# ==========================================
# game_engine.py - Calcul OCR v3.0 CLEAN
# ==========================================

"""
Moteur de jeu mathématique avec traitement parallèle et auto-détection OCR
"""

import random
import time
import datetime
import gradio as gr
import os
import uuid
import gc
import base64
from io import BytesIO
import numpy as np
from PIL import Image
import threading
import queue
from typing import Dict, Tuple, Optional

# Auto-détection propre : GPU OU CPU uniquement
ocr_module = None
ocr_info = {"model_name": "Unknown", "device": "Unknown"}

try:
    # Test GPU : torch + CUDA disponible
    import torch
    if torch.cuda.is_available():
        from image_processing_gpu import (
            recognize_number_fast_with_image, 
            create_thumbnail_fast, 
            create_white_canvas,
            cleanup_memory,
            log_memory_usage,
            get_ocr_model_info
        )
        ocr_module = "gpu"
        print("✅ Game Engine: Mode GPU - TrOCR activé")
    else:
        # Torch installé mais pas de GPU → CPU
        from image_processing_cpu import (
            recognize_number_fast_with_image, 
            create_thumbnail_fast, 
            create_white_canvas,
            cleanup_memory,
            log_memory_usage,
            get_ocr_model_info
        )
        ocr_module = "cpu"
        print("✅ Game Engine: Mode CPU - EasyOCR activé")
        
except ImportError:
    # Torch pas installé → CPU obligatoire
    from image_processing_cpu import (
        recognize_number_fast_with_image, 
        create_thumbnail_fast, 
        create_white_canvas,
        cleanup_memory,
        log_memory_usage,
        get_ocr_model_info
    )
    ocr_module = "cpu"
    print("✅ Game Engine: Mode CPU - EasyOCR activé")

# Récupérer les infos du modèle sélectionné
try:
    ocr_info = get_ocr_model_info()
    print(f"🎯 OCR sélectionné: {ocr_info['model_name']} sur {ocr_info['device']}")
except Exception as e:
    print(f"⚠️ Impossible de récupérer les infos OCR: {e}")
    ocr_info = {"model_name": "Error", "device": "Unknown"}

# Imports dataset avec gestion d'erreur
try:
    from datasets import Dataset, load_dataset
    DATASET_AVAILABLE = True
    print("✅ Modules dataset disponibles")
except ImportError as e:
    DATASET_AVAILABLE = False
    print(f"⚠️ Modules dataset non disponibles: {e}")

# Nom du dataset cohérent avec le space
DATASET_NAME = "hoololi/calcul_ocr_dataset"

# Configuration des difficultés par opération
DIFFICULTY_RANGES = {
    "×": {"Facile": (2, 9), "Difficile": (4, 12)},
    "+": {"Facile": (1, 50), "Difficile": (10, 100)},
    "-": {"Facile": (1, 50), "Difficile": (10, 100)},
    "÷": {"Facile": (1, 10), "Difficile": (2, 12)}
}

def create_result_row_with_images(i: int, image: dict | np.ndarray | Image.Image, expected: int, operation_data: tuple[int, int, str, int]) -> dict:
    
    # OCR optimisé
    recognized, optimized_image, dataset_image_data = recognize_number_fast_with_image(image)
    
    try:
        recognized_num = int(recognized) if recognized.isdigit() else 0
    except:
        recognized_num = 0

    is_correct = recognized_num == expected
    a, b, operation, correct_result = operation_data

    status_icon = "✅" if is_correct else "❌"
    status_text = "Correct" if is_correct else "Incorrect"
    row_color = "#e8f5e8" if is_correct else "#ffe8e8"
    
