Merge branch 'main' into perf/codspeed-ci

Author: FBruzzesi

.github/release-drafter.yml

@@ -9,31 +9,31 @@ autolabeler:
   - label: breaking
     title:
       # Example: feat!: ...
-      - '/^(build|chore|ci|depr|docs|feat|fix|perf|refactor|release|test)(\(.*\))?\!\: /'
+      - '/^([B|b]uild|[C|c]hore|[CI|ci]|[D|d]epr|[D|d]oc|DOC|[F|f]eat|[F|f]ix|[P|p]erf|[R|r]efactor|[R|r]elease|[T|t]est)\! /'
   - label: build
     title:
-      - '/^build/'
+      - '/^[B|b]uild/'
   - label: internal
     title:
-      - '/^(chore|ci|refactor|test|template|bench)/'
+      - '/^[C|c]hore|[CI|ci]|[R|r]efactor|[T|t]est|[T|t]emplate|[B|b]ench/'
   - label: deprecation
     title:
-      - '/^depr/'
+      - '/^[D|d]epr/'
   - label: documentation
     title:
-      - '/(.*doc|.*docstring)/'
+      - '/^[D|d]oc|DOC/'
   - label: enhancement
     title:
-      - '/^(.*feat|.*enh)/'
+      - '/^.*feat|.*enh|Feat|ENH|Enh/'
   - label: fix
     title:
-      - '/^fix/'
+      - '/^[F|f]ix/'
   - label: performance
     title:
-      - '/^perf/'
+      - '/^[P|p]erf/'
   - label: release
     title:
-      - '/^release/'
+      - '/^[R|r]elease/'
 
 version-resolver:
   major:

narwhals/_dask/expr.py

@@ -545,6 +545,40 @@ def func(_input: Any) -> Any:
             returns_scalar=False,
         )
 
+    def is_duplicated(self: Self) -> Self:
+        def func(_input: Any) -> Any:
+            _name = _input.name
+            return (
+                _input.to_frame().groupby(_name).transform("size", meta=(_name, int)) > 1
+            )
+
+        return self._from_call(
+            func,
+            "is_duplicated",
+            returns_scalar=False,
+        )
+
+    def is_unique(self: Self) -> Self:
+        def func(_input: Any) -> Any:
+            _name = _input.name
+            return (
+                _input.to_frame().groupby(_name).transform("size", meta=(_name, int)) == 1
+            )
+
+        return self._from_call(
+            func,
+            "is_unique",
+            returns_scalar=False,
+        )
+
+    def is_in(self: Self, other: Any) -> Self:
+        return self._from_call(
+            lambda _input, other: _input.isin(other),
+            "is_in",
+            other,
+            returns_scalar=False,
+        )
+
     def null_count(self: Self) -> Self:
         return self._from_call(
             lambda _input: _input.isna().sum(),

narwhals/_pandas_like/series.py

@@ -221,18 +221,9 @@ def is_in(self, other: Any) -> PandasLikeSeries:
         return self._from_native_series(res)
 
     def arg_true(self) -> PandasLikeSeries:
-        import numpy as np  # ignore-banned-import
-
         ser = self._native_series
-        res = np.flatnonzero(ser)
-        return self._from_native_series(
-            native_series_from_iterable(
-                res,
-                name=self.name,
-                index=range(len(res)),
-                implementation=self._implementation,
-            )
-        )
+        result = ser.__class__(range(len(ser)), name=ser.name, index=ser.index).loc[ser]
+        return self._from_native_series(result)
 
     # Binary comparisons
 

narwhals/_pandas_like/utils.py

@@ -197,6 +197,10 @@ def set_axis(
     implementation: Implementation,
     backend_version: tuple[int, ...],
 ) -> T:
+    if implementation is Implementation.CUDF:  # pragma: no cover
+        obj = obj.copy(deep=False)  # type: ignore[attr-defined]
+        obj.index = index  # type: ignore[attr-defined]
+        return obj
     if implementation is Implementation.PANDAS and backend_version < (
         1,
     ):  # pragma: no cover
@@ -210,7 +214,7 @@ def set_axis(
         kwargs["copy"] = False
     else:  # pragma: no cover
         pass
-    return obj.set_axis(index, axis=0, **kwargs)  # type: ignore[no-any-return, attr-defined]
+    return obj.set_axis(index, axis=0, **kwargs)  # type: ignore[attr-defined, no-any-return]
 
 
 def translate_dtype(column: Any) -> DType:

narwhals/expr.py

@@ -1114,8 +1114,8 @@ def arg_true(self) -> Self:
 
