Bugfix/check mariadb version

.github/workflows/ci.yml

@@ -25,7 +25,7 @@ jobs:
     strategy:
       matrix:
         python-version: [3.8]
-        backend: ["sqlite", "mariadb"]
+        backend: ["mariadb", "sqlite"]
         mariadb-version: ["10.4"]
         include:
           - python-version: "3.10"  # Future ubuntu 22.04 upgrade.
@@ -53,10 +53,14 @@ jobs:
     env:
       PYTEST_ADDOPTS: "--maxfail=20"  # Stop testing after too many failures.
       # Conditionally set the database url based on the backend.
-      DATABASE_URL: ${{ matrix.backend == 'sqlite' && 'sqlite:///db.sqlite3' || 'mysql://root:[email protected]:3306/mysql' }}
+      DATABASE_URL: ${{ matrix.backend == 'sqlite' && 'sqlite:///db.sqlite3' || 'mysql://root:[email protected]:3306/test_db?charset=utf8mb4' }}
 
     steps:
 
+      - name: Print db options
+        run: |
+          mysql --verbose --help
+
       - name: Checkout Code Repository
         uses: actions/checkout@v4
 

config/settings/base.py

@@ -44,6 +44,17 @@
 DATABASES = {
     "default": env.db("DATABASE_URL", default="sqlite:///primed.db"),
 }
+
+if DATABASES["default"]["ENGINE"] == "django.db.backends.mysql":
+    DATABASES["default"]["TEST"] = {
+        "CHARSET": "utf8mb4",
+        "COLLATION": "utf8mb4_general_ci",  # or other collation
+    }
+    DATABASES["default"]["OPTIONS"] = {
+        "charset": "utf8mb4",  # or other charset
+        "collation": "utf8mb4_general_ci",  # or other collation
+    }
+
 # DATABASES["default"]["ATOMIC_REQUESTS"] = True
 # # https://docs.djangoproject.com/en/stable/ref/settings/#std:setting-DEFAULT_AUTO_FIELD
 DEFAULT_AUTO_FIELD = "django.db.models.BigAutoField"

primed/primed_anvil/tests/test_helpers.py

@@ -789,3 +789,74 @@ def test_multiple_workspace_types_separate_studies(self):
         self.assertEqual(res["test-bp-cdsa/test-ws-cdsa"], "TEST 2")
         self.assertIn("test-bp-open/test-ws-open", res)
         self.assertEqual(res["test-bp-open/test-ws-open"], "TEST 3")
+
+    def test_non_consecutive_grouping(self):
+        """Studies are grouped even if workspaces are listed non-consecutively."""
+        # This replicates an issue seen in prod:
+        # 1) there are multiple workspaces for the same set of studies
+        # 2) objects are created in a specific order with specific alphabetizing.
+        # The difference in behavior between sqlite and mariadb is likely due to different ordering
+        # when the queryset results are returned, so debugging is tricky.
+        study_1 = StudyFactory.create(short_name="TEST_2")
+        study_2 = StudyFactory.create(short_name="TEST_1")
+        study_accession_1 = dbGaPStudyAccessionFactory.create(dbgap_phs=964, studies=[study_1, study_2])
+        study_accession_2 = dbGaPStudyAccessionFactory.create(dbgap_phs=286, studies=[study_2])
+        # Two workspaces associated with study_accession_1 (with two studies)
+        workspace_1_a = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-b",
+            workspace__name="test-a-b_c3",
+            dbgap_study_accession=study_accession_1,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_1_a.workspace, group=self.primed_all_group)
+        workspace_2_a = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-a",
+            workspace__name="test-a_c3",
+            dbgap_study_accession=study_accession_2,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_2_a.workspace, group=self.primed_all_group)
+        workspace_2_b = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-a",
+            workspace__name="test-a_c4",
+            dbgap_study_accession=study_accession_2,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_2_b.workspace, group=self.primed_all_group)
+        workspace_2_c = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-a",
+            workspace__name="test-a_c2",
+            dbgap_study_accession=study_accession_2,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_2_c.workspace, group=self.primed_all_group)
+        workspace_2_d = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-a",
+            workspace__name="test-a_c1",
+            dbgap_study_accession=study_accession_2,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_2_d.workspace, group=self.primed_all_group)
+        workspace_1_b = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-b",
+            workspace__name="test-a-b_c4",
+            dbgap_study_accession=study_accession_1,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_1_b.workspace, group=self.primed_all_group)
+        workspace_1_c = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-b",
+            workspace__name="test-a-b_c1",
+            dbgap_study_accession=study_accession_1,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_1_c.workspace, group=self.primed_all_group)
+        workspace_1_d = dbGaPWorkspaceFactory.create(
+            workspace__billing_project__name="test-b",
+            workspace__name="test-a-b_c2",
+            dbgap_study_accession=study_accession_1,
+        )
+        WorkspaceGroupSharingFactory.create(workspace=workspace_1_d.workspace, group=self.primed_all_group)
+        res = helpers.get_workspaces_for_phenotype_inventory()
+        self.assertEqual(len(res), 8)
+        self.assertEqual(res["test-b/test-a-b_c1"], "TEST_1, TEST_2")
+        self.assertEqual(res["test-b/test-a-b_c2"], "TEST_1, TEST_2")
+        self.assertEqual(res["test-b/test-a-b_c3"], "TEST_1, TEST_2")
+        self.assertEqual(res["test-b/test-a-b_c4"], "TEST_1, TEST_2")
+        self.assertEqual(res["test-a/test-a_c1"], "TEST_1")
+        self.assertEqual(res["test-a/test-a_c2"], "TEST_1")
+        self.assertEqual(res["test-a/test-a_c3"], "TEST_1")
+        self.assertEqual(res["test-a/test-a_c4"], "TEST_1")

primed/tests/test_db_config.py

@@ -0,0 +1,42 @@
+# test_db_settings.py
+import pytest
+from django.conf import settings
+from django.db import connection
+
+
+@pytest.mark.django_db
+def test_character_set_and_storage_engine():
+    with connection.cursor() as cursor:
+        if settings.DATABASES["default"]["ENGINE"] == "django.db.backends.mysql":
+            print("Database options: {}".format(settings.DATABASES["default"].get("OPTIONS")))
+            # For MariaDB/MySQL
+            cursor.execute("SHOW VARIABLES LIKE 'character_set_server';")
+            charset = cursor.fetchone()
+            print(f"Character set: {charset[1]}")
+
+            cursor.execute("SHOW VARIABLES LIKE 'collation_server';")
+            collation = cursor.fetchone()
+            print(f"Collation: {collation[1]}")
+
+            cursor.execute("SHOW TABLE STATUS;")
+            tables = cursor.fetchall()
+            for table in tables:
+                print(f"Table: {table[0]}, Engine: {table[1]}")
+                assert table[1] == "InnoDB"
+
+            assert charset[1] == "utf8mb4"
+            assert collation[1] == "utf8mb4_general_ci"
+
+        elif settings.DATABASES["default"]["ENGINE"] == "django.db.backends.sqlite3":
+            # For SQLite
+            cursor.execute("PRAGMA encoding;")
+            charset = cursor.fetchone()
+            print(f"Character set: {charset[0]}")
+
+            cursor.execute("SELECT name FROM sqlite_master WHERE type='table';")
+            tables = cursor.fetchall()
+            for table in tables:
+                print(f"Table: {table[0]}, Engine: SQLite")
+
+            # SQLite always uses UTF-8 encoding
+            assert charset[0].lower() == "utf-8"