    # Miniature
    image_thumbnail = create_thumbnail_fast(optimized_image, size=(50, 50))
    
    # Libérer mémoire
    if optimized_image and hasattr(optimized_image, 'close'):
        try:
            optimized_image.close()
        except:
            pass

    return {
        'html_row': f"""
            <tr style="background-color: {row_color};">
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd; color: #333;">{i+1}</td>
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd; font-weight: bold; color: #333;">{a}</td>
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd; font-weight: bold; color: #333;">{operation}</td>
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd; font-weight: bold; color: #333;">{b}</td>
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd; font-weight: bold; color: #333;">{expected}</td>
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd;">{image_thumbnail}</td>
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd; font-weight: bold; color: #333;">{recognized_num}</td>
                <td style="text-align: center; padding: 8px; border: 1px solid #ddd; color: #333;">{status_icon} {status_text}</td>
            </tr>
        """,
        'is_correct': is_correct,
        'recognized': recognized,
        'recognized_num': recognized_num,
        'dataset_image_data': dataset_image_data
    }


class MathGame:
    """Moteur de jeu mathématique avec traitement parallèle"""
    
    def __init__(self):
        self.is_running = False
        self.start_time = 0
        self.current_operation = ""
        self.correct_answer = 0
        self.user_images = []
        self.expected_answers = []
        self.operations_history = []
        self.question_count = 0
        self.time_remaining = 30
        self.session_data = []
        
        # Configuration session
        self.duration = 30
        self.operation_type = "×"
        self.difficulty = "Facile"
        
        # Gestion export
        self.export_status = "not_exported"
        self.export_timestamp = None
        self.export_result = None
        
        # Traitement parallèle
        self.processing_queue = queue.Queue()
        self.results_cache: Dict[int, dict] = {}
        self.worker_thread: Optional[threading.Thread] = None
        self.processing_active = False

    def get_export_status(self) -> dict[str, str | bool | None]:
        return {
            "status": self.export_status,
            "timestamp": self.export_timestamp,
            "result": self.export_result,
            "can_export": self.export_status == "not_exported" and len(self.session_data) > 0
        }

    def mark_export_in_progress(self) -> None:
        self.export_status = "exporting"
        self.export_timestamp = datetime.datetime.now().isoformat()

    def mark_export_completed(self, result: str) -> None:
        self.export_status = "exported"
        self.export_result = result

    def _start_background_processing(self) -> None:
        """Démarre le thread de traitement en arrière-plan"""
        if self.worker_thread is None or not self.worker_thread.is_alive():
            self.processing_active = True
            self.worker_thread = threading.Thread(target=self._process_images_worker, daemon=True)
            self.worker_thread.start()
            print("🔄 Thread de traitement parallèle démarré")

    def _stop_background_processing(self) -> None:
        """Arrête le thread de traitement"""
        self.processing_active = False
        if self.worker_thread and self.worker_thread.is_alive():
            print("⏹️ Arrêt du thread de traitement parallèle")

    def _process_images_worker(self) -> None:
        """Worker thread qui traite les images en arrière-plan"""
        print("🚀 Worker thread démarré")
        while self.processing_active:
            try:
                if not self.processing_queue.empty():
                    question_num, image, expected, operation_data = self.processing_queue.get(timeout=1)
                    print(f"🔄 Traitement parallèle image {question_num}...")
                    
                    start_time = time.time()
                    result_data = create_result_row_with_images(question_num, image, expected, operation_data)
                    processing_time = time.time() - start_time
                    
                    # Stocker le résultat
                    self.results_cache[question_num] = result_data
                    print(f"✅ Image {question_num} traitée en {processing_time:.1f}s (parallèle)")
                    
                else:
                    time.sleep(0.1)
                    
            except queue.Empty:
                continue
            except Exception as e:
                print(f"❌ Erreur traitement parallèle: {e}")
                
        print("🛑 Worker thread terminé")

    def _add_image_to_processing_queue(self, question_num: int, image: dict | np.ndarray | Image.Image, 
                                     expected: int, operation_data: tuple) -> None:
        """Ajoute une image à la queue de traitement"""
        if image is not None:
            self.processing_queue.put((question_num, image, expected, operation_data))
            print(f"📝 Image {question_num} ajoutée à la queue de traitement")

    def generate_multiplication(self, difficulty: str) -> tuple[str, int]:
        """Génère une multiplication"""
        min_val, max_val = DIFFICULTY_RANGES["×"][difficulty]
        a = random.randint(min_val, max_val)
        b = random.randint(min_val, max_val)
        return f"{a} × {b}", a * b

