#!/bin/bash
# Reprocessing des 8 sessions existantes
# Génère embeddings + workflows rétroactivement

set -e

echo "=== Reprocessing Sessions Existantes ==="
echo ""

# Couleurs
GREEN='\033[0;32m'
BLUE='\033[0;34m'
YELLOW='\033[1;33m'
NC='\033[0m'

# Trouver toutes les sessions dans screen_states
echo -e "${BLUE}Recherche des sessions à reprocesser...${NC}"
cd /opt/rpa_vision_v3/data/training/screen_states

# Extraire session_ids uniques depuis les screen_states
session_ids=$(find . -name "*.json" -exec grep -h '"session_id"' {} \; | \
              sed 's/.*"session_id": "\(sess_[^"]*\)".*/\1/' | \
              sort -u)

session_count=$(echo "$session_ids" | wc -l)
echo "  ✓ Trouvé $session_count sessions à reprocesser"
echo ""

# Compter pour stats
total=0
success=0
failed=0

echo -e "${BLUE}Lancement du reprocessing...${NC}"
echo ""

for session_id in $session_ids; do
    total=$((total + 1))
    echo "[$total/$session_count] Processing $session_id..."

    # Appeler le pipeline de traitement
    if sudo -u rpa /opt/rpa_vision_v3/venv_v3/bin/python \
         /opt/rpa_vision_v3/server/processing_pipeline.py "$session_id" 2>&1 | \
         grep -E "(ScreenStates|Embeddings|Workflow)"; then
        success=$((success + 1))
        echo -e "  ${GREEN}✓ Succès${NC}"
    else
        failed=$((failed + 1))
        echo -e "  ${YELLOW}⚠ Échec ou déjà traité${NC}"
    fi
    echo ""
done

echo ""
echo -e "${GREEN}=== Reprocessing Terminé ===${NC}"
echo ""
echo "Statistiques:"
echo "  Total sessions     : $total"
echo "  Succès             : $success"
echo "  Échecs             : $failed"
echo ""

echo "Vérification des résultats:"
echo ""
echo "Embeddings générés:"
embedding_count=$(find /opt/rpa_vision_v3/data/training/embeddings -name "*.npy" 2>/dev/null | wc -l)
echo "  $embedding_count fichiers .npy"
echo ""

echo "Workflows créés:"
workflow_count=$(find /opt/rpa_vision_v3/data/training/workflows -name "*.json" 2>/dev/null | wc -l)
echo "  $workflow_count fichiers .json"
ls -lh /opt/rpa_vision_v3/data/training/workflows/*.json 2>/dev/null || echo "  (aucun)"
echo ""

echo "Prototypes sauvegardés:"
prototype_count=$(find /opt/rpa_vision_v3/data/training/prototypes -name "*.npy" 2>/dev/null | wc -l)
echo "  $prototype_count fichiers .npy"
echo ""

echo "Pour voir les workflows dans le dashboard:"
echo "  http://localhost:5001 → Onglet 'Workflows'"