From 08d10668035fd888ea494876411ee4df1c3bc2f1 Mon Sep 17 00:00:00 2001
From: Jeff <138259519+jdotc1@users.noreply.github.com>
Date: Fri, 23 Oct 2026 01:46:40 -0400
Subject: [PATCH 1/2] BUG: support variable-width strings in str.extract

---
 doc/whats-new.rst                 |  4 ++++
 xarray/core/accessor_str.py       |  4 ++--
 xarray/tests/test_accessor_str.py | 20 ++++++++++++++++++++
 3 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/doc/whats-new.rst b/doc/whats-new.rst
index 21f3671188a..3662d5cb00e 100644
--- a/doc/whats-new.rst
+++ b/doc/whats-new.rst
@@ -165,6 +165,10 @@ Deprecations
 
 Bug Fixes
 ~~~~~~~~~
+- ``DataArray.str.extract`` with a named output dimension now supports NumPy
+  variable-width strings, including unmatched values and optional capture groups
+  (:issue:`11785`).
+  By `jdotc1 <https://github.com/jdotc1>`_.
 - Results of the string accessor methods ``strip``, ``pad``, ``center``,
   ``ljust``, ``rjust``, ``zfill`` and ``replace`` on dask arrays are no longer
   truncated to a single character, and ``repeat`` no longer fails on dask arrays
diff --git a/xarray/core/accessor_str.py b/xarray/core/accessor_str.py
index a51c3822bc2..5be9bff2dcb 100644
--- a/xarray/core/accessor_str.py
+++ b/xarray/core/accessor_str.py
@@ -2154,10 +2154,10 @@ def _get_res_single(val, pat):
         def _get_res_multi(val, pat):
             match = pat.search(val)
             if match is None:
-                return np.array([""], val.dtype)
+                return np.array([""], dtype=np.object_)
             match = match.groups()
             match = [grp if grp is not None else "" for grp in match]
-            return np.array(match, val.dtype)
+            return np.array(match, dtype=np.object_)
 
         if dim is None:
             return self._apply(func=_get_res_single, func_args=(pat,))
diff --git a/xarray/tests/test_accessor_str.py b/xarray/tests/test_accessor_str.py
index d49f8bca2c9..a34329d88da 100644
--- a/xarray/tests/test_accessor_str.py
+++ b/xarray/tests/test_accessor_str.py
@@ -852,6 +852,26 @@ def test_extract_single_nocase(dtype) -> None:
     assert_equal(res_re_dim, targ_dim)
 
 
+@pytest.mark.parametrize("string_dtype", ["T", "U", "S"])
+@pytest.mark.parametrize("compiled", [False, True])
+def test_extract_multi_string_dtypes(string_dtype, compiled) -> None:
+    value = xr.DataArray(
+        np.array(["alpha_1234", "beta", "!"], dtype=string_dtype), dims="X"
+    )
+    pattern = r"([a-z]+)(?:_(\d+))?"
+    if compiled:
+        pattern = re.compile(pattern.encode() if string_dtype == "S" else pattern)
+    expected = xr.DataArray(
+        np.array([["alpha", "1234"], ["beta", ""], ["", ""]], dtype=string_dtype),
+        dims=["X", "group"],
+    )
+
+    result = value.str.extract(pattern, dim="group")
+
+    assert result.dtype == expected.dtype
+    assert_equal(result, expected)
+
+
 def test_extract_multi_case(dtype) -> None:
     pat_str = r"(\w+)_Xy_(\d*)"
     pat_re: str | bytes = (

From 49cfacb21b2ba142e289cafd7eb7c6d68fd9e0f0 Mon Sep 17 00:00:00 2001
From: Jeff <138259519+jdotc1@users.noreply.github.com>
Date: Fri, 23 Oct 2026 02:10:24 -0400
Subject: [PATCH 2/2] TST: fix inferred regex pattern type in string extraction
 test

---
 xarray/tests/test_accessor_str.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/xarray/tests/test_accessor_str.py b/xarray/tests/test_accessor_str.py
index a34329d88da..e89db57794c 100644
--- a/xarray/tests/test_accessor_str.py
+++ b/xarray/tests/test_accessor_str.py
@@ -858,9 +858,12 @@ def test_extract_multi_string_dtypes(string_dtype, compiled) -> None:
     value = xr.DataArray(
         np.array(["alpha_1234", "beta", "!"], dtype=string_dtype), dims="X"
     )
-    pattern = r"([a-z]+)(?:_(\d+))?"
-    if compiled:
-        pattern = re.compile(pattern.encode() if string_dtype == "S" else pattern)
+    pattern_str = r"([a-z]+)(?:_(\d+))?"
+    pattern = (
+        re.compile(pattern_str.encode() if string_dtype == "S" else pattern_str)
+        if compiled
+        else pattern_str
+    )
     expected = xr.DataArray(
         np.array([["alpha", "1234"], ["beta", ""], ["", ""]], dtype=string_dtype),
         dims=["X", "group"],