Merge remote-tracking branch 'upstream/main' into oh-nodes

Author: dangotbanned

pyproject.toml

@@ -383,9 +383,8 @@ exclude_also = [
 files = ["src/narwhals", "tests", "test-plugin"]
 pretty = true
 strict = true
+allow_redefinition = true  # https://mypy.readthedocs.io/en/stable/command_line.html#cmdoption-mypy-allow-redefinition
 check_untyped_defs = true
-allow_redefinition_new = true
-local_partial_types = true
 
 [[tool.mypy.overrides]]
 module = [

src/narwhals/_arrow/series.py

@@ -600,13 +600,10 @@ def value_counts(
 
         val_counts = pc.value_counts(self.native)
         values = val_counts.field("values")
-        counts = cast("ChunkedArrayAny", val_counts.field("counts"))
-
+        counts = val_counts.field("counts")
         if normalize:
-            arrays = [values, pc.divide(*cast_for_truediv(counts, pc.sum(counts)))]
-        else:
-            arrays = [values, counts]
-
+            counts = pc.divide(*cast_for_truediv(counts, pc.sum(counts)))
+        arrays = values, counts
         val_count = pa.Table.from_arrays(arrays, names=[index_name_, value_name_])
 
         if sort:

src/narwhals/_pandas_like/utils.py

@@ -195,7 +195,7 @@ def rename(
         result = obj.rename(*args, **kwargs, copy=False, inplace=False)
     else:
         result = obj.rename(*args, **kwargs)
-    return cast("NativeNDFrameT", result)  # type: ignore[redundant-cast]
+    return cast("NativeNDFrameT", result)
 
 
 @functools.lru_cache(maxsize=16)

src/narwhals/_polars/namespace.py

@@ -186,7 +186,7 @@ def concat_str(
                     pl.when(~nm).then(sep).otherwise(pl.lit("")) for nm in null_mask[:-1]
                 ]
 
-                result = pl.fold(  # type: ignore[assignment]
+                result = pl.fold(
                     acc=init_value,
                     function=operator.add,
                     exprs=[s + v for s, v in zip(separators, values, strict=True)],

src/narwhals/dtypes.py

@@ -970,12 +970,18 @@ def __hash__(self) -> int:
 
     def __repr__(self) -> str:
         # Get leaf type
-        dtype_ = self
+        dtype: Self | IntoDType = self
+        tp_self = type(self)
+        # NOTE: This might look odd, but if we used a `while` loop
+        # it won't trigger a `RecursionError` like `for` does.
+        # https://github.com/narwhals-dev/narwhals/pull/3651
         for _ in self.shape:
-            dtype_ = dtype_.inner  # type: ignore[assignment]
-
-        class_name = self.__class__.__name__
-        return f"{class_name}({dtype_!r}, shape={self.shape})"
+            if isinstance(dtype, tp_self):
+                dtype = dtype.inner
+            else:  # pragma: no cover
+                msg = "Either something has gone horribly wrong in narwhals, or you've fiddled with the internals!"
+                raise TypeError(msg)
+        return f"{tp_self.__name__}({dtype!r}, shape={self.shape})"
 
 
 class Date(TemporalType):

tests/dtypes/dtypes_test.py

@@ -180,6 +180,25 @@ def test_2d_array(constructor: Constructor, request: pytest.FixtureRequest) -> N
     assert df.collect_schema()["a"] == nw.Array(nw.Array(nw.Int64(), 2), 3)
 
 
+def test_array_inner_recursive() -> None:
+    # https://github.com/pola-rs/polars/blob/210b34bee60db71ece147db060f570a41ba63c5d/py-polars/tests/unit/datatypes/test_array.py#L360-L365
+    shape = (2, 3, 4, 5)
+    dtype = nw.Array(nw.Int64, shape=shape)
+    for dim in shape:
+        assert isinstance(dtype, nw.Array)
+        assert dtype.size == dim
+        dtype = dtype.inner
+
+
+def test_array_inner_recursive_repr() -> None:
+    dtype_1 = nw.Array(nw.Int64, 2)
+    dtype_2 = nw.Array(dtype_1, 1)
+    dtype_3 = nw.Array(dtype_2, 2)
+    dtype_1.inner = dtype_3.inner
+    with pytest.raises(RecursionError):
+        repr(dtype_3)
+
+
 def test_second_time_unit() -> None:
     pytest.importorskip("pandas")
     pytest.importorskip("pyarrow")

tests/frame/group_by_test.py

@@ -81,7 +81,7 @@ def test_group_by_iter(constructor_eager: ConstructorEager) -> None:
             assert isinstance(sub_df, nw.DataFrame)
         keys.append(key)
     assert sorted(keys) == sorted(expected_keys)
-    expected_keys = [(1, 4), (3, 6)]  # type: ignore[list-item]
+    expected_keys = [(1, 4), (3, 6)]
     keys = []
     for key, _ in df.group_by("a", "b"):
         keys.append(key)

tests/frame/reindex_test.py

@@ -27,7 +27,7 @@ def test_reindex(df_raw: Any) -> None:
     assert result_s[1]
     assert not result_s[2]
     result = df.with_columns(s.sort())
-    expected = {"a": [1, 2, 3], "b": [4, 4, 6], "z": [7.0, 8.0, 9.0]}  # type: ignore[list-item]
+    expected = {"a": [1, 2, 3], "b": [4, 4, 6], "z": [7.0, 8.0, 9.0]}
     assert_equal_data(result, expected)
     with pytest.raises(MultiOutputExpressionError):
         nw.to_native(df.with_columns(nw.all() + nw.all()))

tests/series_only/value_counts_test.py

@@ -29,7 +29,7 @@ def test_value_counts(
     expected_index = [4, 1, 6]
 
     if normalize:
-        expected_count = [v / len(data) for v in expected_count]  # type: ignore[misc]
+        expected_count = [v / len(data) for v in expected_count]
 
     expected_name = name or ("proportion" if normalize else "count")
     expected = {"a": expected_index, expected_name: expected_count}