Disallow NaN in StringArray constructor

Closes pandas-dev#30966

Author: TomAugspurger

pandas/_libs/lib.pyx

@@ -1472,12 +1472,30 @@ cdef class Validator:
         Py_ssize_t n
         dtype dtype
         bint skipna
+        bint na_only
 
     def __cinit__(self, Py_ssize_t n, dtype dtype=np.dtype(np.object_),
-                  bint skipna=False):
+                  bint skipna=False,
+                  bint na_only=False):
+        """
+
+        Parameters
+        ----------
+        n
+        dtype
+        skipna
+        na_only : bool, default False
+            Whether to only treat pandas.NA as NA. Values like None
+            and NaN won't be treated as NA.
+
+        Returns
+        -------
+
+        """
         self.n = n
         self.dtype = dtype
         self.skipna = skipna
+        self.na_only = na_only
 
     cdef bint validate(self, ndarray values) except -1:
         if not self.n:
@@ -1530,7 +1548,10 @@ cdef class Validator:
                                   "must define is_value_typed")
 
     cdef bint is_valid_null(self, object value) except -1:
-        return value is None or value is C_NA or util.is_nan(value)
+        if self.na_only:
+            return value is C_NA
+        else:
+            return value is None or value is C_NA or util.is_nan(value)
 
     cdef bint is_array_typed(self) except -1:
         return False
@@ -1625,11 +1646,12 @@ cdef class StringValidator(Validator):
         return issubclass(self.dtype.type, np.str_)
 
 
-cpdef bint is_string_array(ndarray values, bint skipna=False):
+cpdef bint is_string_array(ndarray values, bint skipna=False, na_only=False):
     cdef:
         StringValidator validator = StringValidator(len(values),
                                                     values.dtype,
-                                                    skipna=skipna)
+                                                    skipna=skipna,
+                                                    na_only=na_only)
     return validator.validate(values)
 
 

pandas/core/arrays/string_.py

@@ -93,9 +93,6 @@ class StringArray(PandasArray):
        StringArray is considered experimental. The implementation and
        parts of the API may change without warning.
 
-       In particular, the NA value used may change to no longer be
-       ``numpy.nan``.
-
     Parameters
     ----------
     values : array-like
@@ -104,8 +101,11 @@ class StringArray(PandasArray):
         .. warning::
 
            Currently, this expects an object-dtype ndarray
-           where the elements are Python strings. This may
-           change without warning in the future.
+           where the elements are Python strings or :attr:`pandas.NA`.
+           This may change without warning in the future. Use
+           :meth:`pandas.array` with ``dtype="string"`` for a stable way of
+           creating a `StringArray` from any sequence.
+
     copy : bool, default False
         Whether to copy the array of data.
 
@@ -119,6 +119,8 @@ class StringArray(PandasArray):
 
     See Also
     --------
+    pandas.array
+        The recommended function for creating a StringArray.
     Series.str
         The string methods are available on Series backed by
         a StringArray.
@@ -164,13 +166,13 @@ def __init__(self, values, copy=False):
 
     def _validate(self):
         """Validate that we only store NA or strings."""
-        if len(self._ndarray) and not lib.is_string_array(self._ndarray, skipna=True):
-            raise ValueError(
-                "StringArray requires a sequence of strings or missing values."
-            )
+        if len(self._ndarray) and not lib.is_string_array(
+            self._ndarray, skipna=True, na_only=True
+        ):
+            raise ValueError("StringArray requires a sequence of strings or pandas.NA")
         if self._ndarray.dtype != "object":
             raise ValueError(
-                "StringArray requires a sequence of strings. Got "
+                "StringArray requires a sequence of strings or pandas.NA. Got "
                 f"'{self._ndarray.dtype}' dtype instead."
             )
 

pandas/tests/arrays/string_/test_string.py

@@ -194,6 +194,12 @@ def test_constructor_raises():
     with pytest.raises(ValueError, match="sequence of strings"):
         pd.arrays.StringArray(np.array([]))
 
+    with pytest.raises(ValueError, match="strings or pandas.NA"):
+        pd.arrays.StringArray(np.array(["a", np.nan], dtype=object))
+
+    with pytest.raises(ValueError, match="strings or pandas.NA"):
+        pd.arrays.StringArray(np.array(["a", None], dtype=object))
+
 
 @pytest.mark.parametrize("skipna", [True, False])
 @pytest.mark.xfail(reason="Not implemented StringArray.sum")