chore(python): Use Dataframe.item internally and in tests (#6109)

py-polars/polars/internals/series/series.py

@@ -1283,7 +1283,7 @@ def nan_max(self) -> int | float | date | datetime | timedelta | str:
         whereas polars defaults to ignoring them.
 
         """
-        return self.to_frame().select(pli.col(self.name).nan_max())[0, 0]
+        return self.to_frame().select(pli.col(self.name).nan_max()).item()
 
     def nan_min(self) -> int | float | date | datetime | timedelta | str:
         """
@@ -1293,7 +1293,7 @@ def nan_min(self) -> int | float | date | datetime | timedelta | str:
         whereas polars defaults to ignoring them.
 
         """
-        return self.to_frame().select(pli.col(self.name).nan_min())[0, 0]
+        return self.to_frame().select(pli.col(self.name).nan_min()).item()
 
     def std(self, ddof: int = 1) -> float | None:
         """
@@ -2095,7 +2095,7 @@ def search_sorted(
 
         """
         if isinstance(element, (int, float)):
-            return pli.select(pli.lit(self).search_sorted(element, side))[0, 0]
+            return pli.select(pli.lit(self).search_sorted(element, side)).item()
         element = Series(element)
         return pli.select(pli.lit(self).search_sorted(element, side)).to_series()
 

py-polars/tests/unit/test_datelike.py

@@ -1771,12 +1771,12 @@ def test_from_dict_tu_consistency() -> None:
 
 def test_date_parse_omit_day() -> None:
     df = pl.DataFrame({"month": ["2022-01"]})
-    assert df.select(pl.col("month").str.strptime(pl.Date, fmt="%Y-%m"))[0, 0] == date(
+    assert df.select(pl.col("month").str.strptime(pl.Date, fmt="%Y-%m")).item() == date(
         2022, 1, 1
     )
-    assert df.select(pl.col("month").str.strptime(pl.Datetime, fmt="%Y-%m"))[
-        0, 0
-    ] == datetime(2022, 1, 1)
+    assert df.select(
+        pl.col("month").str.strptime(pl.Datetime, fmt="%Y-%m")
+    ).item() == datetime(2022, 1, 1)
 
 
 def test_shift_and_fill_group_logicals() -> None:

py-polars/tests/unit/test_exprs.py

@@ -90,7 +90,7 @@ def test_count_expr() -> None:
 
     out = df.select(pl.count())
     assert out.shape == (1, 1)
-    assert cast(int, out[0, 0]) == 5
+    assert cast(int, out.item()) == 5
 
     out = df.groupby("b", maintain_order=True).agg(pl.count())
     assert out["b"].to_list() == ["a", "b"]

py-polars/tests/unit/test_functions.py

@@ -139,9 +139,9 @@ def test_null_handling_correlation() -> None:
     df1 = pl.DataFrame({"a": [None, 1, 2], "b": [None, 2, 1]})
     df2 = pl.DataFrame({"a": [np.nan, 1, 2], "b": [np.nan, 2, 1]})
 
-    assert np.isclose(df1.select(pl.spearman_rank_corr("a", "b"))[0, 0], -1.0)
+    assert np.isclose(df1.select(pl.spearman_rank_corr("a", "b")).item(), -1.0)
     assert (
-        str(df2.select(pl.spearman_rank_corr("a", "b", propagate_nans=True))[0, 0])
+        str(df2.select(pl.spearman_rank_corr("a", "b", propagate_nans=True)).item())
         == "nan"
     )
 

py-polars/tests/unit/test_lazy.py

@@ -501,7 +501,7 @@ def test_is_finite_is_infinite() -> None:
 
 def test_len() -> None:
     df = pl.DataFrame({"nrs": [1, 2, 3]})
-    assert cast(int, df.select(col("nrs").len())[0, 0]) == 3
+    assert cast(int, df.select(col("nrs").len()).item()) == 3
 
 
 def test_cum_agg() -> None:
@@ -534,7 +534,7 @@ def test_round() -> None:
 
 def test_dot() -> None:
     df = pl.DataFrame({"a": [1.8, 1.2, 3.0], "b": [3.2, 1, 2]})
-    assert cast(float, df.select(pl.col("a").dot(pl.col("b")))[0, 0]) == 12.96
+    assert cast(float, df.select(pl.col("a").dot(pl.col("b"))).item()) == 12.96
 
 
 def test_sort() -> None:
@@ -1055,16 +1055,16 @@ def test_join_suffix() -> None:
 def test_str_concat() -> None:
     df = pl.DataFrame({"foo": [1, None, 2]})
     df = df.select(pl.col("foo").str.concat("-"))
-    assert cast(str, df[0, 0]) == "1-null-2"
+    assert cast(str, df.item()) == "1-null-2"
 
 
 @pytest.mark.parametrize("no_optimization", [False, True])
 def test_collect_all(df: pl.DataFrame, no_optimization: bool) -> None:
     lf1 = df.lazy().select(pl.col("int").sum())
     lf2 = df.lazy().select((pl.col("floats") * 2).sum())
     out = pl.collect_all([lf1, lf2], no_optimization=no_optimization)
-    assert cast(int, out[0][0, 0]) == 6
-    assert cast(float, out[1][0, 0]) == 12.0
+    assert cast(int, out[0].item()) == 6
+    assert cast(float, out[1].item()) == 12.0
 
 
 def test_spearman_corr() -> None:
@@ -1120,9 +1120,9 @@ def test_pearson_corr() -> None:
 
 
 def test_cov(fruits_cars: pl.DataFrame) -> None:
-    assert cast(float, fruits_cars.select(pl.cov("A", "B"))[0, 0]) == -2.5
+    assert cast(float, fruits_cars.select(pl.cov("A", "B")).item()) == -2.5
     assert (
-        cast(float, fruits_cars.select(pl.cov(pl.col("A"), pl.col("B")))[0, 0]) == -2.5
+        cast(float, fruits_cars.select(pl.cov(pl.col("A"), pl.col("B"))).item()) == -2.5
     )
 
 

py-polars/tests/unit/test_schema.py

@@ -69,7 +69,8 @@ def test_bool_numeric_supertype() -> None:
         pl.Int64,
     ]:
         assert (
-            df.select([(pl.col("v") < 3).sum().cast(dt) / pl.count()])[0, 0] - 0.3333333
+            df.select([(pl.col("v") < 3).sum().cast(dt) / pl.count()]).item()
+            - 0.3333333
             <= 0.00001
         )