Fix regression when deserializing python rows into cython

Fixed regression involving pickling of Python rows between the cython and
pure Python implementations of :class:`.Row`, which occurred as part of
refactoring code for version 2.0 with typing. A particular constant were
turned into a string based ``Enum`` for the pure Python version of
:class:`.Row` whereas the cython version continued to use an integer
constant, leading to deserialization failures.

Regression occurred in a4bb502cf95ea3523e4d383c4377e50f402d7d52

Fixes: #9423
Change-Id: Icbd85cacb2d589cef7c246de7064249926146f2e

1 files changed, 4 insertions, 1 deletions

diff --git a/lib/sqlalchemy/engine/_py_row.py b/lib/sqlalchemy/engine/_py_row.py
index 7cbac552f..1b952fe4c 100644
--- a/lib/sqlalchemy/engine/_py_row.py
+++ b/lib/sqlalchemy/engine/_py_row.py
@@ -24,7 +24,7 @@ if typing.TYPE_CHECKING:
 MD_INDEX = 0  # integer index in cursor.description
 
 
-class _KeyStyle(enum.Enum):
+class _KeyStyle(enum.IntEnum):
     KEY_INTEGER_ONLY = 0
     """__getitem__ only allows integer values and slices, raises TypeError
     otherwise"""
@@ -121,6 +121,9 @@ class BaseRow:
         mdindex = rec[MD_INDEX]
         if mdindex is None:
             self._parent._raise_for_ambiguous_column_name(rec)
+        # NOTE: keep "== KEY_OBJECTS_ONLY" instead of "is KEY_OBJECTS_ONLY"
+        # since deserializing the class from cython will load an int in
+        # _key_style, not an instance of _KeyStyle
         elif self._key_style == KEY_OBJECTS_ONLY and isinstance(key, int):
             raise KeyError(key)