Fix CI: pin ruff, make type inference tests pandas 3 compatible

.github/workflows/lint.yml

@@ -12,4 +12,4 @@ jobs:
     - uses: actions/checkout@v4
 
     - name: Run ruff
-      run: pipx run ruff check --output-format=github explorer
+      run: pipx run ruff==0.16.8 check --output-format=github explorer

explorer/__init__.py

@@ -9,15 +9,11 @@
 
 def get_version(short=False):
     assert __version_info__["releaselevel"] in ("alpha", "beta", "final")
-    vers = ["%(major)i.%(minor)i" % __version_info__, ]
+    vers = [f"{__version_info__['major']}.{__version_info__['minor']}", ]
     if __version_info__["patch"]:
-        vers.append(".%(patch)i" % __version_info__)
+        vers.append(f".{__version_info__['patch']}")
     if __version_info__["releaselevel"] != "final" and not short:
-        vers.append(
-            "%s%i" % (
-                __version_info__["releaselevel"][0],
-                __version_info__["serial"])
-        )
+        vers.append(f"{__version_info__['releaselevel'][0]}{__version_info__['serial']}")
     return "".join(vers)
 
 

explorer/tests/factories.py

@@ -12,7 +12,7 @@ class UserFactory(DjangoModelFactory):
     class Meta:
         model = settings.AUTH_USER_MODEL
 
-    username = Sequence(lambda n: "User %03d" % n)
+    username = Sequence(lambda n: f"User {n:03d}")
     is_staff = True
 
 

explorer/tests/test_type_infer.py

@@ -18,6 +18,11 @@ def _get_csv(csv_name):
     return csv_bytes
 
 
+def _is_text_dtype(series):
+    # pandas < 3 stores text as object; pandas 3 uses a dedicated string dtype
+    return pd.api.types.is_object_dtype(series) or pd.api.types.is_string_dtype(series)
+
+
 def _get_json(json_name):
     current_script_dir = os.path.dirname(os.path.abspath(__file__))
     file_path = os.path.join(current_script_dir, "json", json_name)
@@ -34,16 +39,16 @@ class TestCsvToTypedDf(TestCase):
 
     def test_mixed_types(self):
         df = csv_to_typed_df(_get_csv("mixed.csv"))
-        self.assertTrue(pd.api.types.is_object_dtype(df["Value1"]))
-        self.assertTrue(pd.api.types.is_object_dtype(df["Value2"]))
-        self.assertTrue(pd.api.types.is_object_dtype(df["Value3"]))
+        self.assertTrue(_is_text_dtype(df["Value1"]))
+        self.assertTrue(_is_text_dtype(df["Value2"]))
+        self.assertTrue(_is_text_dtype(df["Value3"]))
 
     def test_all_types(self):
         df = csv_to_typed_df(_get_csv("all_types.csv"))
-        self.assertTrue(pd.api.types.is_datetime64_ns_dtype(df["Dates"]))
+        self.assertTrue(pd.api.types.is_datetime64_any_dtype(df["Dates"]))
         self.assertTrue(pd.api.types.is_integer_dtype(df["Integers"]))
         self.assertTrue(pd.api.types.is_float_dtype(df["Floats"]))
-        self.assertTrue(pd.api.types.is_object_dtype(df["Strings"]))
+        self.assertTrue(_is_text_dtype(df["Strings"]))
 
     def test_integer_parsing(self):
         df = csv_to_typed_df(_get_csv("integers.csv"))
@@ -62,27 +67,27 @@ def test_date_parsing(self):
         # Day of Year: 2024-024 (Year-DayOfYear)
 
         df = csv_to_typed_df(_get_csv("dates.csv"))
-        self.assertTrue(pd.api.types.is_datetime64_ns_dtype(df["Dates"]))
+        self.assertTrue(pd.api.types.is_datetime64_any_dtype(df["Dates"]))
 
 
 @skipIf(not EXPLORER_USER_UPLOADS_ENABLED, "User uploads not enabled")
 class TestJsonToTypedDf(TestCase):
 
     def test_basic_json(self):
         df = json_to_typed_df(_get_json("kings.json"))
-        self.assertTrue(pd.api.types.is_object_dtype(df["Name"]))
-        self.assertTrue(pd.api.types.is_object_dtype(df["Country"]))
+        self.assertTrue(_is_text_dtype(df["Name"]))
+        self.assertTrue(_is_text_dtype(df["Country"]))
         self.assertTrue(pd.api.types.is_integer_dtype(df["ID"]))
 
     def test_nested_json(self):
         df = json_to_typed_df(_get_json("github.json"))
-        self.assertTrue(pd.api.types.is_object_dtype(df["subscription_url"]))
-        self.assertTrue(pd.api.types.is_object_dtype(df["topics"]))
+        self.assertTrue(_is_text_dtype(df["subscription_url"]))
+        self.assertTrue(_is_text_dtype(df["topics"]))
         self.assertTrue(pd.api.types.is_integer_dtype(df["size"]))
         self.assertTrue(pd.api.types.is_integer_dtype(df["owner.id"]))
 
     def test_json_list(self):
         df = json_list_to_typed_df(_get_json("list.json"))
         self.assertTrue(pd.api.types.is_integer_dtype(df["Item.value.M.unique_connection_count.N"]))
-        self.assertTrue(pd.api.types.is_object_dtype(df["Item.instanceId.S"]))
+        self.assertTrue(_is_text_dtype(df["Item.instanceId.S"]))
         self.assertEqual(len(df), 5)

explorer/views/utils.py

@@ -9,7 +9,7 @@
 logger = logging.getLogger(__name__)
 
 
-def query_viewmodel(request, query, title=None, form=None, message=None,
+def query_viewmodel(request, query, title=None, form=None, message=None,  # noqa: PLR0917
                     run_query=True, error=None,
                     rows=app_settings.EXPLORER_DEFAULT_ROWS):
     """

ruff.toml

@@ -1,4 +1,6 @@
 line-length = 120
+# Keep in sync with python_requires in setup.py; ruff can't read it from there.
+target-version = "py38"
 
 extend-exclude = [
   ".ruff_cache",