Add logging for legacy private dataset access (#13)

portal-backend/depmap/interactive/config/models.py

@@ -506,6 +506,12 @@ def _get_nonstandard_noncustom_datasets(self):
             nonstandard_datasets[key] = Config(**nonstandard_dict)
         return nonstandard_datasets
 
+    def is_legacy_private_dataset(self, dataset_id: str) -> bool:
+        all_private_dataset_ids = [
+            dataset.dataset_id for dataset in PrivateDatasetMetadata.get_all()
+        ]
+        return dataset_id in all_private_dataset_ids
+
     def get_allowed_private_datasets(self) -> Dict[str, Config]:
         """
         WARNING: This CANNOT BE CACHED and must be re-retrieved on every request.

portal-backend/depmap/interactive/config/utils.py

@@ -18,6 +18,7 @@
 from depmap.taiga_id import utils as taiga_utils
 from depmap.utilities import entity_utils
 from depmap.utilities.exception import InteractiveDatasetNotFound
+from depmap.utilities.data_access_log import log_legacy_private_dataset_access
 
 
 def __get_config() -> InteractiveConfig:
@@ -73,6 +74,8 @@ def get_dataset_label(dataset_id) -> str:
     """
     Returns label of dataset
     """
+    if __get_config().is_legacy_private_dataset(dataset_id):
+        log_legacy_private_dataset_access("get_dataset_label", dataset_ids=[dataset_id])
     return __get_config().get(dataset_id).label
 
 

portal-backend/depmap/interactive/interactive_utils/get_and_process_data.py

@@ -43,6 +43,7 @@
 from depmap.vector_catalog.nodes.categorical_tree_nodes import (
     MUTATION_DETAILS_DATASET_ID,
 )
+from depmap.utilities.data_access_log import log_legacy_private_dataset_access
 
 
 def get_matrix(dataset_id):
@@ -207,6 +208,11 @@ def get_subsetted_df_by_labels(
     """
     Get a filtered dataframe with rows indexed by entity labels and columns indexed by depmap ids. 
     """
+    if __get_config().is_legacy_private_dataset(dataset_id):
+        log_legacy_private_dataset_access(
+            "get_subsetted_df_by_labels", dataset_ids=[dataset_id]
+        )
+
     row_index_to_entity_label = {}
     col_index_to_depmap_id = {}
     feature_row_labels_set = set(feature_row_labels) if feature_row_labels else set()
@@ -271,6 +277,11 @@ def get_subsetted_df_by_ids(
     :param cell_line_ids: depmap ids of cell lines to return.  If None, return all cell lines
     :return: dataframe where rows are entities and columns are cell lines
     """
+    if __get_config().is_legacy_private_dataset(dataset_id):
+        log_legacy_private_dataset_access(
+            "get_subsetted_df_by_ids", dataset_ids=[dataset_id]
+        )
+
     row_index_to_entity_label = {}
     col_index_to_depmap_id = {}
     entity_ids_set = set(entity_ids) if entity_ids else set()
@@ -318,6 +329,10 @@ def get_subsetted_df(dataset_id, row_indices, col_indices):
     if is_standard(dataset_id):
         df = standard_utils.get_subsetted_df(dataset_id, row_indices, col_indices)
     else:
+        if __get_config().is_legacy_private_dataset(dataset_id):
+            log_legacy_private_dataset_access(
+                "get_subsetted_df", dataset_ids=[dataset_id]
+            )
         df = nonstandard_utils.get_subsetted_df(dataset_id, row_indices, col_indices)
 
     return df

portal-backend/depmap/private_dataset/views.py

@@ -14,6 +14,7 @@
 )
 from depmap.vector_catalog.trees import OTHER_DATASET_NON_PREPOPULATE_ID_BASE
 from depmap.interactive.nonstandard import nonstandard_utils
+from depmap.utilities.data_access_log import log_legacy_private_dataset_access
 
 blueprint = Blueprint(
     "private_dataset",
@@ -29,6 +30,9 @@ def home():
         abort(404)
 
     datasets = data_access.get_private_datasets()
+    dataset_ids = list(datasets.keys())
+    log_legacy_private_dataset_access("get_dataset_label", dataset_ids)
+
     datasets = [
         {
             "dataset_id": dataset_id,

portal-backend/depmap/utilities/data_access_log.py

@@ -68,3 +68,25 @@ def log_bulk_download_csv():
             )
         ),
     )
+
+
+def log_legacy_private_dataset_access(function_name, dataset_ids):
+    """
+    In theory, once we switch to the new private dataset UI,
+    private datasets should no longer be accessed through the legacy system.
+    If we see logs where they are being accessed, then we'll know there are other 
+    features that need to be updated. 
+    """
+    log.info(
+        "%s",
+        json.dumps(
+            dict(
+                timestamp=datetime.datetime.now().isoformat(),
+                type="legacy-private-dataset-access",
+                endpoint=request.endpoint,
+                function=function_name,
+                dataset_ids=dataset_ids,
+                user=get_authenticated_user(),
+            )
+        ),
+    )

portal-backend/tests/conftest.py

@@ -178,6 +178,9 @@ def __format_all_private_dataset_settings(self):
             "private-b4d7094196889fa4614409570bb12ab5c09c9cc00388deb7c13ec57fd2996461": None
         }
 
+    def is_legacy_private_dataset(self, dataset_id: str) -> bool:
+        return False
+
     @classmethod
     def _get_mutations_taiga_id(cls):
         return "this-is-test-nonsense-that-should-never-be-checked-against.1/except-for-taiga-alias-loading"