New content format

Sources/OpenAI/Public/Models/ChatQuery.swift

@@ -7,6 +7,8 @@
 
 import Foundation
 
+
+
 // See more https://platform.openai.com/docs/guides/text-generation/json-mode
 public struct ResponseFormat: Codable, Equatable {
     public static let jsonObject = ResponseFormat(type: .jsonObject)
@@ -20,10 +22,89 @@ public struct ResponseFormat: Codable, Equatable {
     }
 }
 
+public struct ChatContent: Codable, Equatable {
+    let type: ChatContentType
+    let value: String
+
+    public enum ChatContentType: String, Codable {
+        case text
+        case imageUrl = "image_url"
+    }
+
+    public struct ImageUrl: Codable, Equatable {
+        let url: String
+
+        enum CodingKeys: CodingKey {
+            case url
+        }
+    }
+
+
+    enum CodingKeys: CodingKey {
+        case type
+        case value
+    }
+
+    public static func text(_ text: String) -> Self {
+        Self.init(text)
+    }
+
+    public static func imageUrl(_ url: String) -> Self {
+        Self.init(type: .imageUrl, value: url)
+    }
+
+    public init(type: ChatContentType, value: String) {
+        self.type = type
+        self.value = value
+    }
+
+    public init(_ text: String) {
+        self.type = .text
+        self.value = text
+    }
+
+    // we need to perform a custom encoding since the `value` key is variable based on the `type`
+    public func encode(to encoder: Encoder) throws {
+        var container = encoder.container(keyedBy: ChatContent.CodingKeys.self)
+        var dynamicContainer = encoder.container(keyedBy: DynamicKey.self)
+
+        try container.encode(type, forKey: .type)
+
+        switch self.type {
+        case .text:
+            try dynamicContainer.encode(value, forKey: .init(stringValue: "text"))
+            break
+        case .imageUrl:
+            var nested = dynamicContainer.nestedContainer(keyedBy: ImageUrl.CodingKeys.self, forKey: .init(stringValue: "image_url"))
+            try nested.encode(value, forKey: .url)
+            break
+        }
+    }
+
+    public init(from decoder: Decoder) throws {
+        let container = try decoder.container(keyedBy: CodingKeys.self)
+        self.type = try container.decode(ChatContentType.self, forKey: .type)
+
+        let dynamicContainer = try decoder.container(keyedBy: DynamicKey.self)
+
+        switch self.type {
+        case .text:
+            self.value = try dynamicContainer.decode(String.self, forKey: .init(stringValue: "text"))
+            break
+        case .imageUrl:
+            let nested = try dynamicContainer.nestedContainer(keyedBy: ImageUrl.CodingKeys.self, forKey: .init(stringValue: "image_url"))
+            self.value = try nested.decode(String.self, forKey: .url)
+            break
+        }
+    }
+}
+
+
 public struct Chat: Codable, Equatable {
     public let role: Role
     /// The contents of the message. `content` is required for all messages except assistant messages with function calls.
-    public let content: String?
+    public let content: StringOrCodable<[ChatContent]>?
+
     /// The name of the author of this message. `name` is required if role is `function`, and it should be the name of the function whose response is in the `content`. May contain a-z, A-Z, 0-9, and underscores, with a maximum length of 64 characters.
     public let name: String?
     public let functionCall: ChatFunctionCall?
@@ -42,9 +123,33 @@ public struct Chat: Codable, Equatable {
         case functionCall = "function_call"
     }
 
-    public init(role: Role, content: String? = nil, name: String? = nil, functionCall: ChatFunctionCall? = nil) {
+    public init(role: Role, content stringContent: String? = nil, name: String? = nil, functionCall: ChatFunctionCall? = nil) {
+        let stringOrCodable: StringOrCodable<[ChatContent]>?;
+
+        if let string = stringContent {
+            stringOrCodable = .string(string)
+        } else {
+            stringOrCodable = nil
+        }
+
+        self.init(role: role, contents: stringOrCodable, name: name, functionCall: functionCall)
+    }
+
+    public init(role: Role, content arr: [ChatContent]? = nil ,name: String? = nil, functionCall: ChatFunctionCall? = nil) {
+        let stringOrCodable: StringOrCodable<[ChatContent]>?
+
+        if let arr = arr {
+            stringOrCodable = .object(arr)
+        } else {
+            stringOrCodable = nil
+        }
+
+        self.init(role: role, contents: stringOrCodable, name: name, functionCall: functionCall)
+    }
+
+    public init(role: Role, contents: StringOrCodable<[ChatContent]>? = nil, name: String? = nil, functionCall: ChatFunctionCall? = nil) {
         self.role = role
-        self.content = content
+        self.content = contents
         self.name = name
         self.functionCall = functionCall
     }

