 418, I don't want a unique associated value for each Teapot instance.
Is there a better way than adding computed properties to the enum that switched on self to look up the appropriate value?
You have a couple options. But neither of them involve raw values. Raw values are just not the right tool for the task.
Option 1 (so-so): Associated Values
I personally highly recommend against there being more than one associated value per enum case. Associated values should be dead obvious (since they don't have arguments/names), and having more than one heavily muddies the water.
That said, it's something the language lets you do. This allows you to have each case defined differently as well, if that was something you needed. Example:
enum ErrorType {
case teapot(String, Int)
case skillet(UInt, [CGFloat])
}
Option 2 (better): Tuples! And computed properties!
Tuples are a great feature of Swift because they give you the power of creating ad-hoc types. That means you can define it in-line. Sweet!
If each of your error types are going to have a code and a description, then you could have a computed info property (hopefully with a better name?). See below:
enum ErrorType {
case teapot
case skillet
var info: (code: Int, description: String) {
switch self {
case .teapot:
return (418, "Hear me shout!")
case .skillet:
return (326, "I'm big and heavy.")
}
}
}
Calling this would be much easier because you could use tasty, tasty dot syntax:
let errorCode = myErrorType.info.code
No, an enum cannot have multiple raw values - it has to be a single value, implementing the Equatable protocol, and be literal-convertible as described in the documentation.
I think the best approach in your case is to use the error code as raw value, and a property backed by a prepopulated static dictionary with the error code as key and the text as value.
I created a way of simulating this (No different than what Marcos Crispino suggested on his answer). Far from a perfect solution but allows us to avoid those nasty switch cases for every different property we want to get.
The trick is to use a struct as the "properties/data" holder and using it as a RawValue in the enum itself.
It has a bit of duplication but it's serving me well so far. Every time you want to add a new enum case, the compiler will remind you to fill in the extra case in the rawValue getter, which should remind you to update the init? which would remind you to create the new static property on the struct.
Gist
Code to the Gist:
enum VehicleType : RawRepresentable {
struct Vehicle : Equatable {
let name: String
let wheels: Int
static func ==(l: Vehicle, r: Vehicle) -> Bool {
return l.name == r.name && l.wheels == r.wheels
}
static var bike: Vehicle {
return Vehicle(name: "Bicycle", wheels: 2)
}
static var car: Vehicle {
return Vehicle(name: "Automobile", wheels: 4)
}
static var bus: Vehicle {
return Vehicle(name: "Autobus", wheels: 8)
}
}
typealias RawValue = Vehicle
case car
case bus
case bike
var rawValue: RawValue {
switch self {
case .car:
return Vehicle.car
case .bike:
return Vehicle.bike
case .bus:
return Vehicle.bus
}
}
init?(rawValue: RawValue) {
switch rawValue {
case Vehicle.bike:
self = .bike
case Vehicle.car:
self = .car
case Vehicle.bus:
self = .bus
default: return nil
}
}
}
VehicleType.bike.rawValue.name
VehicleType.bike.rawValue.wheels
VehicleType.car.rawValue.wheels
VehicleType(rawValue: .bike)?.rawValue.name => "Bicycle"
VehicleType(rawValue: .bike)?.rawValue.wheels => 2
VehicleType(rawValue: .car)?.rawValue.name => "Automobile"
VehicleType(rawValue: .car)?.rawValue.wheels => 4
VehicleType(rawValue: .bus)?.rawValue.name => "Autobus"
VehicleType(rawValue: .bus)?.rawValue.wheels => 8
No, you cannot have multiple raw values associated with an enum.
In your case, you could have the raw value to be equal to the code, and have an associated value with the description. But I think the computed properties approach is the best option here.
One workaround if you wanted to have many static properties for a YourError could be to import a property list; you could set the root object to a dictionary, with your enum raw value as the key for each object, allowing you to easily retrieve static structured data for the object.
This has an example of importing and using a plist: http://www.spritekitlessons.com/parsing-a-property-list-using-swift/
That might be overkill for simply an error description, for which you could just use a hardcoded static function with a switch statement for your enum values, that returns the error string you need. Simply place the static function in the same .swift file as your enum.