             >>> func(df_pd)
                a
-            0  1
-            1  2
+            1  1
+            2  2
             >>> func(df_pl)
             shape: (2, 1)
             ┌─────┐

narwhals/functions.py

@@ -133,7 +133,7 @@ def new_series(
         raise NotImplementedError(msg)
     else:  # pragma: no cover
         try:
-            # implementation is UNKNOWN, Narhwals extension using this feature should
+            # implementation is UNKNOWN, Narwhals extension using this feature should
             # implement `from_dict` function in the top-level namespace.
             native_series = native_namespace.new_series(name, values, dtype)
         except AttributeError as e:
@@ -256,7 +256,7 @@ def from_dict(
         native_frame = native_namespace.table(data, schema=schema)
     else:  # pragma: no cover
         try:
-            # implementation is UNKNOWN, Narhwals extension using this feature should
+            # implementation is UNKNOWN, Narwhals extension using this feature should
             # implement `from_dict` function in the top-level namespace.
             native_frame = native_namespace.from_dict(data)
         except AttributeError as e:

narwhals/series.py

@@ -766,8 +766,8 @@ def arg_true(self) -> Self:
             We can then pass either pandas or Polars to `func`:
 
             >>> func(s_pd)
-            0    1
-            1    2
+            1    1
+            2    2
             Name: a, dtype: int64
             >>> func(s_pl)  # doctest: +NORMALIZE_WHITESPACE
             shape: (2,)

narwhals/stable/v1.py

@@ -1,6 +1,5 @@
 from __future__ import annotations
 
-from functools import wraps
 from typing import TYPE_CHECKING
 from typing import Any
 from typing import Callable
@@ -39,6 +38,7 @@
 from narwhals.schema import Schema as NwSchema
 from narwhals.series import Series as NwSeries
 from narwhals.translate import get_native_namespace as nw_get_native_namespace
+from narwhals.translate import narwhalify as nw_narwhalify
 from narwhals.translate import to_native
 from narwhals.typing import IntoDataFrameT
 from narwhals.typing import IntoFrameT
@@ -734,6 +734,7 @@ def narwhalify(
     *,
     strict: bool = False,
     eager_only: bool | None = False,
+    eager_or_interchange_only: bool | None = False,
     series_only: bool | None = False,
     allow_series: bool | None = True,
 ) -> Callable[..., Any]:
@@ -776,7 +777,7 @@ def func(df):
     You can also pass in extra arguments, e.g.
 
     ```python
-    @nw.narhwalify(eager_only=True)
+    @nw.narwhalify(eager_only=True)
     ```
 
     that will get passed down to `nw.from_native`.
@@ -792,53 +793,14 @@ def func(df):
         allow_series: Whether to allow series (default is only dataframe / lazyframe).
     """
 
-    # TODO(Unassigned): do we have a way to de-dupe this a bit?
-    def decorator(func: Callable[..., Any]) -> Callable[..., Any]:
-        @wraps(func)
-        def wrapper(*args: Any, **kwargs: Any) -> Any:
-            args = [
-                from_native(
-                    arg,
-                    strict=strict,
-                    eager_only=eager_only,
-                    series_only=series_only,
-                    allow_series=allow_series,
-                )
-                for arg in args
-            ]  # type: ignore[assignment]
-
-            kwargs = {
-                name: from_native(
-                    value,
-                    strict=strict,
-                    eager_only=eager_only,
-                    series_only=series_only,
-                    allow_series=allow_series,
-                )
-                for name, value in kwargs.items()
-            }
-
-            backends = {
-                b()
-                for v in [*args, *kwargs.values()]
-                if (b := getattr(v, "__native_namespace__", None))
-            }
-
-            if backends.__len__() > 1:
-                msg = "Found multiple backends. Make sure that all dataframe/series inputs come from the same backend."
-                raise ValueError(msg)
-
-            result = func(*args, **kwargs)
-
-            return to_native(result, strict=strict)
-
-        return wrapper
-
-    if func is None:
-        return decorator
-    else:
-        # If func is not None, it means the decorator is used without arguments
-        return decorator(func)
+    return nw_narwhalify(
+        func=func,
+        strict=strict,
+        eager_only=eager_only,
+        eager_or_interchange_only=eager_or_interchange_only,
+        series_only=series_only,
+        allow_series=allow_series,
+    )
 
 
 def all() -> Expr:

narwhals/translate.py

@@ -595,7 +595,7 @@ def func(df):
     You can also pass in extra arguments, e.g.
 
