Latest 0.0.14
License MIT
Platforms ios 8.0, osx 10.9, watchos 2.0, tvos 9.0

Code Climate
Issue Count
Travis CI

There is a plenty of JSON libraries for Swift already. Why this one?

  • It is using pure Swift and only Swift. It will work anywhere Swift works. Even Foundation framework is not necessary.
  • It enforces strict types conformance by wrapping each possible case of JSON in enum. Therefore, no more need to resort to obscure AnyObject.
  • It takes convenience and readability quite seriously. In particular, it provides subcript access to nested JSON sub-components, easy encoding and decoding of custom types, out-of-the-box serialization for the types that adopted encoding protocols, etc.

This library was inspired by SwiftyJSON, Gloss and TidyJSON.



The core of Iolcus framework is JSON type. In practice it’s just an enum:

enum JSON {
    case Null
    case Boolean(Bool)
    case Integer(Int)
    case Float(Double)
    case String(String )
    case Array([JSON])
    case Object([String: JSON])

JSON literals

It’s possible to create an instance of JSON by just assigning some property with a literal:

import Iolcus

let jsonBoolean: JSON = true
let jsonInteger: JSON = 42
let jsonFloat: JSON = 36.6
let jsonString: JSON = "Lorem ipsum dolor sit amet"
let jsonArray: JSON = [false, 1, 42.0, "3"]
let jsonObject: JSON = [
    "boolean" : false,
    "integer" : -42,
    "string"  : "Lorem ipsum"

JSON initializers

Some other ways ways to instantiate JSON are:

// Direcly via enum's case
let jsonLUAE = JSON.Integer(42)

// By encoding JSON from an JSONEncodable-conforming instance
let jsonNotFalse = JSON(encoding: true)

// By copying another JSON's sub-element
let jsonSentence: JSON = ["Lorem", "ipsum", "dolor", "sit", "amet"]
let jsonFirstWord = try! JSON(json: jsonSentence, at: 0)    // .String("Lorem")

let jsonPerson: JSON = ["name": "John Doe", "age": 42]
let jsonAge = try! JSON(json: jsonPerson, at: "age")        // .Integer(42)

// By deserializing string representation of JSON
let serializedFibonacci = "[1, 1, 2, 3, 5, 8]"
let jsonFibonacci = try! JSON(jsonSerialization: serializedFibonacci)

Inspecting JSON

Although, JSON is enum and as such can be used in standard Swift switch statement, or in if/case pattern, it might be eaiser to use the set of is... (isNull, isBoolean, isInteger, etc) computed properties to check which particular JSON kind is being handled at the moment.

For example:

let jsonStuff: JSON = [true, "Ook", 36.6, 999, "Eek"]       // Prints:
for (_, element) in jsonStuff where element.isNumber {      // Float 36.6
    print(element.kind, element)                            // Integer 999

Unwrapping and coercing JSON values

Every JSON value, except .Null, wraps either some elementary value (Bool, Int, Double or String) or a container ([JSON] or [String: JSON]). It is possible to access the value wrapped by JSON via special set of computed properties unwrapped... (unwrappedBoolean, unwrappedInteger, unwrappedFloat , etc). These properties return either a wrapped value (if the type matches) or nil (if there is no match).

For example:

let jsonEarth: JSON = [
    "isFlat" : false,
    "mass"   : JSON(encoding: ["value": 5.972e24, "uom": "kg"]),
    "radius" : JSON(encoding: ["value": 6371, "uom": "km"])

if let earthIsFlat = jsonEarth["isFlat"].unwrappedBoolean {         // Prints:
    if !earthIsFlat {                                               //
        print("Earth is not flat")                                  // Earth is not flat

if let earthMass = jsonEarth["mass"]["value"].unwrappedString {     // Prints nothing because
    print("Earth's mass is (earthMass)")                           // unwrappedString returns nil
}                                                                   // for JSON.Float values

In addition to unwrapped... set of properties there is another set of coerced... properties. This set provides implicit conversion from actually wrapped elementary value into a target type. If such conversion is not possible then the result is, again, nil.

