Chat AI

The project integrates the ChatGPT model with GPT-4 and the Chat API, all while using SwiftUI for a sleek user interface.

Usage

We're going to get our endpoint for our chat API, which we can initialize as follows:

private let endpointURL = "https://api.openai.com/v1/chat/completions"

and you can get the API key from OpenAI.

The function sendMessage takes an array of Message objects as input and is marked with the async keyword, indicating that it can be used with asynchronous programming.

func sendMessage(messages: [Message]) async -> OpenAIChatResponse? {
        // Map input messages to OpenAIChatMessage format
        let openAIMessages = messages.map({OpenAIChatMessage(role: $0.role, content: $0.content)})

        // Create the request body with the chosen model and formatted messages
        // model gpt-3.5.-turbo or gpt-4
        let body = OpenAIChatBody(model: "gpt-3.5-turbo", messages: openAIMessages)

        // Set headers for the API request
        let headers: HTTPHeaders = [
            "Authorization": "Bearer \(Constants.OpenAIApiKey)"
        ]

        // Send the API request and decode the response into OpenAIChatResponse
        return try? await AF.request(endpointURL, method: .post, parameters: body, encoder: .json, headers: headers).serializingDecodable(OpenAIChatResponse.self).value
}

You'll notice inside the function that the input messages are transformed into an array of OpenAIChatMessage objects using the map function. Each OpenAIChatMessage has a role and content, which are taken from the corresponding properties of the input Message objects.

The response from the API request is then decoded into an instance of OpenAIChatResponse using the serializingDecodable function provided by the Alamofire library.

Learn how to use OpenAI in your project

Package

The package that we used:

• Alamofire: Networking framework