Implementation of a clean up function for a stray collection processing

Web/app.py

@@ -5828,7 +5828,9 @@ def upload_item():
 
     if item_id:
         success_msg = f'Element wurde erfolgreich hinzugefügt ({len(created_item_ids)} erstellt)'
+        fs = get_gridfs()  # Deine GridFS Verbindung
 
+        cleanup_orphaned_images(fs, dry_run=True)
         if upload_mode == 'library':
             try:
                 _append_audit_event_standalone(

Web/modules/database/items.py

@@ -25,6 +25,7 @@ import datetime
 import Web.modules.database.settings as cfg
 from Web.modules.database.settings import MongoClient
 import Web.modules.inventarsystem.data_protection as dp
+import logging
 
 
 def is_library_item(item):
@@ -1293,3 +1294,136 @@ def sync_group_codes(primary_obj_id, base_code, individual_codes_list):
     except Exception as e:
         print(f"Error syncing group codes: {e}")
         return False
+
+
+logging.basicConfig(level=logging.INFO, format='%(levelname)s: %(message)s')
+
+
+def cleanup_orphaned_images(fs, dry_run=True):
+    """
+    Finds images in GridFS that are no longer referenced by any item
+    and optionally deletes them.
+
+    Supported item fields:
+        - book_cover_image
+        - image
+        - images
+
+    The fields are expected to contain GridFS file ObjectIds.
+
+    :param fs: GridFS instance, e.g. gridfs.GridFS(db)
+    :param dry_run: If True, only reports what would be deleted.
+                    If False, actually deletes the files.
+    """
+
+    logging.info("Starte Cleanup-Skript...")
+
+    client = MongoClient(cfg.MONGODB_HOST, cfg.MONGODB_PORT)
+
+    try:
+        db = client[cfg.MONGODB_DB]
+        items_collection = db["items"]
+
+        referenced_files = set()
+
+        for item in items_collection.find(
+            {},
+            {
+                "Images": 1
+            }
+        ):
+            images = item.get("Images")
+
+            if isinstance(images, list):
+                for img in images:
+                    if img:
+                        referenced_files.add(img)
+
+        referenced_files.discard(None)
+
+        logging.info(
+            f"{len(referenced_files)} referenzierte GridFS-Dateien gefunden."
+        )
+
+        orphaned_files = []
+
+        for grid_file in fs.find():
+            file_id = grid_file._id
+            filename = grid_file.filename or ""
+
+            if not filename.lower().endswith(
+                (".jpg", ".jpeg", ".png", ".gif", ".webp")
+            ):
+                continue
+
+            if file_id not in referenced_files:
+                orphaned_files.append(
+                    {
+                        "_id": file_id,
+                        "filename": filename,
+                        "upload_date": grid_file.upload_date,
+                    }
+                )
+
+        logging.info(
+            f"Gefundene verwaiste Bilder: {len(orphaned_files)}"
+        )
+
+        if dry_run:
+            logging.info(
+                "--- DRY RUN AKTIV - Es wird nichts gelöscht ---"
+            )
+
+            for file in orphaned_files:
+                logging.info(
+                    f"Würde löschen: "
+                    f"{file['filename']} "
+                    f"(ID: {file['_id']}, "
+                    f"Hochgeladen: {file['upload_date']})"
+                )
+
+            logging.info(
+                "--- Setze dry_run=False, um physisch zu löschen ---"
+            )
+
+        else:
+            logging.warning("--- LÖSCHVORGANG AKTIV ---")
+
+            deleted_count = 0
+            failed_count = 0
+
+            for file in orphaned_files:
+                try:
+                    fs.delete(file["_id"])
+
+                    deleted_count += 1
+
+                    logging.info(
+                        f"Gelöscht: {file['filename']} "
+                        f"(ID: {file['_id']})"
+                    )
+
+                except Exception as e:
+                    failed_count += 1
+
+                    logging.error(
+                        f"Fehler beim Löschen von "
+                        f"{file['filename']} "
+                        f"(ID: {file['_id']}): {e}"
+                    )
+
+            logging.info(
+                f"Cleanup beendet. "
+                f"{deleted_count} Bilder gelöscht, "
+                f"{failed_count} Fehler."
+            )
+
+        return {
+            "status": "success",
+            "referenced_count": len(referenced_files),
+            "orphaned_count": len(orphaned_files),
+            "dry_run": dry_run,
+        }
+
+    finally:
+        client.close()