For instance,
static func codeForError(error : YourErrorType) -> Int {
switch(error) {
case .Teapot:
return "I'm a Teapot"
case .Teacup:
return "I'm a Teacup"
...
default:
return "Unknown Teaware Error"
}
}
This has the benefit (compared to the .plist solution) of better accomodating localization. However, a .plist could just contain a key used for retrieving the proper localization, instead of a error string, for this purpose.
For beginning, assuming you want to store a code and a message, you can use a struct for RawValue
struct ErrorInfo {
let code: Int
let message: String
}
Next step is to define the enum as being RawRepresentable, and use ErrorInfo as the raw value:
enum MyError: RawRepresentable {
typealias RawValue = ErrorInfo
case teapot
What remains is to map between instances of MyError and ErrorInfo:
static private let mappings: [(ErrorInfo, MyError)] = [
(ErrorInfo(code: 418, message: "I'm a teapot"), .teapot)
]
With the above, let's build the full definition of the enum:
enum MyError: RawRepresentable {
static private let mappings: [(ErrorInfo, MyError)] = [
(ErrorInfo(code: 418, message: "I'm a teapot"), .teapot)
]
case teapot
init?(rawValue: ErrorInfo) {
guard let match = MyError.mappings.first(where: { $0.0.code == rawValue.code && $0.0.message == rawValue.message}) else {
return nil
}
self = match.1
}
var rawValue: ErrorInfo {
return MyError.mappings.first(where: { $0.1 == self })!.0
}
}
Some notes:
you could use only the error code for matching, however this might result in inconsistent raw values if the messages differ
the amount of boilerplate code required to have raw values of some custom type might not outcome the benefits of using associated values.
Possible work around may to associate custom functions with enum
enum ToolbarType : String{
case Case = "Case", View="View", Information="Information"
static let allValues = [Case, View, Information]
func ordinal() -> Int{
return ToolbarType.allValues.index(of: self)!
}
}
Can be used as
for item in ToolbarType.allValues {
print("\(item.rawValue): \(item.ordinal())")
}
Output
Case: 0
View: 1
Information: 2
Possibly you can have additional functions to associate enum type to different values
This doesn't particularly answer your question, which was asking to find a better way than switching through self to look up the appropriate value but this answer may still be useful for someone looking in the future that needs a simple way to get a string from an enum which is defined as an integer type.
enum Error: UInt {
case Teapot = 418
case Kettle = 419
static func errorMessage(code: UInt) -> String {
guard let error = Error(rawValue: code) else {
return "Unknown Error Code"
}
switch error {
case .Teapot:
return "I'm a teapot!"
case .Kettle:
return "I'm a kettle!"
}
}
}
This way, we can get the errorMessage two ways:
With an integer (eg. that was returned as an error code from a server)
With an enum value (the rawValue we define for the enum)
Option 1:
let option1 = Error.errorMessage(code: 418)
print(option1) //prints "I'm a teapot!"
Option 2:
let option2 = Error.errorMessage(code: Error.Teapot.rawValue)
print(option2) //prints "I'm a teapot!"
In modern versions of Swift it's possible to get the string value of an enum case label, even without that enum being declared with a : String rawValue.
How to get the name of enumeration value in Swift?
So there is no longer a need to define and maintain a convenience function that switches on each case to return a string literal. In addition, this works automatically for any enum, even if no raw-value type is specified.
This, at least, allows you to have "multiple raw values" by having both a real : Int rawValue as well as the string used as the case label.
I think it just tricky, and I have create my own idea like below:
enum Gender:NSNumber
{
case male = 1
case female = 0
init?(strValue: String?) {
switch strValue {
case Message.male.value:
self = .male
case Message.female.value:
self = .female
default: return nil
}
}
var strValue: String {
switch self {
case .male:
return Message.male.value
case .female:
return Message.female.value
}
}
}
First of all, enums should only have one raw value. However if you want to have something that can use multiple raw values... there is a way to 'hack' this, but you have to make it codable and hashable yourself, implement custom init's etc.
enum MyCustomEnum: Codable, Hashable {
// duplicate every case with associated value of Codable.Type
case myFirstCase, _myFirstCase(Codable.Type)
case mySecondCase, _mySecondCase(Codable.Type)
case myThirdCase, _myThirdCase(Codable.Type)
case unknown(Any), _unknown(Codable.Type, Any) // handles unknown values
// define an allCases value to determine the only values your app 'sees'.
static var allCases: [Self] {
return [
.myFirstCase,
.mySecondCase,
.myThirdCase
// unknown(String) // you can add unknown as well, but this is too mask any unknown values.
