import 'dart:convert';
import 'package:http/http.dart' as http;

// --- AJOUTS NÉCESSAIRES ---
import '../domain/catalogs/filter_catalog.dart'; // 1. Importer le catalogue de filtres// --- MISE À JOUR DU CONTRAT (SIGNATURE DE LA MÉTHODE) ---

/// L'objet retourné par l'analyse de l'IA
typedef ImageAnalysisResult = ({String prompt, List<String> filterIds});

/// Définit un contrat pour tout service capable d'analyser une image.
abstract class ImageAnalysisService {
  /// Prend une image en base64 et retourne un prompt ET une liste d'ID de filtres.
  Future<ImageAnalysisResult> analyzeImage(String base64Image); // 2. Mettre à jour la signature
}

// --- MISE À JOUR DE L'IMPLÉMENTATION ---

/// L'implémentation Ollama de ce service.
class OllamaImageAnalysisService implements ImageAnalysisService {
  final String _apiUrl = 'http://192.168.20.200:11434/api/generate';
  final String _visionModel = 'llava:7b';

  @override
  Future<ImageAnalysisResult> analyzeImage(String base64Image) async {
    print("[OllamaImageAnalysisService] 🚀 Lancement de l'analyse et de la suggestion de filtres...");

    // 3. Préparer la liste des filtres pour l'IA
    final filterDescriptions = availableFilters.map((filter) {
      return "- ID: \"${filter.id}\"\n  Description: ${filter.description}";
    }).join('\n\n');

    // 4. Mettre à jour le prompt système pour demander un JSON
    final requestPrompt = '''
You are an expert image analyst and social media content director.
Analyze the provided image and perform two tasks:
1.  Generate a detailed, high-quality, descriptive prompt in English for an image generation model like Stable Diffusion. The prompt must focus on subject, style, lighting, and composition.
2.  Choose the 3 BEST filters to enhance this image from the list below.

Available filters:
$filterDescriptions

You MUST reply ONLY with a valid JSON object in the following format. Do not include any other text, markdown, or explanations.
{
  "prompt": "your detailed image prompt here...",
  "filters": ["ID_OF_BEST_FILTER_1", "ID_OF_SECOND_BEST_FILTER_2", "ID_OF_THIRD_BEST_FILTER_3"]
}
''';

    final requestBody = {
      'model': _visionModel,
      'prompt': requestPrompt,
      'images': [base64Image],
      'stream': false,
      // 5. Demander explicitement un retour en JSON
      'format': 'json',
    };

    try {
      final response = await http.post(
        Uri.parse(_apiUrl),
        headers: {'Content-Type': 'application/json'},
        body: jsonEncode(requestBody),
      ).timeout(const Duration(minutes: 2));

      if (response.statusCode == 200) {
        // 6. Parser la réponse JSON
        final responseBodyString = jsonDecode(response.body)['response'] as String;
        final responseJson = jsonDecode(responseBodyString) as Map<String, dynamic>;

        final prompt = responseJson['prompt'] as String? ?? 'No prompt generated.';
        final filterIds = (responseJson['filters'] as List<dynamic>? ?? []).cast<String>();

        print('[OllamaImageAnalysisService] ✅ Analyse terminée. Prompt: $prompt, Filtres: $filterIds');
        return (prompt: prompt, filterIds: filterIds);
      } else {
        throw Exception('Erreur Ollama (analyzeImage) ${response.statusCode}: ${response.body}');
      }
    } catch (e) {
      print('[OllamaImageAnalysisService] ❌ Exception : ${e.toString()}');
      rethrow;
    }
  }
}