基于嵌套类型属性反序列化 JSON 数组

Question

考虑这个例子JSON：

{
"sections": [{
"title": "Sign up",
"rows": [
{
"type": "image",
"imageURL": "https://example.com/image.jpg"
},
{
"type": "textField",
"value": "",
"placeholder": "Username"
},
{
"type": "textField",
"placeholder": "password"
},
{
"type": "textField",
"placeholder": "confirmPassword"
},
{
"type": "button",
"placeholder": "Register!"
}
]
}]
}

假设我想将上面的 JSON 解析为以下模型(我知道由于Row协议不对应于Decodable，它无法编译(：

enum RowType: String, Codable {
case textField
case image
case button
}
protocol Row: Codable {
var type: RowType { get }
}
struct TextFieldRow: Row {
let type: RowType
let placeholder: String
let value: String
enum CodingKey: String {
case type
case placeholder
case value
}
}
struct ImageRow: Row {
let type: RowType
let imageURL: URL
enum CodingKey: String {
case type
case imageURL
}
}
struct ButtonRow: Row {
let type: RowType
let title: String
enum CodingKey: String {
case type
case title
}
}
struct Section: Codable {
let rows: [Row]
let title: String
enum CodingKey: String {
case rows
case title
}
}
struct Response: Codable {
let sections: [Section]
enum CodingKey: String {
case sections
}
}

// Parsing the response using the Foundation JSONDecoder
let data: Data // From network
let decoder = JSONDecoder()
do {
let response = try decoder.decode(Response.self, from: data)
} catch {
print("error: (error)")
}

有没有办法使上面的 Swift 代码符合 Codable 标准？ 我知道你可以手动解决这个问题，首先抓取每个Row的type字符串，然后创建正确类型的Row模型，并将它们从结构更改为类，让Row协议成为一个超类。但是有没有一种方法需要更少的体力劳动？

Answer 1

使用具有关联值的enum是最佳选择：

考虑一下这个枚举：

enum Row: Decodable {
case textField(TextFieldRow)
case image(ImageRow)
// and other cases
case unknown
enum CodingKeys: String, CodingKey {
case type
}
public init(from decoder: Decoder) throws {
do {
let selfContainer = try decoder.singleValueContainer()
let typeContainer = try decoder.container(keyedBy: CodingKeys.self)
let type = try typeContainer.decode(String.self, forKey: .type)
switch type {
case "textField": self = .textField( try selfContainer.decode(TextFieldRow.self) )
case "Image": self = .image( try selfContainer.decode(ImageRow.self) )
// and other cases
default: self = .unknown
}
}
}
}

通过这些更改：

struct TextFieldRow: Decodable {
let placeholder: String?
let value: String?
}
struct ImageRow: Decodable {
let imageURL: URL
}
// and so on

现在这将像魅力一样解码：

// Minmal testing JSON
let json = """
[
{
"type": "image",
"imageURL": "https://example.com/image.jpg"
},
{
"type": "textField",
"value": "",
"placeholder": "Username"
},
{
"type": "textField",
"placeholder": "password"
}
]
""".data(using: .utf8)!

let decoder = JSONDecoder()
print( try! decoder.decode([Row].self, from: json) )

现在，您可以将所需的任何其他案例添加到解码器中，以构建应用程序构建器应用程序。