    def generate_addition(self, difficulty: str) -> tuple[str, int]:
        """Génère une addition"""
        min_val, max_val = DIFFICULTY_RANGES["+"][difficulty]
        a = random.randint(min_val, max_val)
        b = random.randint(min_val, max_val)
        return f"{a} + {b}", a + b

    def generate_subtraction(self, difficulty: str) -> tuple[str, int]:
        """Génère une soustraction (résultat toujours positif)"""
        min_val, max_val = DIFFICULTY_RANGES["-"][difficulty]
        a = random.randint(min_val, max_val)
        b = random.randint(min_val, a)
        return f"{a} - {b}", a - b

    def generate_division(self, difficulty: str) -> tuple[str, int]:
        """Génère une division exacte"""
        min_result, max_result = DIFFICULTY_RANGES["÷"][difficulty]
        result = random.randint(min_result, max_result)
        if difficulty == "Facile":
            divisor = random.randint(2, 9)
        else:
            divisor = random.randint(2, 12)
        dividend = result * divisor
        return f"{dividend} ÷ {divisor}", result

    def generate_operation(self, operation_type: str, difficulty: str) -> tuple[str, int]:
        """Génère une opération selon le type et la difficulté"""
        if operation_type == "×":
            return self.generate_multiplication(difficulty)
        elif operation_type == "+":
            return self.generate_addition(difficulty)
        elif operation_type == "-":
            return self.generate_subtraction(difficulty)
        elif operation_type == "÷":
            return self.generate_division(difficulty)
        elif operation_type == "Aléatoire":
            random_op = random.choice(["×", "+", "-", "÷"])
            return self.generate_operation(random_op, difficulty)
        else:
            return self.generate_multiplication(difficulty)

    def start_game(self, duration: str, operation: str, difficulty: str) -> tuple[str, Image.Image, str, str, gr.update, gr.update, str]:
        """Démarre le jeu avec la configuration choisie"""
        
        # Configuration
        self.duration = 60 if duration == "60 secondes" else 30
        self.operation_type = operation
        self.difficulty = difficulty
        
        # Nettoyage
        if hasattr(self, 'user_images') and self.user_images:
            for img in self.user_images:
                if hasattr(img, 'close'):
                    try:
                        img.close()
                    except:
                        pass
        
        if hasattr(self, 'session_data') and self.session_data:
            for entry in self.session_data:
                if 'user_drawing' in entry and entry['user_drawing']:
                    entry['user_drawing'] = None
            self.session_data.clear()
        
        # Réinit avec nettoyage parallèle
        self._stop_background_processing()
        self.results_cache.clear()
        while not self.processing_queue.empty():
            try:
                self.processing_queue.get_nowait()
            except queue.Empty:
                break
        
        self.is_running = True
        self.start_time = time.time()
        self.user_images = []
        self.expected_answers = []
        self.operations_history = []
        self.question_count = 0
        self.time_remaining = self.duration
        self.session_data = []
        
        # Reset export
        self.export_status = "not_exported"
        self.export_timestamp = None
        self.export_result = None
        
        # Démarrer le traitement parallèle
        self._start_background_processing()
        
        gc.collect()

        # Première opération
        operation_str, answer = self.generate_operation(self.operation_type, self.difficulty)
        self.current_operation = operation_str
        self.correct_answer = answer

        # Parser l'opération pour l'historique
        parts = operation_str.split()
        a, op, b = int(parts[0]), parts[1], int(parts[2])
        self.operations_history.append((a, b, op, answer))