     ```python
-    @nw.narhwalify(eager_only=True)
+    @nw.narwhalify(eager_only=True)
     ```
 
     that will get passed down to `nw.from_native`.

requirements-dev.txt

@@ -1,6 +1,6 @@
 covdefaults
 pandas
-polars[timezones]
+polars
 pre-commit
 pyarrow
 pytest

tests/conftest.py

@@ -7,6 +7,7 @@
 import pyarrow as pa
 import pytest
 
+from narwhals.dependencies import get_cudf
 from narwhals.dependencies import get_dask_dataframe
 from narwhals.dependencies import get_modin
 from narwhals.typing import IntoDataFrame
@@ -17,6 +18,8 @@
     import modin.pandas  # noqa: F401
 with contextlib.suppress(ImportError):
     import dask.dataframe  # noqa: F401
+with contextlib.suppress(ImportError):
+    import cudf  # noqa: F401
 
 
 def pytest_addoption(parser: Any) -> None:
@@ -56,6 +59,11 @@ def modin_constructor(obj: Any) -> IntoDataFrame:  # pragma: no cover
     return mpd.DataFrame(pd.DataFrame(obj)).convert_dtypes(dtype_backend="pyarrow")  # type: ignore[no-any-return]
 
 
+def cudf_constructor(obj: Any) -> IntoDataFrame:  # pragma: no cover
+    cudf = get_cudf()
+    return cudf.DataFrame(obj)  # type: ignore[no-any-return]
+
+
 def polars_eager_constructor(obj: Any) -> IntoDataFrame:
     return pl.DataFrame(obj)
 
@@ -87,7 +95,8 @@ def pyarrow_table_constructor(obj: Any) -> IntoDataFrame:
 
 if get_modin() is not None:  # pragma: no cover
     eager_constructors.append(modin_constructor)
-# TODO(unassigned): when Dask gets better support, remove the "False and" part
+if get_cudf() is not None:
+    eager_constructors.append(cudf_constructor)  # pragma: no cover
 if get_dask_dataframe() is not None:  # pragma: no cover
     lazy_constructors.append(dask_lazy_constructor)  # type: ignore  # noqa: PGH003
 

tests/expr_and_series/arithmetic_test.py

@@ -143,10 +143,10 @@ def test_truediv_same_dims(constructor_eager: Any, request: Any) -> None:
         request.applymarker(pytest.mark.xfail)
     s_left = nw.from_native(constructor_eager({"a": [1, 2, 3]}), eager_only=True)["a"]
     s_right = nw.from_native(constructor_eager({"a": [2, 2, 1]}), eager_only=True)["a"]
-    result = (s_left / s_right).to_list()
-    assert result == [0.5, 1.0, 3.0]
-    result = (s_left.__rtruediv__(s_right)).to_list()
-    assert result == [2, 1, 1 / 3]
+    result = s_left / s_right
+    compare_dicts({"a": result}, {"a": [0.5, 1.0, 3.0]})
+    result = s_left.__rtruediv__(s_right)
+    compare_dicts({"a": result}, {"a": [2, 1, 1 / 3]})
 
 
 @pytest.mark.slow()

tests/expr_and_series/cat/get_categories_test.py

@@ -25,8 +25,8 @@ def test_get_categories(request: Any, constructor_eager: Any) -> None:
     result_expr = df.select(nw.col("a").cat.get_categories())
     compare_dicts(result_expr, expected)
 
-    result_series = df["a"].cat.get_categories().to_list()
-    assert set(result_series) == set(expected["a"])
+    result_series = df["a"].cat.get_categories()
+    compare_dicts({"a": result_series}, expected)
 
 
 def test_get_categories_pyarrow() -> None:
@@ -41,5 +41,5 @@ def test_get_categories_pyarrow() -> None:
     result_expr = df.select(nw.col("a").cat.get_categories())
     compare_dicts(result_expr, expected)
 
-    result_series = df["a"].cat.get_categories().to_list()
-    assert result_series == expected["a"]
+    result_series = df["a"].cat.get_categories()
+    compare_dicts({"a": result_series}, expected)