PERF: StringArray from np.str_ array

asv_bench/benchmarks/array.py

@@ -44,6 +44,24 @@ def time_from_integer_array(self):
         pd.array(self.values_integer, dtype="Int64")
 
 
+class StringArray:
+    def setup(self):
+        N = 100_000
+        values = tm.rands_array(3, N)
+        self.values_obj = np.array(values, dtype="object")
+        self.values_str = np.array(values, dtype="U")
+        self.values_list = values.tolist()
+
+    def time_from_np_object_array(self):
+        pd.array(self.values_obj, dtype="string")
+
+    def time_from_np_str_array(self):
+        pd.array(self.values_str, dtype="string")
+
+    def time_from_list(self):
+        pd.array(self.values_list, dtype="string")
+
+
 class ArrowStringArray:
 
     params = [False, True]

doc/source/whatsnew/v2.0.0.rst

@@ -156,6 +156,7 @@ Performance improvements
 - Performance improvement in :func:`merge` and :meth:`DataFrame.join` when joining on a sorted :class:`MultiIndex` (:issue:`48504`)
 - Performance improvement in :meth:`DataFrame.loc` and :meth:`Series.loc` for tuple-based indexing of a :class:`MultiIndex` (:issue:`48384`)
 - Performance improvement for :meth:`MultiIndex.unique` (:issue:`48335`)
+- Performance improvement for :class:`~arrays.StringArray` constructor passing a numpy array with type ``np.str_`` (:issue:`49109`)
 - Performance improvement for :func:`concat` with extension array backed indexes (:issue:`49128`)
 - Performance improvement in :meth:`DataFrame.join` when joining on a subset of a :class:`MultiIndex` (:issue:`48611`)
 - Performance improvement for :meth:`MultiIndex.intersection` (:issue:`48604`)

pandas/_libs/lib.pyx

@@ -703,6 +703,10 @@ cpdef ndarray[object] ensure_string_array(
     if copy and result is arr:
         result = result.copy()
 
+    if issubclass(arr.dtype.type, np.str_):
+        # short-circuit, all elements are str
+        return result
+
     for i in range(n):
         val = arr[i]
 

pandas/tests/arrays/string_/test_string.py

@@ -597,6 +597,14 @@ def test_setitem_scalar_with_mask_validation(dtype):
         ser[mask] = 1
 
 
+def test_from_numpy_str(dtype):
+    vals = ["a", "b", "c"]
+    arr = np.array(vals, dtype=np.str_)
+    result = pd.array(arr, dtype=dtype)
+    expected = pd.array(vals, dtype=dtype)
+    tm.assert_extension_array_equal(result, expected)
+
+
 def test_tolist(dtype):
     vals = ["a", "b", "c"]
     arr = pd.array(vals, dtype=dtype)