Return list of synapse ids instead of dataframe

.gitignore

@@ -0,0 +1,5 @@
+__pycache__
+*.egg-info
+dist
+build
+.vscode

synapsemonitor/monitor.py

@@ -66,41 +66,53 @@ def _render_fileview(
     return viewdf
 
 
-def _find_modified_entities_fileview(
-    syn: Synapse, syn_id: str, days: int = 1
-) -> pd.DataFrame:
-    """Performs query to find modified entities in id and render columns
-    These modified entities include newly uploaded ones
+def _find_modified_entities_fileview(syn: Synapse, syn_id: str, days: int = 1) -> list:
+    """Finds entities scoped in a fileview modified in the past N number of days
 
     Args:
         syn: Synapse connection
-        view_id: Synapse View Id
-        epochtime: Epoch time in milliseconds
+        syn_id: Synapse Fileview Id
+        days: N number of days
 
     Returns:
-        Dataframe with updated entities
+        List of synapse ids
     """
     # Update the view
     # _force_update_view(syn, view_id)
     query = (
-        "select id, name, currentVersion, modifiedOn, modifiedBy, "
-        f"createdOn, projectId, type from {syn_id} where "
+        f"select id from {syn_id} where "
         f"modifiedOn > unix_timestamp(NOW() - INTERVAL {days} DAY)*1000"
     )
     results = syn.tableQuery(query)
     resultsdf = results.asDataFrame()
-    return _render_fileview(syn, viewdf=resultsdf)
+    return resultsdf["id"].tolist()
 
 
-def _find_modified_entities_file(
-    syn: Synapse, syn_id: str, days: int = 1
-) -> pd.DataFrame:
+def _find_modified_entities_file(syn: Synapse, syn_id: str, days: int = 1) -> list:
+    """Determines if entity was modified in the past N number of days
+
+    Args:
+        syn: Synapse connection
+        syn_id: Synapse File Id
+        days: N number of days
+
+    Returns:
+        List of synapse ids
+    """
     raise NotImplementedError("Files not supported yet")
 
 
-def _find_modified_entities_container(
-    syn: Synapse, syn_id: str, days: int = 1
-) -> pd.DataFrame:
+def _find_modified_entities_container(syn: Synapse, syn_id: str, days: int = 1) -> list:
+    """Finds entities in a folder or project modified in the past N number of days
+
+    Args:
+        syn: Synapse connection
+        syn_id: Synapse Folder or Project Id
+        days: N number of days
+
+    Returns:
+        List of synapse ids
+    """
     raise NotImplementedError("Projects and folders not supported yet")
 
 
@@ -135,8 +147,17 @@ def _get_user_ids(syn: Synapse, users: list = None):
     return user_ids
 
 
-def find_modified_entities(syn: Synapse, syn_id: str, days: int) -> pd.DataFrame:
-    """Find modified entities based on the type of the input"""
+def find_modified_entities(syn: Synapse, syn_id: str, days: int) -> list:
+    """Find modified entities based on the type of the input
+
+    Args:
+        syn: Synapse connection
+        syn_id: Synapse Entity Id
+        days: N number of days
+
+    Returns:
+        List of synapse ids
+    """
     entity = syn.get(syn_id, downloadFile=False)
     if isinstance(entity, synapseclient.EntityViewSchema):
         return _find_modified_entities_fileview(syn=syn, syn_id=syn_id, days=days)
@@ -170,21 +191,21 @@ def monitoring(
         Dataframe with files modified within last N days
     """
     # get dataframe of files
-    filesdf = find_modified_entities(syn=syn, syn_id=syn_id, days=days)
+    modified_entities = find_modified_entities(syn=syn, syn_id=syn_id, days=days)
     # Filter out projects and folders
-    print(f"Total number of entities = {len(filesdf.index)}")
+    print(f"Total number of entities = {len(modified_entities)}")
 
     # get user ids
     user_ids = _get_user_ids(syn, users)
 
     # TODO: Add function to beautify email message
 
     # Prepare and send Message
-    if not filesdf.empty:
+    if modified_entities:
         syn.sendMessage(
             user_ids,
             email_subject,
-            filesdf.to_html(index=False),
+            ", ".join(modified_entities),
             contentType="text/html",
         )
-    return filesdf
+    return modified_entities

test/test_monitor.py

@@ -52,22 +52,21 @@ def test__find_modified_entities_fileview(self):
         with patch.object(self.syn, "tableQuery",
                           return_value=self.table_query_results) as patch_q,\
             patch.object(self.table_query_results, "asDataFrame",
-                         return_value=self.query_resultsdf) as patch_asdf,\
-            patch.object(monitor, "_render_fileview",
-                         return_value=self.expecteddf) as patch_render:
-            resultdf = monitor._find_modified_entities_fileview(
+                         return_value=self.query_resultsdf) as patch_asdf:
+            # patch.object(monitor, "_render_fileview",
+            #              return_value=self.expecteddf) as patch_render:
+            modified_list = monitor._find_modified_entities_fileview(
                 self.syn, "syn44444", days=2
             )
             patch_q.assert_called_once_with(
-                "select id, name, currentVersion, modifiedOn, modifiedBy, "
-                "createdOn, projectId, type from syn44444 where "
+                "select id from syn44444 where "
                 "modifiedOn > unix_timestamp(NOW() - INTERVAL 2 DAY)*1000"
             )
             patch_asdf.assert_called_once_with()
-            assert resultdf.equals(self.expecteddf)
-            patch_render.assert_called_once_with(
-                self.syn, viewdf=self.query_resultsdf
-            )
+            assert modified_list == ["syn23333"]
+            # patch_render.assert_called_once_with(
+            #     self.syn, viewdf=self.query_resultsdf
+            # )
 
 
 def test__get_user_ids_none():
@@ -148,9 +147,9 @@ def setup_method(self):
 
     def test_monitoring_fail_integration(self):
         """Test all monitoring functions are called"""
-        returndf = pd.DataFrame({"test": ["foo"]})
+        modified_list = ["syn2222", "syn33333"]
         with patch.object(monitor, "find_modified_entities",
-                          return_value=returndf) as patch_find,\
+                          return_value=modified_list) as patch_find,\
              patch.object(monitor, "_get_user_ids",
                           return_value=[111]) as patch_get_user,\
              patch.object(self.syn, "sendMessage") as patch_send:
@@ -159,5 +158,5 @@ def test_monitoring_fail_integration(self):
             patch_find.assert_called_once_with(syn=self.syn, syn_id="syn12345", days=15)
             patch_get_user.assert_called_once_with(self.syn, ["2222", "fooo"])
             patch_send.assert_called_once_with([111], "new subject",
-                                               returndf.to_html(index=False),
+                                               "syn2222, syn33333",
                                                contentType='text/html')