backend/server.py

import os
import socketio
import uvicorn
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
import asyncio
import base64
import cv2
import numpy as np
import random
import time
from deepface import DeepFace
from supabase import create_client, Client

# --- 1. SETUP SUPABASE ---
SUPABASE_URL = os.getenv("SUPABASE_URL", "https://gwjrwejdjpctizolfkcz.supabase.co")
SUPABASE_KEY = os.getenv("SUPABASE_KEY", "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6Imd3anJ3ZWpkanBjdGl6b2xma2N6Iiwicm9sZSI6InNlcnZpY2Vfcm9sZSIsImlhdCI6MTc2OTA5ODEyNCwiZXhwIjoyMDg0Njc0MTI0fQ.EjU1DGTN-jrdkaC6nJWilFtYZgtu-NKjnfiMVMnHal0")

try:
    supabase: Client = create_client(SUPABASE_URL, SUPABASE_KEY)
    print("☁️ SUPABASE CONNECTÉ")
except Exception as e:
    print(f"❌ ERREUR SUPABASE: {e}")

# --- 2. CONFIG SOCKET ---
sio = socketio.AsyncServer(
    async_mode='asgi', 
    cors_allowed_origins='*',
    ping_timeout=60,
    max_http_buffer_size=10000000
)

app = FastAPI()
app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_credentials=True, allow_methods=["*"], allow_headers=["*"])
socket_app = socketio.ASGIApp(sio, app)

# --- 3. API ADMIN ---
@app.get("/api/sessions")
def get_sessions():
    try:
        response = supabase.table('sessions').select("*").order('id', desc=True).execute()
        return response.data
    except: return []

@app.get("/api/sessions/{session_id}")
def get_session_details(session_id: int):
    try:
        sess = supabase.table('sessions').select("*").eq('id', session_id).execute()
        if not sess.data: raise HTTPException(status_code=404, detail="Session introuvable")
        meas = supabase.table('measurements').select("*").eq('session_id', session_id).order('session_time', desc=False).execute()
        return {"info": sess.data[0], "data": meas.data}
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.delete("/api/sessions/{session_id}")
def delete_session(session_id: int):
    try:
        supabase.table('sessions').delete().eq('id', session_id).execute()
        return {"message": "Supprimé"}
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

# --- 4. GESTION DES SESSIONS ---
sessions = {}

@sio.event
async def connect(sid, environ):
    print(f"✅ Client: {sid}")
    sessions[sid] = {
        "active": False, 
        "start_time": 0, 
        "db_id": None,
        "last_save_time": 0 # Pour éviter de saturer la DB
    }

@sio.event
async def disconnect(sid):
    if sid in sessions: del sessions[sid]

@sio.event
async def start_session(sid, data):
    print(f"▶️ START: {sid}")
    if sid in sessions:
        sessions[sid]["active"] = True
        sessions[sid]["start_time"] = time.time()
        sessions[sid]["last_save_time"] = 0
        
        try:
            # Création de la ligne "Session"
            new_session = { 
                "first_name": data.get('firstName', 'Inconnu'), 
                "last_name": data.get('lastName', ''), 
                "client_id": data.get('clientId', '') 
            }
            res = supabase.table('sessions').insert(new_session).execute()
            sessions[sid]["db_id"] = res.data[0]['id']
            print(f"💾 Session ID {sessions[sid]['db_id']} créée.")
        except Exception as e:
            print(f"⚠️ Erreur Création Session: {e}")

@sio.event
async def stop_session(sid):
    print(f"⏹️ STOP: {sid}")
    if sid in sessions: sessions[sid]["active"] = False

@sio.event
async def process_frame(sid, data_uri):
    if sid not in sessions: return

    try:
        # A. Décodage
        encoded_data = data_uri.split(',')[1]
        nparr = np.frombuffer(base64.b64decode(encoded_data), np.uint8)
        frame = cv2.imdecode(nparr, cv2.IMREAD_COLOR)

        # B. Analyse IA
        result = DeepFace.analyze(frame, actions=['emotion'], enforce_detection=False, silent=True)
        data = result[0] if isinstance(result, list) else result
        
        emotion = data['dominant_emotion']
        emotion_score = data['emotion'][emotion] # On récupère le score de confiance (ex: 98.5)
        
        # C. Calcul KPIs & Temps
        current_time = 0
        if sessions[sid]["active"]:
            current_time = int(time.time() - sessions[sid]["start_time"])
        
        # Coordonnées visage
        region = data['region']
        face_coords = {'x': region['x'], 'y': region['y'], 'w': region['w'], 'h': region['h']} if region['w'] > 0 else None
        
        # Algorithme Métriques
        valence = 0.8 if emotion == "happy" else (-0.6 if emotion in ["sad", "angry", "fear"] else 0.0)
        arousal = 0.8 if emotion in ["angry", "fear", "surprise"] else 0.3
        
        def clamp(n): return max(0, min(100, int(n)))
        
        val_eng = clamp((arousal * 100) + random.uniform(0, 10))
        val_sat = clamp(((valence + 1) / 2) * 100)
        val_tru = clamp(50 + (valence * 20))
        val_loy = clamp(50 + (valence * 10))
        val_opi = clamp(((valence + 1) / 2) * 100)
        
        # Labels Textuels
        lbl_eng = "Fort 🔥" if val_eng > 60 else ("Moyen" if val_eng > 30 else "Faible 💤")
        lbl_sat = "Positif 😃" if val_sat > 60 else ("Négatif 😡" if val_sat < 40 else "Neutre 😐")

        metrics = {
            "engagement": val_eng, "satisfaction": val_sat, "trust": val_tru, "loyalty": val_loy, "opinion": val_opi
        }
        
        # D. Envoi au Frontend (Temps Réel)
        payload = {
            "emotion": emotion,
            "face_coords": face_coords,
            "metrics": metrics,
            "session_time": current_time,
            "is_recording": sessions[sid]["active"]
        }
        await sio.emit('metrics_update', payload, room=sid)
        
        # E. SAUVEGARDE EN BASE DE DONNÉES (La partie qui manquait !)
        # On sauvegarde seulement si : 
        # 1. L'enregistrement est actif
        # 2. On a un ID de session valide
        # 3. Ça fait plus d'1 seconde depuis la dernière sauvegarde (pour ne pas saturer)
        now = time.time()
        if sessions[sid]["active"] and sessions[sid]["db_id"]:
            if now - sessions[sid]["last_save_time"] >= 1.0:
                sessions[sid]["last_save_time"] = now
                
                # Préparation de la ligne à insérer
                row_data = {
                    "session_id": sessions[sid]["db_id"],
                    "session_time": current_time,
                    "emotion": emotion,
                    "emotion_score": float(emotion_score),
                    "engagement_val": val_eng, "engagement_lbl": lbl_eng,
                    "satisfaction_val": val_sat, "satisfaction_lbl": lbl_sat,
                    "trust_val": val_tru, "loyalty_val": val_loy, "opinion_val": val_opi
                }
                
                # Insertion asynchrone (on ne bloque pas)
                try:
                    supabase.table('measurements').insert(row_data).execute()
                    # print(f"💾 Data saved T={current_time}") # Décommenter pour vérifier
                except Exception as db_err:
                    print(f"⚠️ Erreur Insert Mesure: {db_err}")

    except Exception as e:
        pass # On continue même si une frame est mauvaise

if __name__ == "__main__":
    try:
        DeepFace.build_model("Emotion")
        print("✅ Modèle chargé !")
    except: pass
    uvicorn.run(socket_app, host="0.0.0.0", port=7860)