Sources/OpenAI/Public/Utilities/CodableUtilities.swift

@@ -0,0 +1,118 @@
+//
+//  CodableUtilities.swift
+//
+//
+//  Created by Federico Vitale on 09/11/23.
+//
+
+import Foundation
+
+
+/// Allows having dynamic keys in codables. 
+struct DynamicKey: CodingKey {
+    var stringValue: String
+    var intValue: Int?
+
+    init(stringValue: String) {
+        self.stringValue = stringValue
+    }
+
+    init?(intValue: Int) {
+        self.intValue = intValue
+        self.stringValue = "\(intValue)"
+    }
+}
+
+
+///  Allows to encode/decode ``Chat`` or ``Codable`` (T)
+///     ```swift
+///     struct Person: Codable, Equatable {
+///         let name: StringOrCodable<FullName>
+///
+///         struct FullName {
+///             firstName: String
+///             lastName: String
+///         }
+///     }
+///
+///     let person = Person(name: .string("John Doe"))
+///     let fullNamePerson = Person(name: .object(.init(firstName: "John", lastName: "Doe")))
+///     ```
+public enum StringOrCodable<T: Codable>: Equatable, Codable where T: Equatable {
+    case string(String)
+    case object(T)
+
+    enum CodingKeys: CodingKey {
+        case string
+        case object
+    }
+
+    public func encode(to encoder: Encoder) throws {
+        var container = encoder.singleValueContainer()
+
+        switch self {
+        case .string(let string):
+            try container.encode(string)
+        case .object(let object):
+            try container.encode(object)
+        }
+    }
+
+
+    public init(from decoder: Decoder) throws {
+        let container = try decoder.singleValueContainer()
+
+        if let string = try? container.decode(String.self) {
+            self = .string(string)
+        } else if let object = try? container.decode(T.self) {
+            self = .object(object)
+        } else {
+            throw DecodingError.dataCorrupted(
+                DecodingError.Context(
+                    codingPath: decoder.codingPath,
+                    debugDescription: "Invalid data encountered when decoding StringOrCodable<T>"
+                )
+            )
+        }
+    }
+}
+
+/// Same as ``StringOrCodable`` but accepts 2 codable generics instead of String as first generic argument
+public enum AnyOf<T: Codable, U: Codable>: Equatable, Codable where T: Equatable, U: Equatable {
+    case objectA(T)
+    case objectB(U)
+
+    enum CodingKeys: CodingKey {
+        case objectA
+        case objectB
+    }
+
+    public func encode(to encoder: Encoder) throws {
+        var container = encoder.singleValueContainer()
+
+        switch self {
+        case .objectB(let value):
+            try container.encode(value)
+        case .objectA(let value):
+            try container.encode(value)
+        }
+    }
+
+
+    public init(from decoder: Decoder) throws {
+        let container = try decoder.singleValueContainer()
+
+        if let valueT = try? container.decode(T.self) {
+            self = .objectA(valueT)
+        } else if let valueU = try? container.decode(U.self) {
+            self = .objectB(valueU)
+        } else {
+            throw DecodingError.dataCorrupted(
+                DecodingError.Context(
+                    codingPath: decoder.codingPath,
+                    debugDescription: "Invalid data encountered when decoding StringOrCodable<T>"
+                )
+            )
+        }
+    }
+}

