feat(python): Support Numpy ufunc with more than one expression

We use the pl.reduce trick basically, where the limitation is that non-expressions have to be passed in as kwargs. That is probably the safest anyway.

Related issues:
#6770 : brought up no support for multiple expression, have added a ValueError in response

#5713 : reminder that there is the `pl.reduce` trick

py-polars/polars/expr/expr.py

@@ -7,7 +7,7 @@
 import random
 import warnings
 from datetime import timedelta
-from functools import reduce
+from functools import partial, reduce
 from typing import (
     TYPE_CHECKING,
     Any,
@@ -47,7 +47,7 @@
 
 with contextlib.suppress(ImportError):  # Module not available when building docs
     from polars.polars import arg_where as py_arg_where
-
+    from polars.polars import reduce as pyreduce
 if TYPE_CHECKING:
     import sys
 
@@ -282,9 +282,13 @@ def __array_ufunc__(
         """Numpy universal functions."""
         num_expr = sum(isinstance(inp, Expr) for inp in inputs)
         if num_expr > 1:
-            raise ValueError(
-                f"Numpy ufunc can only be used with one expression, {num_expr} given. Use `pl.reduce` to call numpy functions over multiple expressions."
-            )
+            if num_expr < len(inputs):
+                raise ValueError(
+                    "Numpy ufunc with more than one expression can only be used if all non-expression inputs are provided as keyword arguments only"
+                )
+
+            exprs = selection_to_pyexpr_list(inputs)
+            return self._from_pyexpr(pyreduce(partial(ufunc, **kwargs), exprs))
 
         def function(s: Series) -> Series:  # pragma: no cover
             args = [inp if not isinstance(inp, Expr) else s for inp in inputs]

py-polars/tests/unit/test_df.py

@@ -3654,6 +3654,41 @@ def test_ufunc_expr_not_first() -> None:
     assert_frame_equal(out, expected)
 
 
+def test_ufunc_multiple_expressions() -> None:
+    # example from https://github.com/pola-rs/polars/issues/6770
+    df = pl.DataFrame(
+        {
+            "v": [
+                -4.293,
+                -2.4659,
+                -1.8378,
+                -0.2821,
+                -4.5649,
+                -3.8128,
+                -7.4274,
+                3.3443,
+                3.8604,
+                -4.2200,
+            ],
+            "u": [
+                -11.2268,
+                6.3478,
+                7.1681,
+                3.4986,
+                2.7320,
+                -1.0695,
+                -10.1408,
+                11.2327,
+                6.6623,
+                -8.1412,
+            ],
+        }
+    )
+    expected = np.arctan2(df.get_column("v"), df.get_column("u"))
+    result = df.select(np.arctan2(pl.col("v"), pl.col("u")))[:, 0]  # type: ignore[call-overload]
+    assert_series_equal(expected, result)  # type: ignore[arg-type]
+
+
 def test_window_deadlock() -> None:
     np.random.seed(12)