BUG: Fix dtype=str converts NaN to 'n' (#22564)

More specifically the cases that seem to have an issue
are when:
- the series in empty
- it's a single element series

* Closes #22477

doc/source/whatsnew/v0.24.0.rst

@@ -1443,6 +1443,7 @@ Reshaping
 - Bug in :func:`merge_asof` where confusing error message raised when attempting to merge with missing values (:issue:`23189`)
 - Bug in :meth:`DataFrame.nsmallest` and :meth:`DataFrame.nlargest` for dataframes that have a :class:`MultiIndex` for columns (:issue:`23033`).
 - Bug in :meth:`DataFrame.append` with a :class:`Series` with a dateutil timezone would raise a ``TypeError`` (:issue:`23682`)
+- Bug in ``Series`` construction when passing no data and ``dtype=str`` (:issue:`22477`)
 
 .. _whatsnew_0240.bug_fixes.sparse:
 

pandas/core/dtypes/cast.py

@@ -6,7 +6,7 @@
 
 from pandas._libs import lib, tslib, tslibs
 from pandas._libs.tslibs import OutOfBoundsDatetime, Period, iNaT
-from pandas.compat import PY3, string_types, text_type
+from pandas.compat import PY3, string_types, text_type, to_str
 
 from .common import (
     _INT64_DTYPE, _NS_DTYPE, _POSSIBLY_CAST_DTYPES, _TD_DTYPE, _string_dtypes,
@@ -1216,11 +1216,16 @@ def construct_1d_arraylike_from_scalar(value, length, dtype):
         if not isinstance(dtype, (np.dtype, type(np.dtype))):
             dtype = dtype.dtype
 
-        # coerce if we have nan for an integer dtype
-        # GH 22858: only cast to float if an index
-        # (passed here as length) is specified
         if length and is_integer_dtype(dtype) and isna(value):
-            dtype = np.float64
+            # coerce if we have nan for an integer dtype
+            dtype = np.dtype('float64')
+        elif isinstance(dtype, np.dtype) and dtype.kind in ("U", "S"):
+            # we need to coerce to object dtype to avoid
+            # to allow numpy to take our string as a scalar value
+            dtype = object
+            if not isna(value):
+                value = to_str(value)
+
         subarr = np.empty(length, dtype=dtype)
         subarr.fill(value)
 

pandas/core/dtypes/common.py

@@ -419,7 +419,7 @@ def is_datetime64_dtype(arr_or_dtype):
         return False
     try:
         tipo = _get_dtype_type(arr_or_dtype)
-    except TypeError:
+    except (TypeError, UnicodeEncodeError):
         return False
     return issubclass(tipo, np.datetime64)
 

pandas/tests/series/test_constructors.py

@@ -134,6 +134,17 @@ def test_constructor_no_data_index_order(self):
         result = pd.Series(index=['b', 'a', 'c'])
         assert result.index.tolist() == ['b', 'a', 'c']
 
+    def test_constructor_no_data_string_type(self):
+        # GH 22477
+        result = pd.Series(index=[1], dtype=str)
+        assert np.isnan(result.iloc[0])
+
+    @pytest.mark.parametrize('item', ['entry', 'ѐ', 13])
+    def test_constructor_string_element_string_type(self, item):
+        # GH 22477
+        result = pd.Series(item, index=[1], dtype=str)
+        assert result.iloc[0] == str(item)
+
     def test_constructor_dtype_str_na_values(self, string_dtype):
         # https://github.com/pandas-dev/pandas/issues/21083
         ser = Series(['x', None], dtype=string_dtype)