Text Generation

import Kuzco
// Basic session
let session = try await KuzcoSession(model: .qwen3_4b)
// Session with custom configuration
let config = KuzcoConfiguration(
    temperature: 0.7,
    maxTokens: 2048,
    topP: 0.9
)
let session = try await KuzcoSession(model: .qwen3_8b, configuration: config)

import SwiftUI
import Kuzco
struct ChatView: View {
    @State private var messages: [ChatMessage] = []
    @State private var currentResponse = ""
    @State private var isStreaming = false
    @State private var session: KuzcoSession?
    var body: some View {
        VStack {
            ScrollView {
                ForEach(messages) { message in
                    MessageBubble(message: message)
                }
                if !currentResponse.isEmpty {
                    MessageBubble(
                        message: ChatMessage(role: .assistant, content: currentResponse)
                    )
                }
            }
            ChatInput { text in
                sendMessage(text)
            }
            .disabled(isStreaming)
        }
        .task {
            session = try? await KuzcoSession(model: .qwen3_4b)
        }
    }
    func sendMessage(_ text: String) {
        guard let session = session else { return }
        messages.append(ChatMessage(role: .user, content: text))
        isStreaming = true
        currentResponse = ""
        Task {
            do {
                for try await partial in session.streamResponse(to: text) {
                    currentResponse += partial.text
                }
                // Move completed response to messages
                messages.append(ChatMessage(role: .assistant, content: currentResponse))
                currentResponse = ""
            } catch {
                currentResponse = "Error: \(error.localizedDescription)"
            }
            isStreaming = false
        }
    }
}

for try await partial in session.streamResponse(to: prompt) {
    print(partial.text, terminator: "")
    // Token usage is available on each partial response
    if let usage = partial.tokenUsage {
        print("Tokens used: \(usage.totalTokens)")
    }
    // Check if generation is complete
    if partial.isComplete {
        print("\nGeneration finished!")
    }
}

// Simple one-shot
let response = try await session.oneShot("What is the speed of light?")
print(response.text)
// With token usage
print("Prompt tokens: \(response.tokenUsage?.promptTokens ?? 0)")
print("Completion tokens: \(response.tokenUsage?.completionTokens ?? 0)")

let session = try await KuzcoSession(model: .qwen3_4b)
// Set system prompt
session.setSystemPrompt("""
You are a helpful fitness coach. Provide exercise recommendations
based on the user's goals. Always emphasize proper form and safety.
Include warm-up suggestions when relevant.
""")
// The model will now respond as a fitness coach
let response = try await session.oneShot("I want to build upper body strength")
print(response.text)

let session = try await KuzcoSession(model: .qwen3_4b)
// First message
let r1 = try await session.oneShot("My name is Alex")
print(r1.text) // "Nice to meet you, Alex!"
// Follow-up - context is remembered
let r2 = try await session.oneShot("What's my name?")
print(r2.text) // "Your name is Alex."
// Access conversation history
let history = session.conversationHistory
for message in history {
    print("\(message.role): \(message.content)")
}
// Clear history to start fresh
session.clearHistory()

let session = try await KuzcoSession(model: .qwen3_4b)
// Add messages manually
session.addMessage(role: .system, content: "You are a pirate captain.")
session.addMessage(role: .user, content: "Where should we sail?")
session.addMessage(role: .assistant, content: "Arrr! To the Caribbean!")
session.addMessage(role: .user, content: "What treasure should we seek?")
// Generate response with full context
let response = try await session.generateResponse()
print(response.text) // Response in pirate voice with context

// Store the task reference
var generationTask: Task<Void, Never>?
func startGeneration() {
    generationTask = Task {
        do {
            for try await partial in session.streamResponse(to: prompt) {
                // Check for cancellation
                try Task.checkCancellation()
                response += partial.text
            }
        } catch is CancellationError {
            print("Generation cancelled")
        } catch {
            print("Error: \(error)")
        }
    }
}
func cancelGeneration() {
    generationTask?.cancel()
}