        # Affichage adapté selon l'opération
        operation_emoji = {
            "×": "✖️", "+": "➕", "-": "➖", "÷": "➗", "Aléatoire": "🎲"
        }
        emoji = operation_emoji.get(self.operation_type, "🔢")

        return (
            f'<div style="font-size: 3em; font-weight: bold; text-align: center; padding: 20px; background: linear-gradient(45deg, #667eea 0%, #764ba2 100%); color: white; border-radius: 10px;">{operation_str}</div>',
            create_white_canvas(),
            f"🎯 {emoji} {self.operation_type} • {self.difficulty} • Écrivez votre réponse !",
            f"⏱️ Temps restant: {self.time_remaining}s",
            gr.update(interactive=False),
            gr.update(interactive=True),
            ""
        )

    def next_question(self, image_data: dict | np.ndarray | Image.Image | None) -> tuple[str, Image.Image, str, str, gr.update, gr.update, str]:
        if not self.is_running:
            return (
                f'<div style="font-size: 3em; font-weight: bold; text-align: center; padding: 20px; background: linear-gradient(45deg, #667eea 0%, #764ba2 100%); color: white; border-radius: 10px;">{self.current_operation}</div>',
                image_data,
                "❌ Le jeu n'est pas en cours !",
                "⏱️ Temps: 0s",
                gr.update(interactive=True),
                gr.update(interactive=False),
                ""
            )

        elapsed_time = time.time() - self.start_time
        if elapsed_time >= self.duration:
            return self.end_game(image_data)

        if image_data is not None:
            # Ajouter l'image à la liste ET au traitement parallèle
            self.user_images.append(image_data)
            self.expected_answers.append(self.correct_answer)
            
            # Parser l'opération actuelle pour le traitement
            parts = self.current_operation.split()
            a, op, b = int(parts[0]), parts[1], int(parts[2])
            current_operation_data = (a, b, op, self.correct_answer)
            
            # Lancer le traitement en parallèle de l'image qu'on vient de recevoir
            self._add_image_to_processing_queue(self.question_count, image_data, self.correct_answer, current_operation_data)
            
            self.question_count += 1

        # Nouvelle opération
        operation_str, answer = self.generate_operation(self.operation_type, self.difficulty)
        self.current_operation = operation_str
        self.correct_answer = answer

        # Parser pour l'historique
        parts = operation_str.split()
        a, op, b = int(parts[0]), parts[1], int(parts[2])
        self.operations_history.append((a, b, op, answer))

        time_remaining = max(0, self.duration - int(elapsed_time))
        self.time_remaining = time_remaining

        if time_remaining <= 0:
            return self.end_game(image_data)

        # Emoji pour l'opération
        operation_emoji = {
            "×": "✖️", "+": "➕", "-": "➖", "÷": "➗", "Aléatoire": "🎲"
        }
        emoji = operation_emoji.get(self.operation_type, "🔢")

        return (
            f'<div style="font-size: 3em; font-weight: bold; text-align: center; padding: 20px; background: linear-gradient(45deg, #667eea 0%, #764ba2 100%); color: white; border-radius: 10px;">{operation_str}</div>',
            create_white_canvas(),
            f"🎯 {emoji} Question {self.question_count + 1} • {self.difficulty}",
            f"⏱️ Temps restant: {time_remaining}s",
            gr.update(interactive=False),
            gr.update(interactive=True),
            ""
        )

    def end_game(self, final_image: dict | np.ndarray | Image.Image | None) -> tuple[str, Image.Image, str, str, gr.update, gr.update, str]:

        self.is_running = False
        
        # Arrêter le traitement parallèle
        self._stop_background_processing()
        
        print("🏁 Fin de jeu - Assemblage des résultats...")

        if final_image is not None:
            self.user_images.append(final_image)
            self.expected_answers.append(self.correct_answer)
            
            # Traitement de la dernière image
            parts = self.current_operation.split()
            a, op, b = int(parts[0]), parts[1], int(parts[2])
            final_operation_data = (a, b, op, self.correct_answer)
            
            # Traiter la dernière image immédiatement (pas en parallèle)
            print(f"🔄 Traitement final de l'image {self.question_count}...")
            final_result = create_result_row_with_images(self.question_count, final_image, self.correct_answer, final_operation_data)
            self.results_cache[self.question_count] = final_result
            
            self.question_count += 1
            if len(self.operations_history) < len(self.user_images):
                self.operations_history.append((a, b, op, self.correct_answer))

