rpa_vision_v3/scripts/backup_vwb_and_audit.sh

#!/bin/bash
################################################################################
# backup_vwb_and_audit.sh
#
# Backup quotidien critique avant POC Anouste.
# Cf. challenge du 16 avril 2026 : sans backup, perte de workflows.db =
# perte directe de travail client. Ce script doit tourner AVANT tout
# déploiement chez un client.
#
# Ce qu'il sauvegarde :
#   - visual_workflow_builder/backend/instance/workflows.db
#     → ~/backups/vwb/workflows_YYYY-MM-DD.db
#   - data/audit/*.jsonl
#     → ~/backups/audit/audit_YYYY-MM-DD/
#
# Rétention : 30 jours (suppression automatique des backups plus anciens).
# Log       : ~/backups/backup.log (append, horodaté).
#
# Installation (non automatique — à faire à la main) :
#   crontab -e
#   0 2 * * * /home/dom/ai/rpa_vision_v3/scripts/backup_vwb_and_audit.sh
#   → s'exécute tous les jours à 2h du matin.
#
# Procédure de restore : voir ~/backups/README.md
#
# Auteur : Dom + Claude — 16 avril 2026
################################################################################

set -u  # strict: variable non définie = erreur

# ------------------------------------------------------------------------------
# Chemins
# ------------------------------------------------------------------------------
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"

SRC_WORKFLOWS_DB="$PROJECT_ROOT/visual_workflow_builder/backend/instance/workflows.db"
SRC_AUDIT_DIR="$PROJECT_ROOT/data/audit"

BACKUP_ROOT="${BACKUP_ROOT:-$HOME/backups}"
BACKUP_VWB_DIR="$BACKUP_ROOT/vwb"
BACKUP_AUDIT_DIR="$BACKUP_ROOT/audit"
BACKUP_LOG="$BACKUP_ROOT/backup.log"

RETENTION_DAYS="${RETENTION_DAYS:-30}"
DATE_TAG="$(date +%Y-%m-%d)"
NOW="$(date '+%Y-%m-%d %H:%M:%S')"

# ------------------------------------------------------------------------------
# Helpers
# ------------------------------------------------------------------------------
log() {
    local msg="$1"
    echo "[$NOW] $msg" >> "$BACKUP_LOG"
    echo "$msg"
}

# ------------------------------------------------------------------------------
# Préparation
# ------------------------------------------------------------------------------
mkdir -p "$BACKUP_VWB_DIR" "$BACKUP_AUDIT_DIR"
touch "$BACKUP_LOG"

log "=== Début backup VWB + audit ==="

# ------------------------------------------------------------------------------
# 1. workflows.db
# ------------------------------------------------------------------------------
if [ -f "$SRC_WORKFLOWS_DB" ]; then
    DEST_DB="$BACKUP_VWB_DIR/workflows_${DATE_TAG}.db"
    # On utilise sqlite3 .backup si possible (safe, même si la DB est
    # ouverte par le backend). Fallback : cp simple.
    if command -v sqlite3 > /dev/null 2>&1; then
        if sqlite3 "$SRC_WORKFLOWS_DB" ".backup '$DEST_DB'" 2>/dev/null; then
            size=$(stat -c %s "$DEST_DB" 2>/dev/null || echo "?")
            log "  [OK] workflows.db → $DEST_DB (${size} octets) via sqlite3 .backup"
        else
            cp "$SRC_WORKFLOWS_DB" "$DEST_DB"
            log "  [OK fallback] workflows.db → $DEST_DB via cp"
        fi
    else
        cp "$SRC_WORKFLOWS_DB" "$DEST_DB"
        log "  [OK] workflows.db → $DEST_DB via cp (sqlite3 absent)"
    fi
else
    log "  [WARN] workflows.db introuvable : $SRC_WORKFLOWS_DB"
fi

# ------------------------------------------------------------------------------
# 2. data/audit/*.jsonl
# ------------------------------------------------------------------------------
if [ -d "$SRC_AUDIT_DIR" ]; then
    DEST_AUDIT="$BACKUP_AUDIT_DIR/audit_${DATE_TAG}"
    mkdir -p "$DEST_AUDIT"
    copied=0
    # shellcheck disable=SC2045
    for f in "$SRC_AUDIT_DIR"/*.jsonl; do
        [ -f "$f" ] || continue
        cp "$f" "$DEST_AUDIT/"
        copied=$((copied + 1))
    done
    log "  [OK] $copied fichiers audit → $DEST_AUDIT"
else
    log "  [WARN] dossier audit introuvable : $SRC_AUDIT_DIR"
fi

# ------------------------------------------------------------------------------
# 3. Rétention : suppression des backups > RETENTION_DAYS jours
# ------------------------------------------------------------------------------
# On retire les fichiers .db du dossier vwb
if [ -d "$BACKUP_VWB_DIR" ]; then
    deleted_db=$(find "$BACKUP_VWB_DIR" -maxdepth 1 -name "workflows_*.db" \
        -type f -mtime +"$RETENTION_DAYS" -print -delete 2>/dev/null | wc -l)
    [ "$deleted_db" -gt 0 ] && log "  [CLEAN] $deleted_db backup(s) vwb > ${RETENTION_DAYS}j supprimé(s)"
fi
# On retire les répertoires audit daté
if [ -d "$BACKUP_AUDIT_DIR" ]; then
    deleted_audit=$(find "$BACKUP_AUDIT_DIR" -maxdepth 1 -type d \
        -name "audit_*" -mtime +"$RETENTION_DAYS" -print -exec rm -rf {} \; 2>/dev/null | wc -l)
    [ "$deleted_audit" -gt 0 ] && log "  [CLEAN] $deleted_audit backup(s) audit > ${RETENTION_DAYS}j supprimé(s)"
fi

log "=== Fin backup ==="
exit 0