From a55d441515b27519a6f5fdfa43d620a682608943 Mon Sep 17 00:00:00 2001
From: Joris Van den Bossche <jorisvandenbossche@gmail.com>
Date: Mon, 29 Sep 2025 10:06:40 +0200
Subject: [PATCH 1/2] BUG: avoid validation error for ufunc with string[python]
 array

---
 doc/source/whatsnew/v2.3.3.rst             |  2 +-
 pandas/core/arrays/numpy_.py               |  9 ++++++++
 pandas/tests/arrays/string_/test_string.py | 27 ++++++++++++++++++++++
 3 files changed, 37 insertions(+), 1 deletion(-)

diff --git a/doc/source/whatsnew/v2.3.3.rst b/doc/source/whatsnew/v2.3.3.rst
index 0accd07eb366a..6c106841e970b 100644
--- a/doc/source/whatsnew/v2.3.3.rst
+++ b/doc/source/whatsnew/v2.3.3.rst
@@ -47,7 +47,7 @@ Bug fixes
 - Fix regression in ``~Series.str.contains``, ``~Series.str.match`` and ``~Series.str.fullmatch``
   with a compiled regex and custom flags (:issue:`62240`)
 - Fix :meth:`Series.str.match` and :meth:`Series.str.fullmatch` not matching patterns with groups correctly for the Arrow-backed string dtype (:issue:`61072`)
-
+- Fix error being raised when using a numpy ufunc with a Python-backed string array (:issue:`40800`)
 
 Improvements and fixes for Copy-on-Write
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
diff --git a/pandas/core/arrays/numpy_.py b/pandas/core/arrays/numpy_.py
index ef64bda3dc504..620d0b8696470 100644
--- a/pandas/core/arrays/numpy_.py
+++ b/pandas/core/arrays/numpy_.py
@@ -234,6 +234,15 @@ def __array_ufunc__(self, ufunc: np.ufunc, method: str, *inputs, **kwargs):
             # e.g. test_np_max_nested_tuples
             return result
         else:
+            if self.dtype.type is str:
+                # StringDtype
+                try:
+                    # specify dtype to preserve storage/na_value
+                    return type(self)(result, dtype=self.dtype)
+                except ValueError:
+                    # if validation of input fails (no strings)
+                    # -> fallback to returning raw numpy array
+                    return result
             # one return value; re-box array-like results
             return type(self)(result)
 
diff --git a/pandas/tests/arrays/string_/test_string.py b/pandas/tests/arrays/string_/test_string.py
index 1a71f6c41c4f1..d3effb7c33457 100644
--- a/pandas/tests/arrays/string_/test_string.py
+++ b/pandas/tests/arrays/string_/test_string.py
@@ -840,3 +840,30 @@ def test_string_array_view_type_error():
     arr = pd.array(["a", "b", "c"], dtype="string")
     with pytest.raises(TypeError, match="Cannot change data-type for string array."):
         arr.view("i8")
+
+
+@pytest.mark.parametrize("box", [pd.Series, pd.array])
+def test_numpy_array_ufunc(dtype, box):
+    arr = box(["a", "bb", "ccc"], dtype=dtype)
+
+    # custom ufunc that works with string (object) input -> returning numeric
+    str_len_ufunc = np.frompyfunc(lambda x: len(x), 1, 1)
+    result = str_len_ufunc(arr)
+    expected_cls = pd.Series if box is pd.Series else np.array
+    # TODO we should infer int64 dtype here?
+    expected = expected_cls([1, 2, 3], dtype=object)
+    tm.assert_equal(result, expected)
+
+    # custom ufunc returning strings
+    str_multiply_ufunc = np.frompyfunc(lambda x: x * 2, 1, 1)
+    result = str_multiply_ufunc(arr)
+    expected = box(["aa", "bbbb", "cccccc"], dtype=dtype)
+    if dtype.storage == "pyarrow":
+        # TODO ArrowStringArray should also preserve the class / dtype
+        if box is pd.array:
+            expected = np.array(["aa", "bbbb", "cccccc"], dtype=object)
+        else:
+            # not specifying the dtype because the exact dtype is not yet preserved
+            expected = pd.Series(["aa", "bbbb", "cccccc"])
+
+    tm.assert_equal(result, expected)

From 5bcad4ecbd3f107fad2063274da46e33cb825e40 Mon Sep 17 00:00:00 2001
From: Joris Van den Bossche <jorisvandenbossche@gmail.com>
Date: Mon, 29 Sep 2025 13:44:02 +0200
Subject: [PATCH 2/2] fix typing

---
 pandas/core/arrays/numpy_.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/pandas/core/arrays/numpy_.py b/pandas/core/arrays/numpy_.py
index 620d0b8696470..fab51ffa56919 100644
--- a/pandas/core/arrays/numpy_.py
+++ b/pandas/core/arrays/numpy_.py
@@ -5,6 +5,7 @@
     Any,
     Literal,
     Self,
+    cast,
 )
 
 import numpy as np
@@ -48,6 +49,7 @@
     )
 
     from pandas import Index
+    from pandas.arrays import StringArray
 
 
 class NumpyExtensionArray(
@@ -234,8 +236,9 @@ def __array_ufunc__(self, ufunc: np.ufunc, method: str, *inputs, **kwargs):
             # e.g. test_np_max_nested_tuples
             return result
         else:
-            if self.dtype.type is str:
+            if self.dtype.type is str:  # type: ignore[comparison-overlap]
                 # StringDtype
+                self = cast("StringArray", self)
                 try:
                     # specify dtype to preserve storage/na_value
                     return type(self)(result, dtype=self.dtype)