        # Attendre que toutes les images soient traitées
        max_wait = 10
        wait_start = time.time()
        expected_results = len(self.user_images)
        
        print(f"⏳ Attente de {expected_results} résultats...")
        while len(self.results_cache) < expected_results and (time.time() - wait_start) < max_wait:
            time.sleep(0.1)
            
        results_ready = len(self.results_cache)
        print(f"✅ {results_ready}/{expected_results} résultats prêts")

        # Assembler les résultats dans l'ordre
        correct_answers = 0
        total_questions = len(self.user_images)
        table_rows_html = ""
        
        session_timestamp = datetime.datetime.now().isoformat()
        session_id = f"session_{int(datetime.datetime.now().timestamp())}_{str(uuid.uuid4())[:8]}"
        
        self.session_data = []
        images_saved = 0
        total_image_size_kb = 0
        
        print(f"📊 Assemblage de {total_questions} résultats...")
        
        for i in range(total_questions):
            if i in self.results_cache:
                row_data = self.results_cache[i]
                print(f"  ✅ Résultat {i} du cache parallèle")
            else:
                print(f"  🔄 Traitement fallback pour résultat {i}...")
                if i < len(self.operations_history):
                    row_data = create_result_row_with_images(i, self.user_images[i], self.expected_answers[i], self.operations_history[i])
                else:
                    row_data = {
                        'html_row': f'<tr><td>{i+1}</td><td colspan="7">Erreur traitement</td></tr>',
                        'is_correct': False,
                        'recognized': "0",
                        'recognized_num': 0,
                        'dataset_image_data': None
                    }
            
            table_rows_html += row_data['html_row']
            
            if row_data['is_correct']:
                correct_answers += 1

            # Structure pour dataset avec debug OCR
            a, b, operation, correct_result = self.operations_history[i] if i < len(self.operations_history) else (0, 0, "×", 0)
            
            try:
                ocr_info_data = get_ocr_model_info()
                print(f"🔍 Debug OCR info: {ocr_info_data}")
            except Exception as e:
                print(f"❌ Erreur get_ocr_model_info: {e}")
                ocr_info_data = {"model_name": "Error", "device": "Unknown"}
            
            entry = {
                "session_id": session_id,
                "timestamp": session_timestamp,
                "question_number": i + 1,
                "session_duration": self.duration,
                "operation_type": self.operation_type,
                "difficulty_level": self.difficulty,
                "operand_a": a,
                "operand_b": b,
                "operation": operation,
                "correct_answer": self.expected_answers[i] if i < len(self.expected_answers) else 0,
                "ocr_model": ocr_info_data.get("model_name", "Unknown"),
                "ocr_device": ocr_info_data.get("device", "Unknown"),
                "user_answer_ocr": row_data['recognized'],
                "user_answer_parsed": row_data['recognized_num'],
                "is_correct": row_data['is_correct'],
                "total_questions": total_questions,
                "app_version": "3.0_calcul_ocr_parallel"
            }
            
            print(f"🔍 Debug entry OCR fields: ocr_model={entry['ocr_model']}, ocr_device={entry['ocr_device']}")
            
            
            if row_data['dataset_image_data']:
                entry["handwriting_image"] = row_data['dataset_image_data']["image_base64"]
                entry["image_width"] = int(row_data['dataset_image_data']["compressed_size"][0])
                entry["image_height"] = int(row_data['dataset_image_data']["compressed_size"][1])
                entry["image_size_kb"] = float(row_data['dataset_image_data']["file_size_kb"])
                entry["has_image"] = True
                images_saved += 1
                total_image_size_kb += row_data['dataset_image_data']["file_size_kb"]
            else:
                entry["has_image"] = False
                
            self.session_data.append(entry)

        accuracy = (correct_answers / total_questions * 100) if total_questions > 0 else 0
        
        for entry in self.session_data:
            entry["session_accuracy"] = accuracy

        # Nettoyage mémoire
        for img in self.user_images:
            if hasattr(img, 'close'):
                try:
                    img.close()
                except:
                    pass
        
        gc.collect()

