Better handling for dedicated schemas (#336)

* Better handling for dedicated schemas

* Unit tests, PR feedback

cumulus_library/actions/cleaner.py

@@ -147,7 +147,12 @@ def clean_study(
     if dedicated := manifest.get_dedicated_schema():
         view_table_list = [
             (
-                f"`{dedicated}`.`{x[0]}`",
+                # Athena uses different quoting strategies for drop view statements
+                # versus drop table statements. -_-
+                # TODO: Consider moving this logic to a database object?
+                f"`{dedicated}`.`{x[0]}`"
+                if (x[1] == "TABLE" and config.db.db_type == "athena")
+                else f'"{dedicated}"."{x[0]}"',
                 x[1],
             )
             for x in view_table_list

pyproject.toml

@@ -3,7 +3,7 @@ name = "cumulus-library"
 requires-python = ">= 3.11"
 dependencies = [
     "cumulus-fhir-support >= 1.2",
-    "duckdb >= 1.1",
+    "duckdb >= 1.1.3",
     "Jinja2 > 3",
     "pandas <3, >=2.1.3",
     "psmpy <1, >=0.3.13",

tests/test_actions.py

@@ -103,6 +103,30 @@ def test_clean_study(mock_db_config, verbose, prefix, confirm, stats, target, ra
                 assert ("study_valid__456",) not in remaining_tables
 
 
+def test_clean_dedicated_schema(mock_db_config):
+    with mock.patch.object(builtins, "input", lambda _: False):
+        mock_db_config.schema = "dedicated"
+        manifest = study_manifest.StudyManifest("./tests/test_data/study_dedicated_schema/")
+        mock_db_config.db.cursor().execute("CREATE SCHEMA dedicated")
+        builder.run_protected_table_builder(
+            config=mock_db_config,
+            manifest=manifest,
+        )
+        mock_db_config.db.cursor().execute("CREATE TABLE dedicated.table_1 (test int)")
+        mock_db_config.db.cursor().execute(
+            "CREATE VIEW dedicated.view_2 AS SELECT * FROM dedicated.table_1"
+        )
+        cleaner.clean_study(config=mock_db_config, manifest=manifest)
+        remaining_tables = (
+            mock_db_config.db.cursor()
+            .execute("select distinct(table_name) from information_schema.tables")
+            .fetchall()
+        )
+        assert (f"{enums.ProtectedTables.TRANSACTIONS.value}",) in remaining_tables
+        assert ("table_1",) not in remaining_tables
+        assert ("view_2",) not in remaining_tables
+
+
 def test_clean_throws_error_on_missing_params(mock_db_config):
     with pytest.raises(errors.CumulusLibraryError):
         cleaner.clean_study(config=mock_db_config, manifest=None)

tests/test_cli.py

@@ -822,7 +822,7 @@ def test_dedicated_schema(tmp_path):
     )
     for table in [
         ("dedicated", "table_1"),
-        ("dedicated", "table_2"),
+        ("dedicated", "view_2"),
         ("dedicated", "table_raw_sql"),
         ("main", "core__condition"),
     ]:

tests/test_data/study_dedicated_schema/module2.py

@@ -6,5 +6,6 @@ class ModuleTwoRunner(cumulus_library.BaseTableBuilder):
 
     def prepare_queries(self, *args, **kwargs):
         self.queries.append(
-            "CREATE TABLE IF NOT EXISTS study_dedicated_schema__table_2 (test int);"
+            """CREATE VIEW IF NOT EXISTS study_dedicated_schema__view_2 AS 
+            SELECT * FROM dedicated.table_1;"""
         )