For example:

if let earthMass = jsonEarth["mass"]["value"].coercedString {       // Prints:
    print("Earth's mass is (earthMass)")                           //
}                                                                   // Earth's mass is 5.972e+24


Elements of the container JSON kinds (that is, .Array and .Object) can be directly accessed via subscripts:

let jsonEarthIsFlat = jsonEarth["isFlat"]       // .Boolean(false)

Subscripts can be chained:

let jsonEarthMass = jsonEarth["mass"]["value"]  // .Float(5.972e+24)

Subscripts can be used as setters:

var jsonListOfOrders: JSON = [1000000, 1000001, 1000002]    // Prints:
jsonListOfOrders[2] = 1000005                               // [
jsonListOfOrders[1] = 1000003                               //     1000007,
jsonListOfOrders[0] = 1000007                               //     1000003,
                                                            //     1000005
print(jsonListOfOrders)                                     // ]

In addition to Int and String-indexed subscripts, there is a special case of Void-indexed subscript []:

var jsonDynamicallyGrownArray: JSON = []            // Prints:
jsonDynamicallyGrownArray[].append("start"  )       // [
jsonDynamicallyGrownArray[].append(".")             //     "start",
jsonDynamicallyGrownArray[].append("..")            //     ".",
jsonDynamicallyGrownArray[].append("...end")        //     "..",
                                                    //     "...end"
print(jsonDynamicallyGrownArray)                    // ]

The main purpose of [] subcript is to provide a way to create/grow/shrink JSON.Array:

  • When getter is applied to .Array, it unwraps and returns underlying array of JSON sub-elements.
  • When getter is applied to any other JSON kind, it will create and return single-element array out of it.
  • The setter will take input [JSON] array and create JSON.Array out of it.

For example, the following snippet will produce exactly same output as the one from above:

var jsonImplicitlyConvertedArray: JSON = "start"    // Prints:
jsonImplicitlyConvertedArray[].append("."  )        // [
jsonImplicitlyConvertedArray[].append("..")         //     "start",
jsonImplicitlyConvertedArray[].append("...end")     //     ".",
                                                    //     "..",
print(jsonImplicitlyConvertedArray)                 //     "...end"
                                                    // ]


Encoding of a custom type into JSON is enabled by implementing JSONEncodable protocol.

For example, if we have a structure called Book:

struct Book {
    let title       : String
    let pagesCount  : Int
    let isPaperback : Bool
    let authors     : [String]
    let notes       : [String: String]

… then it can implement JSONEncodable as follows:

extension Book: JSONEncodable {
    func jsonEncoded() -> JSON {
        return [
            "title"       : title,
            "pages"       : pagesCount,
            "isPaperback" : isPaperback,
            "authors"     : JSON(encoding: authors),
            "notes"       : JSON(encoding: notes)

Note: Array authors and dictionary notes have to be explicitly encoded into JSON via JSON(encoding: _) initializer. This is because there is no support (yet) for conditional protocol conformance by generics. It might appear in Swift 3, though.

Obvioulsy, with JSONEncodable implemented, we can encode Book instances into JSON by calling the method we just implemented:

let book = Book(                                            // Prints:
    title       : "Dune",                                   //
    pagesCount  : 896,                                      // {
    isPaperback : true,                                     //     "title": "Dune",
    authors     : ["Frank Herbert"],                        //     "isPaperback": true,
    notes       : [                                         //     "authors": [
        "2015-05-23": "Damaged. Handed over to repair.",    //         "Frank Herbert"
        "2015-06-10": "Repaired. Condition is good."        //     ],
    ]                                                       //     "notes": {
)                                                           //         "2015-05-23": "Damaged. Handed over to repair.",
                                                            //         "2015-06-10": "Repaired. Condition is good."
var jsonBook = book.jsonEncoded()                           //     },
                                                            //     "pages": 896
print(jsonBook)                                             // }

… or alternatively we can use the initializer:

let jsonAnotherBook = JSON(encoding: book)

As soon as Book is JSONEncodable we can also encode arrays and dictionaries that contain Book:

struct Library: JSONEncodable {
    let books: [Book]
    let favorites: [String: Book]
    let series: [String: [Book]]
    func jsonEncoded() -> JSON {
        return [
            "books"     : JSON(encoding: books),
            "favorites" : JSON(encoding: favorites),
            "series"    : JSON(encoding: series)


Decoding of JSON is also made possible via protocol implementation. In this case it’s JSONDecodable protocol.

Keeping up with our Book as an example:

extension Book: JSONDecodable {
    init(json: JSON) throws {
        title       = try json.decode(at: "title")
        pagesCount  = try json.decode(at: "pages")
        isPaperback = try json.decode(at: "isPaperback")
        authors     = try json.decode(at: "authors")
        notes       = try json.decode(at: "notes")

Same thing for Library:

extension Library: JSONDecodable {
    init(json: JSON) throws {
        books     = try json.decode(at: "books")
        favorites = try json.decode(at: "favorites")
        series    = try json.decode(at: "series")

As soon as JSONDecodable is implemented, decoding of a new instance from JSON takes just a call to the initializer:

let anotherBook = try! Book(json: jsonBook)

Serializing and deserializing

Method jsonSerialized() should be used both to serialize JSON values:

let serializedBook = jsonBook.jsonSerialized()

… as well as instances that conform to JSONEncodable protocol:

let anotherSerializedBook = book.jsonSerialized()

Special form of initializer JSON(jsonSerialization: _) can be used to deserialize JSON values:

Note: Theoretically, a string that is used as an input for deserialization can be invalid. That is why JSON(jsonSerialization: _) call can throw an error of JSON.Error.Deserializing type.

do {
    let jsonDeserialized = try JSON(jsonSerialization: serializedBook)
catch let error as JSON.Error.Deserializing {
    // Deserializing JSON failed.  Do something about it.