        # HTML résultats
        table_html = f"""
        <div style="overflow-x: auto; margin: 20px 0;">
            <table style="width: 100%; border-collapse: collapse; border: 2px solid #4a90e2;">
                <thead>
                    <tr style="background: #4a90e2; color: white;">
                        <th style="padding: 8px;">Question</th>
                        <th style="padding: 8px;">A</th>
                        <th style="padding: 8px;">Op</th>
                        <th style="padding: 8px;">B</th>
                        <th style="padding: 8px;">Réponse</th>
                        <th style="padding: 8px;">Votre dessin</th>
                        <th style="padding: 8px;">OCR</th>
                        <th style="padding: 8px;">Statut</th>
                    </tr>
                </thead>
                <tbody>
                    {table_rows_html}
                </tbody>
            </table>
        </div>
        """

        # Configuration session pour affichage
        config_display = f"{self.operation_type} • {self.difficulty} • {self.duration}s"
        operation_emoji = {
            "×": "✖️", "+": "➕", "-": "➖", "÷": "➗", "Aléatoire": "🎲"
        }
        emoji = operation_emoji.get(self.operation_type, "🔢")

        export_info = self.get_export_status()
        if export_info["can_export"]:
            export_section = f"""
            <div style="margin-top: 20px; padding: 15px; background-color: #e8f5e8; border-radius: 8px;">
                <h3 style="color: #2e7d32;">📤 Ajouter cette série au dataset ?</h3>
                <p style="color: #2e7d32;">
                    ✅ {total_questions} réponses • 📊 {accuracy:.1f}% de précision<br>
                    📸 {images_saved} opérations et images sauvegardées ({total_image_size_kb:.1f}KB)<br>
                    ⚙️ Configuration: {config_display}
                </p>
            </div>
            """
        else:
            export_section = ""

        final_results = f"""
        <div style="margin: 20px 0;">
            <h1 style="text-align: center; color: #4a90e2;">🎉 Session terminée !</h1>
            <div style="background: linear-gradient(45deg, #667eea 0%, #764ba2 100%); color: white; padding: 20px; border-radius: 10px; margin: 20px 0;">
                <h2>📈 Résultats</h2>
                <div style="text-align: center; margin-bottom: 15px;">
                    <strong>{emoji} {config_display}</strong>
                </div>
                <div style="display: flex; justify-content: space-around; flex-wrap: wrap;">
                    <div style="text-align: center; margin: 10px;">
                        <div style="font-size: 2em; font-weight: bold;">{total_questions}</div>
                        <div>Questions</div>
                    </div>
                    <div style="text-align: center; margin: 10px;">
                        <div style="font-size: 2em; font-weight: bold; color: #90EE90;">{correct_answers}</div>
                        <div>Correctes</div>
                    </div>
                    <div style="text-align: center; margin: 10px;">
                        <div style="font-size: 2em; font-weight: bold; color: #FFB6C1;">{total_questions - correct_answers}</div>
                        <div>Incorrectes</div>
                    </div>
                    <div style="text-align: center; margin: 10px;">
                        <div style="font-size: 2em; font-weight: bold;">{accuracy:.1f}%</div>
                        <div>Précision</div>
                    </div>
                </div>
            </div>
            <h2 style="color: #4a90e2;">📊 Détail des Réponses</h2>
            {table_html}
            {export_section}
        </div>
        """

        return (
            """<div style="font-size: 3em; font-weight: bold; text-align: center; padding: 20px; background: linear-gradient(45deg, #667eea 0%, #764ba2 100%); color: white; border-radius: 10px;">🏁 C'est fini !</div>""",
            create_white_canvas(),
            f"✨ Session {config_display} terminée !",
            "⏱️ Temps écoulé !",
            gr.update(interactive=True),
            gr.update(interactive=False),
            final_results
        )


def export_to_clean_dataset(session_data: list[dict], dataset_name: str = None) -> str:
    """Export vers le nouveau dataset calcul_ocr_dataset"""
    if dataset_name is None:
        dataset_name = DATASET_NAME  # Utiliser la variable globale
        
    if not DATASET_AVAILABLE:
        return "❌ Modules dataset non disponibles"
    
    hf_token = os.getenv("HF_TOKEN") or os.getenv("tk_calcul_ocr")
    if not hf_token:
        return "❌ Token HuggingFace manquant (HF_TOKEN ou tk_calcul_ocr)"
    
    try:
        print(f"\n🚀 === EXPORT VERS DATASET CALCUL OCR ===")
        print(f"📊 Dataset: {dataset_name}")
        
