fix more tests

Author: TrevorBergeron

tests/system/small/test_dataframe.py

@@ -5324,9 +5324,12 @@ def test_df_to_latex(scalars_df_index, scalars_pandas_df_index):
 
 
 def test_df_to_json_local_str(scalars_df_index, scalars_pandas_df_index):
-    bf_result = scalars_df_index.to_json()
+    # pandas 3.0 bugged for serializing date col
+    bf_result = scalars_df_index.drop(columns="date_col").to_json()
     # default_handler for arrow types that have no default conversion
-    pd_result = scalars_pandas_df_index.to_json(default_handler=str)
+    pd_result = scalars_pandas_df_index.drop(columns="date_col").to_json(
+        default_handler=str
+    )
 
     assert bf_result == pd_result
 
@@ -6170,6 +6173,11 @@ def test_agg_with_dict_lists_strings(scalars_dfs):
     )
 
 
+@pytest.mark.skipif(
+    pandas.__version__.startswith("3"),
+    # See: https://github.com/python/cpython/issues/112282
+    reason="pandas 3.0 miscaculates variance",
+)
 def test_agg_with_dict_lists_callables(scalars_dfs):
     bf_df, pd_df = scalars_dfs
     agg_funcs = {

tests/system/small/test_dataframe_io.py

@@ -510,8 +510,9 @@ def test_to_csv_index(
     dtype = scalars_df.reset_index().dtypes.to_dict()
     dtype.pop("geography_col")
     dtype.pop("rowindex")
-    # read_csv will decode into bytes inproperly, convert_pandas_dtypes will encode properly from string
+    # read_csv will decode into bytes, numeric inproperly, convert_pandas_dtypes will encode properly from string
     dtype.pop("bytes_col")
+    dtype.pop("numeric_col")
     gcs_df = pd.read_csv(
         utils.get_first_file_from_wildcard(path),
         dtype=dtype,
@@ -548,8 +549,9 @@ def test_to_csv_tabs(
     dtype = scalars_df.reset_index().dtypes.to_dict()
     dtype.pop("geography_col")
     dtype.pop("rowindex")
-    # read_csv will decode into bytes inproperly, convert_pandas_dtypes will encode properly from string
-    # dtype.pop("bytes_col")
+    # read_csv will decode into bytes, numeric inproperly, convert_pandas_dtypes will encode properly from string
+    dtype.pop("bytes_col")
+    dtype.pop("numeric_col")
     gcs_df = pd.read_csv(
         utils.get_first_file_from_wildcard(path),
         sep="\t",

tests/system/small/test_pandas.py

@@ -541,7 +541,9 @@ def _convert_pandas_category(pd_s: pd.Series):
             f"Input must be a pandas Series with categorical data: {pd_s.dtype}"
         )
 
-    if pd.api.types.is_object_dtype(pd_s.cat.categories.dtype):
+    if pd.api.types.is_object_dtype(
+        pd_s.cat.categories.dtype
+    ) or pd.api.types.is_string_dtype(pd_s.cat.categories.dtype):
         return pd_s.astype(pd.StringDtype(storage="pyarrow"))
 
     if not isinstance(pd_s.cat.categories.dtype, pd.IntervalDtype):

tests/unit/test_dataframe_polars.py

@@ -4116,9 +4116,12 @@ def test_df_to_dict(scalars_df_index, scalars_pandas_df_index):
 
 
 def test_df_to_json_local_str(scalars_df_index, scalars_pandas_df_index):
-    bf_result = scalars_df_index.to_json()
+    # pandas 3.0 bugged for serializing date col
+    bf_result = scalars_df_index.drop(columns="date_col").to_json()
     # default_handler for arrow types that have no default conversion
-    pd_result = scalars_pandas_df_index.to_json(default_handler=str)
+    pd_result = scalars_pandas_df_index.drop(columns="date_col").to_json(
+        default_handler=str
+    )
 
     assert bf_result == pd_result