]
}
static func == (lhs: MyCustomEnum, rhs: MyCustomEnum) -> Bool {
return lhs.stringValue == rhs.stringValue // can be either one of your custom raw values.
}
// add this per raw value. In this case one for Int and one for String
init(rawValue: Int) {
guard let value = Self.allCases.first(where:{ $0.intValue == rawValue }) else {
self = ._unknown(Int.self, rawValue)
return
}
switch value {
case .myFirstCase: self = ._myFirstCase(Int.self)
case .mySecondCase: self = ._mySecondCase(Int.self)
case .myThirdCase: self = ._myThirdCase(Int.self)
default: self = ._unknown(Int.self, rawValue)
}
}
init(rawValue: String) {
guard let value = Self.allCases.first(where:{ $0.stringValue == rawValue }) else {
self = ._unknown(String.self, rawValue)
return
}
switch value {
case .myFirstCase: self = ._myFirstCase(String.self)
case .mySecondCase: self = ._mySecondCase(String.self)
case .myThirdCase: self = ._myThirdCase(String.self)
default: self = ._unknown(Int.self, rawValue)
}
}
// add this per raw value. In this case one for Int and one for String
var intValue: Int {
switch self {
case .myFirstCase, ._myFirstCase(_): return 1
case .mySecondCase, ._mySecondCase(_): return 2
case .myThirdCase, ._myThirdCase(_): return 3
case .unknown(let value), ._unknown(_, let value): return value as? Int ?? -1 // you can also choose to let intValue return optional Int.
}
}
var stringValue: String {
switch self {
case .myFirstCase, ._myFirstCase(_): return "my first case"
case .mySecondCase, ._mySecondCase(_): return "my second case"
case .myThirdCase, ._myThirdCase(_): return "my third case"
case .unknown(let value), ._unknown(_, let value): return value as? String ?? "not a String" // you can also choose to let stringValue return optional String.
}
}
// determine the codable type using Mirror
private func getCodableType() -> Codable.Type? {
let mirrorOfModuleType = Mirror.init(reflecting: self)
guard let childOfModuleType = mirrorOfModuleType.children.first else { // no children, means no associated values.
return nil
}
let value = childOfModuleType.value // can be either Codable.Type, String or (Codable.Type & String)
if let rawValue = value as? Codable.Type {
return rawValue
} else {
guard let rawValue = value as? (Codable.Type, String) else {
// unknown(String), we don't know the rawValue as given, but try in this part of the code to guess what type fits best.
if self.stringValue != "\(self.intValue)" { // e.g. "1" might match 1 but "1.0" and 1 don't match
return String.self
} else {
return Int.self // return either a default value, or nil. It's your choice.
}
}
return rawValue.0
}
}
// confine to hashable using getCodableType
func hash(into hasher: inout Hasher) {
if self.getCodableType() is String.Type {
hasher.combine(self.stringValue)
} else { // if you don't call hasher.combine at all, you can expect strange issues. If you do not know the type, choose one that is most common.
hasher.combine(self.intValue)
}
}
// confine to Decodable
init(from decoder: Decoder) throws {
if let rawValue = try? Int.init(from: decoder) {
self.init(rawValue: rawValue)
} else if let rawValue = try? String.init(from: decoder) {
self.init(rawValue: rawValue)
} else {
throw DecodingError.valueNotFound(Self.self, DecodingError.Context(codingPath: [], debugDescription: "no matching value was found"))
}
}
// confine to Encodable using getCodableType
func encode(to encoder: Encoder) throws {
let rawValue = self.getCodableType()
if rawValue is String.Type {
try self.stringValue.encode(to: encoder)
} else if rawValue is Int.Type {
try self.intValue.encode(to: encoder)
} else {
// getCodableType returns nil if it does not know what value it is. (e.g. myFirstCase without associated value) If you want to support this as well, you can encode using one of your rawValues to the encoder.
throw EncodingError.invalidValue(Self.self, EncodingError.Context.init(codingPath: [], debugDescription: "this enum does not have a correct value", underlyingError: nil))
}
}
}
this code is scalable to any number of raw value as long as they are Codable