String dtype: honor mode.string_storage option (and change default to None)

Author: jorisvandenbossche

pandas/core/arrays/string_.py

@@ -140,12 +140,14 @@ def __init__(
         # infer defaults
         if storage is None:
             if na_value is not libmissing.NA:
-                if HAS_PYARROW:
-                    storage = "pyarrow"
-                else:
-                    storage = "python"
-            else:
                 storage = get_option("mode.string_storage")
+                if storage is None:
+                    if HAS_PYARROW:
+                        storage = "pyarrow"
+                    else:
+                        storage = "python"
+            else:
+                storage = get_option("mode.string_storage") or "python"
 
         if storage == "pyarrow_numpy":
             # TODO raise a deprecation warning

pandas/core/config_init.py

@@ -452,13 +452,12 @@ def is_terminal() -> bool:
 
 string_storage_doc = """
 : string
-    The default storage for StringDtype. This option is ignored if
-    ``future.infer_string`` is set to True.
+    The default storage for StringDtype.
 """
 
 
 def is_valid_string_storage(value: Any) -> None:
-    legal_values = ["python", "pyarrow"]
+    legal_values = [None, "python", "pyarrow"]
     if value not in legal_values:
         msg = "Value must be one of python|pyarrow"
         if value == "pyarrow_numpy":
@@ -473,7 +472,7 @@ def is_valid_string_storage(value: Any) -> None:
 with cf.config_prefix("mode"):
     cf.register_option(
         "string_storage",
-        "python",
+        None,
         string_storage_doc,
         # validator=is_one_of_factory(["python", "pyarrow"]),
         validator=is_valid_string_storage,

pandas/tests/arrays/string_/test_string_arrow.py

@@ -4,7 +4,6 @@
 import numpy as np
 import pytest
 
-from pandas.compat import HAS_PYARROW
 import pandas.util._test_decorators as td
 
 import pandas as pd
@@ -28,11 +27,6 @@ def test_eq_all_na():
 
 
 def test_config(string_storage, request, using_infer_string):
-    if using_infer_string and string_storage == "python" and HAS_PYARROW:
-        # string storage with na_value=NaN always uses pyarrow if available
-        # -> does not yet honor the option
-        request.applymarker(pytest.mark.xfail(reason="TODO(infer_string)"))
-
     with pd.option_context("string_storage", string_storage):
         assert StringDtype().storage == string_storage
         result = pd.array(["a", "b"])