ENH: Update DataFrame.to_stata to handle pd.NA and None values in strL columns

Author: Jesse

doc/source/whatsnew/v2.3.0.rst

@@ -150,6 +150,7 @@ MultiIndex
 I/O
 ^^^
 - :meth:`DataFrame.to_excel` was storing decimals as strings instead of numbers (:issue:`49598`)
+- :meth:`DataFrame.to_stata` no longer throws a ``TypeError('encoding without a string argument')`` when exporting a column containing both long strings (Stata strL) and :class:`pd.NA` values (:issue:`23633`)
 -
 
 Period

pandas/io/stata.py

@@ -3196,8 +3196,8 @@ def generate_table(self) -> tuple[dict[str, tuple[int, int]], DataFrame]:
         for o, (idx, row) in enumerate(selected.iterrows()):
             for j, (col, v) in enumerate(col_index):
                 val = row[col]
-                # Allow columns with mixed str and None (GH 23633)
-                val = "" if val is None else val
+                # Allow columns with mixed str and None or pd.NA (GH 23633)
+                val = "" if (val is None) or isna(val) else val
                 key = gso_table.get(val, None)
                 if key is None:
                     # Stata prefers human numbers

pandas/tests/io/test_stata.py

@@ -2587,3 +2587,17 @@ def test_many_strl(temp_file, version):
     lbls = ["".join(v) for v in itertools.product(*([string.ascii_letters] * 3))]
     value_labels = {"col": {i: lbls[i] for i in range(n)}}
     df.to_stata(temp_file, value_labels=value_labels, version=version)
+
+
+@pytest.mark.parametrize("version", [117, 118, 119, None])
+def test_strl_missings(temp_file, version):
+    # GH 23633
+    # Check that strl supports None and pd.NA
+    df = DataFrame(
+        [
+            {"str1": "string" * 500, "number": 0},
+            {"str1": None, "number": 1},
+            {"str1": pd.NA, "number": 1},
+        ]
+    )
+    df.to_stata(temp_file, version=version)