Looking for swift Answers? Try Ask4KnowledgeBase
Looking for swift Keywords? Try Ask4Keywords

Swift Language JSON Serialization, Encoding, and Decoding with Apple Foundation and the Swift Standard Library


Example

The JSONSerialization class is built into Apple's Foundation framework.

2.2

Read JSON

The JSONObjectWithData function takes NSData, and returns AnyObject. You can use as? to convert the result to your expected type.

do {
    guard let jsonData = "[\"Hello\", \"JSON\"]".dataUsingEncoding(NSUTF8StringEncoding) else {
        fatalError("couldn't encode string as UTF-8")
    }

    // Convert JSON from NSData to AnyObject
    let jsonObject = try NSJSONSerialization.JSONObjectWithData(jsonData, options: [])
    
    // Try to convert AnyObject to array of strings
    if let stringArray = jsonObject as? [String] {
        print("Got array of strings: \(stringArray.joinWithSeparator(", "))")
    }
} catch {
    print("error reading JSON: \(error)")
}

You can pass options: .AllowFragments instead of options: [] to allow reading JSON when the top-level object isn't an array or dictionary.

Write JSON

Calling dataWithJSONObject converts a JSON-compatible object (nested arrays or dictionaries with strings, numbers, and NSNull) to raw NSData encoded as UTF-8.

do {
    // Convert object to JSON as NSData
    let jsonData = try NSJSONSerialization.dataWithJSONObject(jsonObject, options: [])
    print("JSON data: \(jsonData)")
    
    // Convert NSData to String
    let jsonString = String(data: jsonData, encoding: NSUTF8StringEncoding)!
    print("JSON string: \(jsonString)")
} catch {
    print("error writing JSON: \(error)")
}

You can pass options: .PrettyPrinted instead of options: [] for pretty-printing.

3.0

Same behavior in Swift 3 but with a different syntax.

do {
    guard let jsonData = "[\"Hello\", \"JSON\"]".data(using: String.Encoding.utf8) else {
        fatalError("couldn't encode string as UTF-8")
    }
    
    // Convert JSON from NSData to AnyObject
    let jsonObject = try JSONSerialization.jsonObject(with: jsonData, options: [])
    
    // Try to convert AnyObject to array of strings
    if let stringArray = jsonObject as? [String] {
        print("Got array of strings: \(stringArray.joined(separator: ", "))")
    }
} catch {
    print("error reading JSON: \(error)")
}

do {
    // Convert object to JSON as NSData
    let jsonData = try JSONSerialization.data(withJSONObject: jsonObject, options: [])
    print("JSON data: \(jsonData)")

    // Convert NSData to String
    let jsonString = String(data: jsonData, encoding: .utf8)!
    print("JSON string: \(jsonString)")
} catch {
    print("error writing JSON: \(error)")
}

Note: The Following is currently available only in Swift 4.0 and later.

As of Swift 4.0, the Swift standard library includes the protocols Encodable and Decodable to define a standardized approach to data encoding and decoding. Adopting these protocols will allow implementations of the Encoder and Decoder protocols take your data and encode or decode it to and from an external representation such as JSON. Conformance to the Codable protocol combines both the Encodable and Decodable protocols. This is now the recommended means to handle JSON in your program.

Encode and Decode Automatically

The easiest way to make a type codable is to declare its properties as types that are already Codable. These types include standard library types such as String, Int, and Double; and Foundation types such as Date, Data, and URL. If a type's properties are codable, the type itself will automatically conform to Codable by simply declaring the conformance.

Consider the following example, in which the Book structure conforms to Codable.

struct Book: Codable {
    let title: String
    let authors: [String]
    let publicationDate: Date
}

Note that standard collections such as Array and Dictionary conform to Codable if they contain codable types.

By adopting Codable, the Book structure can now be encoded to and decoded from JSON using the Apple Foundation classes JSONEncoder and JSONDecoder, even though Book itself contains no code to specifically handle JSON. Custom encoders and decoders can be written, as well, by conforming to the Encoder and Decoder protocols, respectively.

Encode to JSON data

// Create an instance of Book called book
let encoder = JSONEncoder()
let data = try! encoder.encode(book) // Do not use try! in production code
print(data)

Set encoder.outputFormatting = .prettyPrinted for easier reading. ## Decode from JSON data

Decode from JSON data

// Retrieve JSON string from some source
let jsonData = jsonString.data(encoding: .utf8)!
let decoder = JSONDecoder()
let book = try! decoder.decode(Book.self, for: jsonData) // Do not use try! in production code
print(book)

In the above example, Book.self informs the decoder of the type to which the JSON should be decoded.

Encoding or Decoding Exclusively

Sometimes you may not need data to be both encodable and decodable, such as when you need only read JSON data from an API, or if your program only submits JSON data to an API.

If you intend only to write JSON data, conform your type to Encodable.

struct Book: Encodable {
    let title: String
    let authors: [String]
    let publicationDate: Date
}

If you intend only to read JSON data, conform your type to Decodable.

struct Book: Decodable {
    let title: String
    let authors: [String]
    let publicationDate: Date
}

Using Custom Key Names

APIs frequently use naming conventions other than the Swift-standard camel case, such as snake case. This can become an issue when it comes to decoding JSON, since by default the JSON keys must align exactly with your type's property names. To handle these scenarios you can create custom keys for your type using the CodingKey protocol.

struct Book: Codable {
    // ...
    enum CodingKeys: String, CodingKey { 
        case title
        case authors
        case publicationDate = "publication_date"
    }
}

CodingKeys are generated automatically for types which adopt the Codable protocol, but by creating our own implementation in the example above we're allow our decoder to match the local camel case publicationDate with the snake case publication_date as it's delivered by the API.