Tests/OpenAITests/CodableUtilsTests.swift

@@ -0,0 +1,84 @@
+//
+//  File.swift
+//  
+//
+//  Created by Federico Vitale on 09/11/23.
+//
+
+import XCTest
+@testable import OpenAI
+
+fileprivate class TestUtils {
+    static func decode<T: Decodable & Equatable>(
+        _ jsonString: String,
+        _ expectedValue: T
+    ) throws {
+        let data = jsonString.data(using: .utf8)!
+        let decoded = try JSONDecoder().decode(T.self, from: data)
+
+        XCTAssertEqual(decoded, expectedValue)
+    }
+
+    static func encode<T: Encodable & Equatable>(
+        _ value: T,
+        _ expectedJson: String
+    ) throws {
+        let source = try jsonDataAsNSDictionary(JSONEncoder().encode(value))
+        let expected = try jsonDataAsNSDictionary(expectedJson.data(using: .utf8)!)
+
+        XCTAssertEqual(source, expected)
+    }
+
+    static func jsonDataAsNSDictionary(_ data: Data) throws -> NSDictionary {
+        return NSDictionary(dictionary: try JSONSerialization.jsonObject(with: data, options: []) as! [String: Any])
+    }
+}
+
+
+class CodableUtilsTests: XCTestCase {
+    func testStringOrCodable_String() throws {
+        struct Person: Codable, Equatable {
+            let name: StringOrCodable<FullName>
+
+            struct FullName: Codable, Equatable {
+                let firstName: String
+                let lastName: String
+            }
+        }
+
+        let jsonString = """
+        {
+            "name": "test"
+        }
+        """
+
+        let value = Person(name: .string("test"))
+
+        try TestUtils.encode(value, jsonString)
+        try TestUtils.decode(jsonString, value)
+    }
+
+    func testStringOrCodable_Object() throws {
+        struct Person: Codable, Equatable {
+            let name: StringOrCodable<FullName>
+
+            struct FullName: Codable, Equatable {
+                let firstName: String
+                let lastName: String
+            }
+        }
+
+        let jsonString = """
+        {
+            "name": { "firstName": "first", "lastName": "last" }
+        }
+        """
+
+        let value = Person(name: .object(.init(firstName: "first", lastName: "last")))
+
+        try TestUtils.encode(value, jsonString)
+        try TestUtils.decode(jsonString, value)
+    }
+}
+
+

Tests/OpenAITests/OpenAITests.swift

@@ -63,19 +63,20 @@ class OpenAITests: XCTestCase {
     }
 
     func testChats() async throws {
-       let query = ChatQuery(model: .gpt4, messages: [
-           .init(role: .system, content: "You are Librarian-GPT. You know everything about the books."),
-           .init(role: .user, content: "Who wrote Harry Potter?")
-       ])
-       let chatResult = ChatResult(id: "id-12312", object: "foo", created: 100, model: .gpt3_5Turbo, choices: [
-        .init(index: 0, message: .init(role: .system, content: "bar"), finishReason: "baz"),
-        .init(index: 0, message: .init(role: .user, content: "bar1"), finishReason: "baz1"),
-        .init(index: 0, message: .init(role: .assistant, content: "bar2"), finishReason: "baz2")
+        let query = ChatQuery(model: .gpt4, messages: [
+            .init(role: .system, content: "You are Librarian-GPT. You know everything about the books."),
+            .init(role: .user, content: "Who wrote Harry Potter?")
+        ])
+        let chatResult = ChatResult(id: "id-12312", object: "foo", created: 100, model: .gpt3_5Turbo, choices: [
+            .init(index: 0, message: .init(role: .system, content: "bar"), finishReason: "baz"),
+            .init(index: 0, message: .init(role: .user, content: "bar1"), finishReason: "baz1"),
+            .init(index: 0, message: .init(role: .assistant, content: "bar2"), finishReason: "baz2")
         ], usage: .init(promptTokens: 100, completionTokens: 200, totalTokens: 300))
-       try self.stub(result: chatResult)
+        try self.stub(result: chatResult)
+
+        let result = try await openAI.chats(query: query)
 
-       let result = try await openAI.chats(query: query)
-       XCTAssertEqual(result, chatResult)
+        XCTAssertEqual(result, chatResult)
     }
 
     func testChatsFunction() async throws {