sleepy_agent_ios/SleepyAgent/Services/TtsService.swift

import Foundation
import AVFoundation

// MARK: - Errors

enum TtsServiceError: LocalizedError {
    case notAvailable
    case synthesisFailed(String)
    
    var errorDescription: String? {
        switch self {
        case .notAvailable:
            return "Text-to-speech is not available on this device"
        case .synthesisFailed(let reason):
            return "Speech synthesis failed: \(reason)"
        }
    }
}

// MARK: - State

enum TtsState: Equatable {
    case initializing
    case ready
    case speaking
    case error(String)
}

// MARK: - TTS Service

/// Text-to-Speech service using AVSpeechSynthesizer.
actor TtsService: NSObject {
    
    // MARK: - Properties
    
    private var synthesizer: AVSpeechSynthesizer?
    private var currentState: TtsState = .initializing
    private var completionContinuation: CheckedContinuation<Void, Never>?
    
    /// Current TTS state
    var state: TtsState { currentState }
    
    /// Whether the synthesizer is currently speaking
    var isSpeaking: Bool {
        synthesizer?.isSpeaking ?? false
    }
    
    /// Whether TTS is available on this device
    var isAvailable: Bool {
        AVSpeechSynthesisVoice.speechVoices().count > 0
    }
    
    // MARK: - Initialization
    
    override init() {
        super.init()
    }
    
    deinit {
        synthesizer?.stopSpeaking(at: .immediate)
    }
    
    // MARK: - Setup
    
    /// Initializes the TTS engine.
    /// - Returns: Async stream of state changes
    func initialize() async -> AsyncStream<TtsState> {
        AsyncStream { continuation in
            Task {
                await setupSynthesizer()
                continuation.yield(currentState)
                continuation.finish()
            }
        }
    }
    
    private func setupSynthesizer() {
        guard isAvailable else {
            currentState = .error("No voices available")
            return
        }
        
        synthesizer = AVSpeechSynthesizer()
        synthesizer?.delegate = self
        currentState = .ready
    }
    
    // MARK: - Speech Methods
    
    /// Speaks the given text.
    /// - Parameters:
    ///   - text: The text to speak
    ///   - language: Optional language code (e.g., "en-US"). Defaults to system language.
    /// - Throws: TtsServiceError if synthesis fails
    func speak(text: String, language: String? = nil) async throws {
        guard let synthesizer = synthesizer else {
            throw TtsServiceError.notAvailable
        }
        
        // Stop any current speech
        synthesizer.stopSpeaking(at: .immediate)
        
        let utterance = AVSpeechUtterance(string: text)
        
        // Configure voice
        if let language = language {
            utterance.voice = AVSpeechSynthesisVoice(language: language)
        } else {
            // Try system language, fallback to US English
            let systemLanguage = Locale.current.language.languageCode?.identifier ?? "en"
            let region = Locale.current.region?.identifier ?? "US"
            let locale = "\(systemLanguage)-\(region)"
            
            if let voice = AVSpeechSynthesisVoice(language: locale) {
                utterance.voice = voice
            } else if let voice = AVSpeechSynthesisVoice(language: "en-US") {
                utterance.voice = voice
            }
        }
        
        // Configure speech parameters
        utterance.rate = AVSpeechUtteranceDefaultSpeechRate
        utterance.pitchMultiplier = 1.0
        utterance.volume = 1.0
        
        // Speak and wait for completion
        await withCheckedContinuation { continuation in
            self.completionContinuation = continuation
            synthesizer.speak(utterance)
        }
    }
    
    /// Speaks the given text with completion callback.
    /// - Parameters:
    ///   - text: The text to speak
    ///   - language: Optional language code
    ///   - onComplete: Called when speech completes or fails
    func speak(text: String, language: String? = nil, onComplete: (() -> Void)? = nil) {
        Task {
            do {
                try await speak(text: text, language: language)
            } catch {
                print("TTS error: \(error)")
            }
            onComplete?()
        }
    }
    
    /// Stops the current speech immediately.
    func stop() {
        synthesizer?.stopSpeaking(at: .immediate)
        completionContinuation?.resume()
        completionContinuation = nil
        currentState = .ready
    }
    
    /// Stops the current speech at the end of the word.
    func stopAtEndOfWord() {
        synthesizer?.stopSpeaking(at: .word)
    }
    
    /// Shuts down the TTS engine and releases resources.
    func shutdown() {
        synthesizer?.stopSpeaking(at: .immediate)
        synthesizer?.delegate = nil
        synthesizer = nil
        completionContinuation = nil
        currentState = .initializing
    }
    
    /// Sets the speech rate (0.0 to 1.0, default is 0.5).
    func setRate(_ rate: Float) {
        // Applied per utterance
    }
    
    /// Sets the speech volume (0.0 to 1.0).
    func setVolume(_ volume: Float) {
        // Applied per utterance
    }
}

// MARK: - AVSpeechSynthesizerDelegate

extension TtsService: AVSpeechSynthesizerDelegate {
    nonisolated func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didStart utterance: AVSpeechUtterance) {
        Task {
            await updateState(.speaking)
        }
    }
    
    nonisolated func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didFinish utterance: AVSpeechUtterance) {
        Task {
            await completeSpeech()
        }
    }
    
    nonisolated func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didCancel utterance: AVSpeechUtterance) {
        Task {
            await completeSpeech()
        }
    }
    
    nonisolated func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didPause utterance: AVSpeechUtterance) {
        // Handle pause if needed
    }
    
    nonisolated func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didContinue utterance: AVSpeechUtterance) {
        Task {
            await updateState(.speaking)
        }
    }
    
    private func updateState(_ state: TtsState) {
        currentState = state
    }
    
    private func completeSpeech() {
        completionContinuation?.resume()
        completionContinuation = nil
        currentState = .ready
    }
}