mymuseum-visitapp/lib/Services/Glasses/engines/impl/gemini_tts_engine.dart

import 'dart:convert';
import 'dart:io';
import 'dart:typed_data';
import 'package:flutter/foundation.dart';
import 'package:http/http.dart' as http;
import 'package:just_audio/just_audio.dart';
import 'package:path_provider/path_provider.dart';
import 'package:mymuseum_visitapp/Services/Glasses/engines/tts_engine.dart';

/// TTS via Gemini 2.5 Flash — voix naturelle, style configurable via prompt.
///
/// Voix disponibles : Algieba, Iapetus, Aoede, Charon, Fenrir, Kore,
///                    Leda, Orus, Puck, Schedar, Sulafat, Umbriel...
///
/// Exemple :
///   GeminiTtsEngine(
///     apiKey: kGeminiApiKey,
///     voiceName: 'Algieba',
///     voicePrompt: 'Voix de guide de musée, ton chaleureux, rythme posé, '
///                  'comme un narrateur de documentaire culturel.',
///   )
class GeminiTtsEngine implements TtsEngine {
  final String apiKey;
  final String voiceName;
  final String voicePrompt;

  static const String _model = 'gemini-2.5-flash-preview-tts';
  static const String _baseUrl = 'https://generativelanguage.googleapis.com/v1beta';

  final AudioPlayer _player = AudioPlayer();
  bool _speaking = false;
  String? _lastTempPath;

  GeminiTtsEngine({
    required this.apiKey,
    this.voiceName = 'Algieba',
    this.voicePrompt = 'Voix de guide de musée, ton chaleureux et bienveillant, '
        'rythme posé et clair, comme un narrateur de documentaire culturel.',
  });

  @override
  bool get isSpeaking => _speaking;

  @override
  Future<void> speak(String text, {String languageCode = 'fr-FR'}) async {
    if (text.isEmpty) return;
    try {
      _speaking = true;
      final wavPath = await _synthesize(text, languageCode);
      if (wavPath == null) return;
      _lastTempPath = wavPath;
      await _player.setFilePath(wavPath);
      await _player.play();
    } catch (e) {
      debugPrint('[GeminiTtsEngine] speak error: $e');
    } finally {
      _speaking = false;
    }
  }

  @override
  Future<void> stop() async {
    await _player.stop();
    _speaking = false;
  }

  @override
  Future<void> replay() async {
    if (_lastTempPath == null) return;
    await _player.seek(Duration.zero);
    await _player.play();
  }

  Future<String?> _synthesize(String text, String languageCode) async {
    final body = jsonEncode({
      'contents': [
        {
          'parts': [{'text': text}]
        }
      ],
      'systemInstruction': {
        'parts': [{'text': '$voicePrompt\nLangue : $languageCode.'}]
      },
      'generationConfig': {
        'responseModalities': ['AUDIO'],
        'speechConfig': {
          'voiceConfig': {
            'prebuiltVoiceConfig': {'voiceName': voiceName}
          }
        }
      }
    });

    final response = await http.post(
      Uri.parse('$_baseUrl/models/$_model:generateContent?key=$apiKey'),
      headers: {'Content-Type': 'application/json'},
      body: body,
    ).timeout(const Duration(seconds: 15));

    if (response.statusCode != 200) {
      debugPrint('[GeminiTtsEngine] HTTP ${response.statusCode}: ${response.body}');
      return null;
    }

    final json = jsonDecode(response.body) as Map<String, dynamic>;
    final candidates = json['candidates'] as List?;
    if (candidates == null || candidates.isEmpty) return null;

    final parts = candidates[0]['content']['parts'] as List?;
    if (parts == null || parts.isEmpty) return null;

    final inlineData = parts[0]['inlineData'] as Map<String, dynamic>?;
    if (inlineData == null) return null;

    final pcmBase64 = inlineData['data'] as String?;
    if (pcmBase64 == null) return null;

    final pcmBytes = base64Decode(pcmBase64);
    return _pcmToWav(pcmBytes, sampleRate: 24000);
  }

  /// Convertit du PCM16 brut en fichier WAV lisible par just_audio.
  Future<String> _pcmToWav(Uint8List pcm, {int sampleRate = 24000}) async {
    const channels = 1;
    const bitsPerSample = 16;
    final byteRate = sampleRate * channels * bitsPerSample ~/ 8;
    final blockAlign = channels * bitsPerSample ~/ 8;
    final dataSize = pcm.length;
    final chunkSize = 36 + dataSize;

    final header = ByteData(44);
    // RIFF chunk
    header.setUint8(0, 0x52); header.setUint8(1, 0x49);
    header.setUint8(2, 0x46); header.setUint8(3, 0x46); // "RIFF"
    header.setUint32(4, chunkSize, Endian.little);
    header.setUint8(8, 0x57); header.setUint8(9, 0x41);
    header.setUint8(10, 0x56); header.setUint8(11, 0x45); // "WAVE"
    // fmt chunk
    header.setUint8(12, 0x66); header.setUint8(13, 0x6D);
    header.setUint8(14, 0x74); header.setUint8(15, 0x20); // "fmt "
    header.setUint32(16, 16, Endian.little); // subchunk size
    header.setUint16(20, 1, Endian.little);  // PCM format
    header.setUint16(22, channels, Endian.little);
    header.setUint32(24, sampleRate, Endian.little);
    header.setUint32(28, byteRate, Endian.little);
    header.setUint16(32, blockAlign, Endian.little);
    header.setUint16(34, bitsPerSample, Endian.little);
    // data chunk
    header.setUint8(36, 0x64); header.setUint8(37, 0x61);
    header.setUint8(38, 0x74); header.setUint8(39, 0x61); // "data"
    header.setUint32(40, dataSize, Endian.little);

    final wav = Uint8List(44 + dataSize)
      ..setAll(0, header.buffer.asUint8List())
      ..setAll(44, pcm);

    final dir = await getTemporaryDirectory();
    final path = '${dir.path}/gemini_tts_${DateTime.now().millisecondsSinceEpoch}.wav';
    await File(path).writeAsBytes(wav, flush: true);
    return path;
  }

  void dispose() => _player.dispose();
}