BUG: Prevent creating v2 data asset without specifying datastore (#854)

Closes #786 

It should now be impossible to create data new data assets without also
specifying a datastore by its specific name, in both v1 and v2 pathways.

Default datastore interactions in all other code paths have been left
unchanged.

hi-ml-azure/src/health_azure/datasets.py

@@ -128,6 +128,11 @@ def _get_or_create_v1_dataset(datastore_name: str, dataset_name: str, workspace:
     try:
         azureml_dataset = _retrieve_v1_dataset(dataset_name, workspace)
     except UserErrorException:
+        if datastore_name == "":
+            raise ValueError(
+                "When creating a new dataset, a datastore name must be provided. Please specify a datastore name using "
+                "the --datastore flag"
+            )
         azureml_dataset = _create_v1_dataset(datastore_name, dataset_name, workspace)
     return azureml_dataset
 
@@ -193,15 +198,17 @@ def _create_v2_data_asset(
     :param datastore_name: The name of the datastore in which to create or update the Data asset.
     :param data_asset_name: The name of the data asset to be created.
     :param version: The version of the data asset to be created.
+    :raises ValueError: If no data asset name is provided to define the new asset.
     :raises ValueError: If no datastore name is provided to define where to create the data.
     :return: The created or updated Data asset.
     """
-    if not data_asset_name:
+    if data_asset_name == "":
         raise ValueError("Cannot create data asset with empty name.")
 
-    if not datastore_name:
-        default_datastore = ml_client.datastores.get_default()
-        datastore_name = default_datastore.name
+    if datastore_name == "":
+        raise ValueError(
+            "Cannot create data asset with empty datastore name. Please specify a datastore name using the --datastore flag."
+        )
 
     logging.info(
         f"Creating a new Data asset from data in folder '{data_asset_name}' in the datastore '{datastore_name}'"

hi-ml-azure/testazure/testazure/test_datasets.py

@@ -310,6 +310,16 @@ def test_get_or_create_v1_dataset() -> None:
         mocks["_create_v1_dataset"].assert_called_once()
 
 
+def test_get_or_create_v1_dataset_empty_datastore_name() -> None:
+    workspace = DEFAULT_WORKSPACE.workspace
+    datastore = ""
+    dataset_name = "foo"
+    with pytest.raises(ValueError) as ex:
+        _get_or_create_v1_dataset(datastore, dataset_name, workspace)
+
+    assert "When creating a new dataset, a datastore name must be provided." in str(ex)
+
+
 def test_get_or_create_v2_data_asset() -> None:
     def _mock_error_from_retrieve_v2_data_asset(
         ml_client: MLClient,
@@ -464,9 +474,15 @@ def test_retrieving_v2_data_asset_does_not_increment() -> None:
 
 
 @pytest.mark.parametrize(
-    ["asset_name", "version"], [(TEST_DATA_ASSET_NAME, None), (TEST_DATA_ASSET_NAME, "1"), ("", 1)]
+    ["asset_name", "datastore_name", "version"],
+    [
+        (TEST_DATA_ASSET_NAME, TEST_DATASTORE_NAME, None),
+        (TEST_DATA_ASSET_NAME, "", None),
+        (TEST_DATA_ASSET_NAME, TEST_DATASTORE_NAME, "1"),
+        ("", TEST_DATASTORE_NAME, 1),
+    ],
 )
-def test_create_v2_data_asset(asset_name: str, version: Optional[str]) -> None:
+def test_create_v2_data_asset(asset_name: str, datastore_name: str, version: Optional[str]) -> None:
     try:
         data_asset = _create_v2_data_asset(
             ml_client=TEST_ML_CLIENT,
@@ -483,6 +499,8 @@ def test_create_v2_data_asset(asset_name: str, version: Optional[str]) -> None:
     except ValueError as ex:
         if asset_name == "":
             assert "Cannot create data asset with empty name." in str(ex)
+        elif datastore_name == "":
+            assert "Cannot create data asset with empty datastore name." in str(ex)
         else:
             pytest.fail(f"Unexpected error: {ex}")
 

hi-ml-azure/testazure/testazure/test_himl.py

@@ -1782,7 +1782,12 @@ def test_create_v2_inputs(already_exists: bool) -> None:
 
     for use_mounting in [True, False]:
         mock_input_dataconfigs = [
-            DatasetConfig(name="dummy_dataset", use_mounting=use_mounting, version=int(mock_data_version))
+            DatasetConfig(
+                name="dummy_dataset",
+                datastore="dummy_datastore",
+                use_mounting=use_mounting,
+                version=int(mock_data_version),
+            )
         ]
         inputs = himl.create_v2_inputs(mock_ml_client, mock_input_dataconfigs)
         assert isinstance(inputs, Dict)