Add check_file_id to insert_datasets

isimip_publisher/commands.py

@@ -213,6 +213,10 @@ def insert_datasets():
 
     session = database.init_database_session(settings.DATABASE)
 
+    for dataset in store.datasets:
+        for file in dataset.files:
+            database.check_file_id(session, file.path, file.uuid)
+
     for dataset in tqdm(store.datasets, desc='insert_datasets'.ljust(18)):
         database.insert_dataset(session, settings.VERSION, settings.RIGHTS, settings.RESTRICTED,
                                 dataset.name, dataset.path, dataset.size, dataset.specifiers)

isimip_publisher/utils/database.py

@@ -389,6 +389,12 @@ def retrieve_datasets(session, path, public=None, follow=False, like=True):
     return datasets
 
 
+def check_file_id(session, path, uuid):
+    file = session.query(File).filter(File.id == uuid).one_or_none()
+    if file:
+        raise RuntimeError(f'File {path} has an id which already exists in the database ({uuid})')
+
+
 def insert_file(session, version, dataset_path, uuid, name, path, size,
                 checksum, checksum_type, netcdf_header, specifiers):
     # get the dataset from the database