Similar form of initializer in available for types conforming to JSONDecodable protocol:

Note: Deserialization of Book can throw a more error types. On top of JSON.Error.Deserializing it can also throw JSON.Error.Decoding and JSON.Error.Subscripting:

do {
    let yetAnotherBook = try Book(jsonSerialization: serializedBook)
catch let error as JSON.Error.Deserializing {
    // Something went wrong during deserialization of JSON from input string
catch let error as JSON.Error.Decoding {
    // Something went wrong while decoding of JSON into target type
catch let error as JSON.Error.Subscripting {
    // Something went wrong while reading values from container JSON elements

Iterating over JSON

Container JSON kinds (.Object and .Array) can be iterated over. For example, with forEach() method:

let jsonTodoList: JSON = ["Groceries", "Pick up kids", "Dinner"]    // Prints:
jsonTodoList.forEach {                                              // [0] "Groceries"
    print($0, $1)                                                   // [1] "Pick up kids"
}                                                                   // [2] "Dinner"

Note: Non-container JSON kinds can be iterated over as well, only there will be just one iteration made.

Note: JSONIndex is enum with three cases: .This used for non-container JSON, .Index(Int) used for .Array and .Key(String) used for .Object.

Naturally, map(), filter() and all other methods applicable to sequences can be used with JSON too:

let pagerCountIndex = jsonBook.filter { $1 == 896 }     // Prints:
    .first?.index                                       //
print(pagerCountIndex)                                  // Optional(["pages"])

Iterating over JSON in a standard for-loop will only go over the surface without going deeper down into nested JSON sub-elements. For this reason JSON offers special method flatten() that constructs an array of (path: JSONPath, elemetaryValue: JSON) tuples. Such array will only contain basic JSON values (.Null, .Boolean, .Integer, .Float and .String), and each such value will be accompanied with a JSONPath that holds the indexes that have to be traversed from the root of self and all the way down to the particular basic element.

For example:

let dune = Book(                                            // Prints:
    title       : "Dune",                                   //
    pagesCount  : 896,                                      // {
    isPaperback : true,                                     //     "title": "Dune",
    authors     : ["Frank Herbert"],                        //     "isPaperback": true,
    notes       : [                                         //     "authors": [
        "2015-05-23": "Damaged. Handed over to repair.",    //         "Frank Herbert"
        "2015-06-10": "Repaired. Condition is good."        //     ],
    ]                                                       //     "notes": {
)                                                           //         "2015-05-23": "Damaged. Handed over to repair.",
                                                            //         "2015-06-10": "Repaired. Condition is good."
let jsonDune = book.jsonEncoded()                           //     },
                                                            //     "pages": 896
print(jsonDune)                                             // }
print()                                                     //
                                                            // jsonDune["title"] == "Dune"
jsonDune.flatten().forEach {                                // jsonDune["isPaperback"] == true
    (path: JSONPath, json: JSON) in                         // jsonDune["authors"][0] == "Frank Herbert"
                                                            // jsonDune["notes"]["2015-05-23"] == "Damaged. Handed over to repair."
    print("jsonDune(path) == (json)")                     // jsonDune["notes"]["2015-06-10"] == "Repaired. Condition is good."
}                                                           // jsonDune["pages"] == 896
                                                            // ["notes"]["2015-05-23"]

Note: JSONPath is essentially an array of JSONIndex values. It can be constructed from the sequence of JSONIndex values or from a literal. It can be used in the subsript accessor as well.

Latest podspec

    "name": "Iolcus",
    "version": "0.0.14",
    "summary": "Swift JSON library",
    "description": "Swift JSON library.nInspired by SwiftyJSON, Gloss and TidyJSON.",
    "homepage": "",
    "license": {
        "type": "MIT",
        "file": "LICENSE"
    "authors": {
        "Anton Bronnikov": "[email protected]"
    "platforms": {
        "ios": "8.0",
        "osx": "10.9",
        "watchos": "2.0",
        "tvos": "9.0"
    "source": {
        "git": "",
        "tag": "0.0.14"
    "source_files": "Sources/**/*.{swift,h}"

Pin It on Pinterest

Share This