MAINT: Fix nightly by skipping and noting problematic code (#301)

* MAINT: Fix nightly

* MAINT: Add skips for problematic tests on nightly

Catch warnings after 1.5.0 is out

* CLN: Refactor to simplify

Author: bashtage

tests/__init__.py

@@ -5,9 +5,13 @@
     Final,
 )
 
+from packaging.version import parse
+import pandas as pd
+
 from pandas._typing import T
 
 TYPE_CHECKING_INVALID_USAGE: Final = TYPE_CHECKING
+PD_LT_15 = parse(pd.__version__) < parse("1.5.0")
 
 
 def check(actual: T, klass: type, dtype: type | None = None, attr: str = "left") -> T:

tests/test_errors.py

@@ -3,12 +3,10 @@
 from typing import TYPE_CHECKING
 import warnings
 
-from packaging.version import parse
-import pandas as pd
 from pandas import errors
 import pytest
 
-PD_LT_15 = parse(pd.__version__) < parse("1.5.0")
+from tests import PD_LT_15
 
 if TYPE_CHECKING or PD_LT_15:
     # TODO: Remove all imports below after switch to 1.5.x, these moved to pandas.errors

tests/test_frame.py

@@ -1625,7 +1625,10 @@ def sample_to_df(x: pd.DataFrame) -> pd.DataFrame:
         return x.sample()
 
     check(
-        assert_type(df.groupby("col1").apply(sample_to_df), pd.DataFrame), pd.DataFrame
+        assert_type(
+            df.groupby("col1", group_keys=False).apply(sample_to_df), pd.DataFrame
+        ),
+        pd.DataFrame,
     )
 
 

tests/test_io.py

@@ -14,13 +14,11 @@
 )
 
 import numpy as np
-from packaging.version import parse
 import pandas as pd
 from pandas import (
     DataFrame,
     HDFStore,
     Series,
-    __version__,
     read_clipboard,
     read_csv,
     read_excel,
@@ -46,7 +44,10 @@
 import sqlalchemy
 from typing_extensions import assert_type
 
-from tests import check
+from tests import (
+    PD_LT_15,
+    check,
+)
 
 from pandas.io.api import to_pickle
 from pandas.io.clipboard import PyperclipException
@@ -64,8 +65,6 @@
 DF = DataFrame({"a": [1, 2, 3], "b": [0.0, 0.0, 0.0]})
 CWD = os.path.split(os.path.abspath(__file__))[0]
 
-PD_LT_15 = parse(__version__) < parse("1.5.0")
-
 
 @pytest.mark.skipif(PD_LT_15, reason="pandas 1.5.0 or later required")
 def test_orc():
@@ -188,6 +187,8 @@ def test_read_stata_df():
         check(assert_type(read_stata(path), pd.DataFrame), pd.DataFrame)
 
 
+# Remove test when pandas 1.5.0 is released
+@pytest.mark.skipif(not PD_LT_15, reason="Keyword only in 1.5.0")
 def test_read_stata_iterator_positional():
     with ensure_clean() as path:
         str_path = str(path)
@@ -657,7 +658,9 @@ def test_excel_writer():
         with pd.ExcelWriter(path) as ew:
             check(assert_type(ew, pd.ExcelWriter), pd.ExcelWriter)
             DF.to_excel(ew, sheet_name="A")
-            check(assert_type(ew.handles, IOHandles[bytes]), IOHandles)
+            if PD_LT_15:
+                # Remove after 1.5 and remove handles from ExcelWriter
+                check(assert_type(ew.handles, IOHandles[bytes]), IOHandles)
         check(assert_type(read_excel(path, sheet_name="A"), DataFrame), DataFrame)
         check(assert_type(read_excel(path), DataFrame), DataFrame)
         ef = pd.ExcelFile(path)

tests/test_plotting.py

@@ -536,7 +536,6 @@ def test_plot_keywords(close_figures):
                 colormap="jet",
                 table=True,
                 stacked=True,
-                sort_columns=True,
                 secondary_y=True,
                 mark_right=True,
                 include_bool=True,

tests/test_resampler.py

@@ -16,11 +16,15 @@
 )
 from pandas.core.groupby.generic import SeriesGroupBy
 from pandas.core.resample import Resampler
+import pytest
 from typing_extensions import assert_type
 
 from pandas._typing import Scalar
 
-from tests import check
+from tests import (
+    PD_LT_15,
+    check,
+)
 
 DR = date_range("1999-1-1", periods=365, freq="D")
 DF_ = DataFrame(np.random.standard_normal((365, 1)), index=DR)
@@ -297,7 +301,11 @@ def s2scalar(val: Series) -> float:
 
     check(S.resample("m").aggregate(np.sum), Series)
     check(S.resample("m").aggregate("sum"), Series)
-    check(S.resample("m").aggregate(s2series), Series)
+    if PD_LT_15:
+        check(S.resample("m").aggregate(s2series), Series)
+    else:
+        with pytest.warns(FutureWarning, match="Not prepending group keys"):
+            check(S.resample("m").aggregate(s2series), Series)
     check(S.resample("m").aggregate(s2scalar), Series)
     check(S.resample("m").aggregate([np.mean]), DataFrame)
     check(S.resample("m").aggregate(["sum", np.mean]), DataFrame)
@@ -317,7 +325,11 @@ def df2scalar(val: DataFrame) -> float:
 
     check(DF.resample("m").aggregate(np.sum), DataFrame)
     check(DF.resample("m").aggregate("sum"), DataFrame)
-    check(DF.resample("m").aggregate(df2frame), DataFrame)
+    if PD_LT_15:
+        check(DF.resample("m").aggregate(df2frame), DataFrame)
+    else:
+        with pytest.warns(FutureWarning, match="Not prepending group keys"):
+            check(DF.resample("m").aggregate(df2frame), DataFrame)
     check(DF.resample("m").aggregate(df2series), DataFrame)
     check(DF.resample("m").aggregate(df2scalar), DataFrame)
     check(DF.resample("m").aggregate([np.mean]), DataFrame)

tests/test_series.py

@@ -34,6 +34,7 @@
 )
 
 from tests import (
+    PD_LT_15,
     TYPE_CHECKING_INVALID_USAGE,
     check,
 )
@@ -259,7 +260,11 @@ def test_types_rank() -> None:
         s.rank(method="min", pct=True)
     with pytest.warns(FutureWarning, match="Dropping of nuisance columns"):
         s.rank(method="dense", ascending=True)
-    s.rank(method="first", numeric_only=True)
+    if PD_LT_15:
+        s.rank(method="first", numeric_only=True)
+    else:
+        with pytest.warns(FutureWarning, match="Calling Series.rank with numeric_only"):
+            s.rank(method="first", numeric_only=True)
 
 
 def test_types_mean() -> None: