Update startup script to load environment variables correctly

Modify `deployment/check_frontend.sh` to source the `.env` file before starting the Node.js development server. This ensures that environment variables, such as `DATABASE_URL`, are properly loaded, resolving startup errors on non-Replit environments. Also updates `replit.md` to document this fix.

Replit-Commit-Author: Agent
Replit-Commit-Session-Id: 7a657272-55ba-4a79-9a2e-f1ed9bc7a528
Replit-Commit-Checkpoint-Type: full_checkpoint
Replit-Commit-Event-Id: 21e2b3ab-3e67-45ff-819b-a2db15c27292
Replit-Commit-Screenshot-Url: https://storage.googleapis.com/screenshot-production-us-central1/449cf7c4-c97a-45ae-8234-e5c5b8d6a84f/7a657272-55ba-4a79-9a2e-f1ed9bc7a528/dI1I79r

.replit

@@ -14,6 +14,10 @@ run = ["npm", "run", "start"]
 localPort = 5000
 externalPort = 80
 
+[[ports]]
+localPort = 34661
+externalPort = 3001
+
 [[ports]]
 localPort = 41303
 externalPort = 3002

deployment/check_frontend.sh

@@ -25,6 +25,11 @@ else
     # Wait a moment
     sleep 2
     
+    # Load environment variables from .env file
+    set -a
+    source "$WORK_DIR/.env" 2>/dev/null || true
+    set +a
+    
     # Start frontend
     cd "$WORK_DIR"
     nohup npm run dev >> "$LOG_FILE" 2>&1 &

replit.md

@@ -43,6 +43,19 @@ Sistema di rilevamento intrusioni per router MikroTik basato su Machine Learning
 
 ## Fix Recenti (Novembre 2025)
 
+### ✅ Frontend Environment Variables Fix (21 Nov 2025 - 16:30)
+- **Problema**: Frontend Node.js non si avviava su server AlmaLinux con errore `DATABASE_URL must be set`
+- **Causa**: Script `check_frontend.sh` non caricava variabili d'ambiente dal file `.env`
+- **Soluzione**: 
+  - Modificato `deployment/check_frontend.sh` per caricare `.env` prima di eseguire `npm run dev`
+  - Usato `set -a; source .env; set +a` per esportare tutte le variabili
+- **Risultato**: Frontend si avvia correttamente leggendo `DATABASE_URL` e altre variabili da `.env`
+
+### ✅ Form Validation Migliorata (21 Nov 2025 - 15:00)
+- **API ML Endpoints**: Timeout configurabili (120s train/detect, 10s stats), validazione input, messaggi errore specifici (504 timeout, 503 backend down, 400 validation)
+- **Whitelist Form**: Convertito a react-hook-form + zodResolver, validazione IP completa (regex + controllo ottetti 0-255)
+- **Training Forms**: Due form separati (trainForm/detectForm), schema Zod con z.coerce.number(), FormDescription per suggerimenti
+
 ### ✅ Sistema Completamente Funzionante (17 Nov 2025 - 19:30)
 - **Backend Python FastAPI**: ✅ Porta 8000, modello ML caricato, endpoint /stats funzionante
 - **Database PostgreSQL**: ✅ 5 tabelle (network_logs, detections, routers, whitelist, training_history)