        # Filtrer les entrées avec images et ajouter les infos OCR globalement
        clean_entries = []
        
        # Récupérer une seule fois les infos OCR pour toute la session
        try:
            global_ocr_info = get_ocr_model_info()
            print(f"🔍 Infos OCR globales: {global_ocr_info}")
        except Exception as e:
            print(f"❌ Erreur infos OCR globales: {e}")
            global_ocr_info = {"model_name": "Unknown", "device": "Unknown"}
        
        for entry in session_data:
            if entry.get('has_image', False):
                # Ajouter explicitement les champs OCR manquants
                entry_with_ocr = entry.copy()
                entry_with_ocr["ocr_model"] = global_ocr_info.get("model_name", "Unknown")
                entry_with_ocr["ocr_device"] = global_ocr_info.get("device", "Unknown")
                
                print(f"🔍 Entry avec OCR: ocr_model={entry_with_ocr['ocr_model']}, ocr_device={entry_with_ocr['ocr_device']}")
                clean_entries.append(entry_with_ocr)
        
        # Créer un dataset de test avec structure forcée
        if len(clean_entries) == 0:
            return "❌ Aucune entrée avec image à exporter"
        
        # Vérifier la structure de la première entrée
        sample_entry = clean_entries[0]
        print(f"🔍 Structure première entrée: {list(sample_entry.keys())}")
        print(f"🔍 OCR dans entrée: ocr_model={sample_entry.get('ocr_model', 'MISSING')}, ocr_device={sample_entry.get('ocr_device', 'MISSING')}")
        
        # Charger dataset existant et combiner (IMPORTANT!)
        try:
            existing_dataset = load_dataset(dataset_name, split="train")
            existing_data = existing_dataset.to_list()
            print(f"📊 {len(existing_data)} entrées existantes trouvées")
            
            # Combiner ancien + nouveau
            combined_data = existing_data + clean_entries
            clean_dataset = Dataset.from_list(combined_data)
            print(f"📊 Dataset combiné: {len(existing_data)} existantes + {len(clean_entries)} nouvelles = {len(combined_data)} total")
            
        except Exception as e:
            print(f"📊 Dataset non trouvé, création nouveau: {e}")
            # Si le dataset n'existe pas, créer depuis les nouvelles entrées
            clean_dataset = Dataset.from_list(clean_entries)
            print(f"📊 Nouveau dataset créé avec {len(clean_entries)} entrées")
        
        print(f"✅ Dataset créé - Features:")
        for feature_name in clean_dataset.features:
            print(f"  - {feature_name}: {clean_dataset.features[feature_name]}")
        
        # Statistiques par opération
        operations_count = {}
        for entry in clean_entries:
            op = entry.get('operation_type', 'unknown')
            operations_count[op] = operations_count.get(op, 0) + 1
        
        operations_summary = ", ".join([f"{op}: {count}" for op, count in operations_count.items()])
        
        # Push vers HuggingFace  
        print(f"📤 Push vers {dataset_name}...")
        clean_dataset.push_to_hub(
            dataset_name,
            private=False,
            token=hf_token,
            commit_message=f"Add {len(clean_entries)} handwriting samples for math OCR ({operations_summary})"
        )
        
        cleanup_memory()
        
        return f"""✅ Session ajoutée au dataset avec succès !

📊 Dataset: {dataset_name}
📸 Images: {len(clean_entries)}
🔢 Opérations: {operations_summary}
📈 Total: {len(clean_dataset)}

🔗 Le dataset est consultable ici : https://huggingface.co/datasets/{dataset_name}"""
        
    except Exception as e:
        print(f"❌ ERREUR: {e}")
        import traceback
        traceback.print_exc()
        return f"❌ Erreur: {str(e)}"