Merge branch 'main' into u/jrbogart/specify_prod_schema

CHANGELOG.md

@@ -9,6 +9,13 @@ its associated production schema and form foreign key constraints correctly.
 Bumped database version to 2.3.0.  This code requires
 database version >= 2.3.0
 
+## Version 0.4.2
+
+- Add check during dataset registration to raise an exception if the `root_dir`
+  does not exist
+- Add check before copying any data (i.e., `old_location != None`) that the
+  user has write permission to the `root_dir` folder.
+
 ## Version 0.4.1
 
 Add ability to register "external" datasets. For example datasets that are not

src/dataregistry/_version.py

@@ -1 +1 @@
-__version__ = "0.5.0"
+__version__ = "0.5.0"

src/dataregistry/registrar/dataset.py

@@ -27,6 +27,12 @@ def __init__(self, db_connection, root_dir, owner, owner_type, execution_table):
         self.which_table = "dataset"
         self.entry_id = "dataset_id"
 
+        # Does the root_dir exist?
+        self.root_dir_exists = os.path.isdir(root_dir)
+
+        # Does the user have write permission to the root_dir?
+        self.root_dir_write_access = os.access(root_dir, os.W_OK)
+
     def register(
         self,
         relative_path,
@@ -121,6 +127,10 @@ def register(
             The execution ID associated with the dataset
         """
 
+        # If the root_dir does not exist, stop
+        if not self.root_dir_exists:
+            raise FileNotFoundError(f"root_dir {self._root_dir} does not exist")
+
         # If external dataset, check for either a `url` or `contact_email`
         if location_type == "external":
             if url is None and contact_email is None:
@@ -369,6 +379,11 @@ def _handle_data(self, relative_path, old_location, owner, owner_type, verbose):
 
         # Copy data into data registry
         if old_location:
+
+            # Stop if we don't have write permission to the root_dir
+            if not self.root_dir_write_access:
+                raise Exception(f"Cannot copy data, no write access to {self._root_dir}")
+
             if verbose:
                 tic = time.time()
                 print(