From 0a4c2f5b33e6966c0dbc9fd847dbb212214413e3 Mon Sep 17 00:00:00 2001
From: Joris Van den Bossche <jorisvandenbossche@gmail.com>
Date: Mon, 29 Sep 2025 18:22:14 +0200
Subject: [PATCH] BUG: avoid validation error for ufunc with string[python]
 array (#62498)

---
 doc/source/whatsnew/v2.3.3.rst             |  3 ++-
 pandas/core/arrays/numpy_.py               |  8 +++++++
 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 bc5a4c5b27a90..db710d29b66a0 100644
--- a/doc/source/whatsnew/v2.3.3.rst
+++ b/doc/source/whatsnew/v2.3.3.rst
@@ -47,7 +47,8 @@ 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 comparing a :class:`StringDtype` Series with mixed objects raising an error (:issue:`60228`)
+- 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 e0031d3db6ca7..07fd207933fc1 100644
--- a/pandas/core/arrays/numpy_.py
+++ b/pandas/core/arrays/numpy_.py
@@ -209,6 +209,14 @@ 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:  # type: ignore[comparison-overlap]
+                # StringDtype
+                try:
+                    return type(self)(result)
+                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 8c15f0c98f90e..b468480cf5f86 100644
--- a/pandas/tests/arrays/string_/test_string.py
+++ b/pandas/tests/arrays/string_/test_string.py
@@ -864,3 +864,30 @@ def test_tolist(dtype):
     result = arr.tolist()
     expected = vals
     tm.assert_equal(result, expected)
+
+
+@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)