Pickle globals read from nested class bodies

cloudpickle/cloudpickle.py

@@ -55,6 +55,7 @@
 import abc
 import builtins
 import copyreg
+import inspect
 import dataclasses
 import dis
 from enum import Enum
@@ -396,6 +397,13 @@ def func():
 _glbl_ops.append(LOAD_GLOBAL)
 GLOBAL_OPS = tuple(_glbl_ops)
 del _glbl_ops
+
+# Class bodies compile global reads as LOAD_NAME (and, on 3.12+, sometimes as
+# LOAD_FROM_DICT_OR_GLOBALS for annotations / type aliases), not LOAD_GLOBAL.
+LOAD_NAME = opcode.opmap["LOAD_NAME"]
+STORE_NAME = opcode.opmap["STORE_NAME"]
+LOAD_FROM_DICT_OR_GLOBALS = opcode.opmap.get("LOAD_FROM_DICT_OR_GLOBALS")
+
 HAVE_ARGUMENT = dis.HAVE_ARGUMENT
 EXTENDED_ARG = dis.EXTENDED_ARG
 
@@ -416,11 +424,35 @@ def _builtin_type(name):
 
 
 def _walk_global_ops(code):
-    """Yield referenced name for global-referencing instructions in code."""
+    """Yield referenced name for global-referencing instructions in code.
+
+    Function code uses LOAD_GLOBAL / STORE_GLOBAL. Class bodies instead use
+    LOAD_NAME for names looked up in the class namespace (falling back to
+    globals). To avoid treating a class-local assignment that shadows an
+    unpicklable global as a global read, LOAD_NAME only counts as a global
+    reference when it appears before any STORE_NAME of the same name in that
+    code object.
+    """
+    # Class bodies are not CO_OPTIMIZED; function/comprehension code is.
+    class_body = not (code.co_flags & inspect.CO_OPTIMIZED)
+    seen_stores = set()
+
     for instr in dis.get_instructions(code):
         op = instr.opcode
         if op in GLOBAL_OPS:
             yield instr.argval
+        elif class_body and op == STORE_NAME:
+            seen_stores.add(instr.argval)
+        elif class_body and op == LOAD_NAME:
+            if instr.argval not in seen_stores:
+                yield instr.argval
+        elif (
+            class_body
+            and LOAD_FROM_DICT_OR_GLOBALS is not None
+            and op == LOAD_FROM_DICT_OR_GLOBALS
+        ):
+            # Type aliases / lazy annotations look up missing names in globals.
+            yield instr.argval
 
 
 def _extract_class_dict(cls):

tests/cloudpickle_test.py

@@ -2526,6 +2526,39 @@ def inner_function():
         inner_func = depickled_factory()
         assert inner_func() == _TEST_GLOBAL_VARIABLE
 
+    def test_class_body_reads_globals_by_value(self):
+        # Issue #609: globals read only inside a nested class body must be
+        # pickled for by-value functions.
+        ns = {"__name__": "not_importable", "CONST": 42}
+        exec(
+            "def f():\n"
+            "    class Rec:\n"
+            "        q = CONST\n"
+            "    return Rec.q\n",
+            ns,
+        )
+        self.assertEqual(ns["f"](), 42)
+        self.assertEqual(pickle_depickle(ns["f"], protocol=self.protocol)(), 42)
+
+    def test_class_body_local_shadow_skips_unpicklable_global(self):
+        # A class-local assignment that shadows an unpicklable global must not
+        # pull that global into the pickled function state.
+        import threading
+
+        ns = {"__name__": "not_importable", "resource": threading.Lock()}
+        exec(
+            "def factory():\n"
+            "    class Record:\n"
+            "        resource = 42\n"
+            "        value = resource\n"
+            "    return Record.value\n",
+            ns,
+        )
+        self.assertEqual(ns["factory"](), 42)
+        self.assertEqual(
+            pickle_depickle(ns["factory"], protocol=self.protocol)(), 42
+        )
+
     # TODO: remove this xfail when we drop support for Python 3.8. We don't
     # plan to fix it because Python 3.8 is EOL.
     @pytest.mark.skipif(