[3.14] gh-135437: Account For Duplicate Names in _PyCode_SetUnboundVarCounts() (gh-135493)

(cherry picked from commit 56eabea, AKA gh-135438)

Co-authored-by: Eric Snow <ericsnowcurrently@gmail.com>

Lib/test/_code_definitions.py

@@ -57,6 +57,13 @@ def spam_with_globals_and_builtins():
     print(res)
 
 
+def spam_with_global_and_attr_same_name():
+    try:
+        spam_minimal.spam_minimal
+    except AttributeError:
+        pass
+
+
 def spam_full_args(a, b, /, c, d, *args, e, f, **kwargs):
     return (a, b, c, d, e, f, args, kwargs)
 
@@ -190,6 +197,7 @@ def ham_C_closure(z):
     spam_minimal,
     spam_with_builtins,
     spam_with_globals_and_builtins,
+    spam_with_global_and_attr_same_name,
     spam_full_args,
     spam_full_args_with_defaults,
     spam_args_attrs_and_builtins,
@@ -258,6 +266,7 @@ def ham_C_closure(z):
     script_with_globals,
     spam_full_args_with_defaults,
     spam_with_globals_and_builtins,
+    spam_with_global_and_attr_same_name,
     spam_full,
 ]
 
@@ -275,6 +284,7 @@ def ham_C_closure(z):
     *PURE_SCRIPT_FUNCTIONS,
     script_with_globals,
     spam_with_globals_and_builtins,
+    spam_with_global_and_attr_same_name,
 ]
 
 

Lib/test/test_code.py

@@ -701,6 +701,7 @@ def test_local_kinds(self):
                 'checks': CO_FAST_LOCAL,
                 'res': CO_FAST_LOCAL,
             },
+            defs.spam_with_global_and_attr_same_name: {},
             defs.spam_full_args: {
                 'a': POSONLY,
                 'b': POSONLY,
@@ -955,6 +956,10 @@ def new_var_counts(*,
                 purelocals=5,
                 globalvars=6,
             ),
+            defs.spam_with_global_and_attr_same_name: new_var_counts(
+                globalvars=2,
+                attrs=1,
+            ),
             defs.spam_full_args: new_var_counts(
                 posonly=2,
                 posorkw=2,

Objects/codeobject.c

@@ -1714,7 +1714,7 @@ static int
 identify_unbound_names(PyThreadState *tstate, PyCodeObject *co,
                        PyObject *globalnames, PyObject *attrnames,
                        PyObject *globalsns, PyObject *builtinsns,
-                       struct co_unbound_counts *counts)
+                       struct co_unbound_counts *counts, int *p_numdupes)
 {
     // This function is inspired by inspect.getclosurevars().
     // It would be nicer if we had something similar to co_localspluskinds,
@@ -1729,6 +1729,7 @@ identify_unbound_names(PyThreadState *tstate, PyCodeObject *co,
     assert(builtinsns == NULL || PyDict_Check(builtinsns));
     assert(counts == NULL || counts->total == 0);
     struct co_unbound_counts unbound = {0};
+    int numdupes = 0;
     Py_ssize_t len = Py_SIZE(co);
     for (int i = 0; i < len; i += _PyInstruction_GetLength(co, i)) {
         _Py_CODEUNIT inst = _Py_GetBaseCodeUnit(co, i);
@@ -1747,6 +1748,12 @@ identify_unbound_names(PyThreadState *tstate, PyCodeObject *co,
             if (PySet_Add(attrnames, name) < 0) {
                 return -1;
             }
+            if (PySet_Contains(globalnames, name)) {
+                if (_PyErr_Occurred(tstate)) {
+                    return -1;
+                }
+                numdupes += 1;
+            }
         }
         else if (inst.op.code == LOAD_GLOBAL) {
             int oparg = GET_OPARG(co, i, inst.op.arg);
@@ -1778,11 +1785,20 @@ identify_unbound_names(PyThreadState *tstate, PyCodeObject *co,
             if (PySet_Add(globalnames, name) < 0) {
                 return -1;
             }
+            if (PySet_Contains(attrnames, name)) {
+                if (_PyErr_Occurred(tstate)) {
+                    return -1;
+                }
+                numdupes += 1;
+            }
         }
     }
     if (counts != NULL) {
         *counts = unbound;
     }
+    if (p_numdupes != NULL) {
+        *p_numdupes = numdupes;
+    }
     return 0;
 }
 
@@ -1932,20 +1948,24 @@ _PyCode_SetUnboundVarCounts(PyThreadState *tstate,
 
     // Fill in unbound.globals and unbound.numattrs.
     struct co_unbound_counts unbound = {0};
+    int numdupes = 0;
     Py_BEGIN_CRITICAL_SECTION(co);
     res = identify_unbound_names(
             tstate, co, globalnames, attrnames, globalsns, builtinsns,
-            &unbound);
+            &unbound, &numdupes);
     Py_END_CRITICAL_SECTION();
     if (res < 0) {
         goto finally;
     }
     assert(unbound.numunknown == 0);
-    assert(unbound.total <= counts->unbound.total);
+    assert(unbound.total - numdupes <= counts->unbound.total);
     assert(counts->unbound.numunknown == counts->unbound.total);
-    unbound.numunknown = counts->unbound.total - unbound.total;
+    // There may be a name that is both a global and an attr.
-    unbound.total = counts->unbound.total;
+    int totalunbound = counts->unbound.total + numdupes;
+    unbound.numunknown = totalunbound - unbound.total;
+    unbound.total = totalunbound;
     counts->unbound = unbound;
+    counts->total += numdupes;
     res = 0;
 
 finally: