bpo-47084: Clear Unicode cached representations on finalization (GH-32032)

1 files changed, 44 insertions, 0 deletions

diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 5dfe6e1..ce3ebce 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -16057,6 +16057,35 @@ _PyUnicode_FiniTypes(PyInterpreterState *interp)
 }
 
 
+static void unicode_static_dealloc(PyObject *op)
+{
+    PyASCIIObject* ascii = (PyASCIIObject*)op;
+
+    assert(ascii->state.compact);
+
+    if (ascii->state.ascii) {
+        if (ascii->wstr) {
+            PyObject_Free(ascii->wstr);
+            ascii->wstr = NULL;
+        }
+    }
+    else {
+        PyCompactUnicodeObject* compact = (PyCompactUnicodeObject*)op;
+        void* data = (void*)(compact + 1);
+        if (ascii->wstr && ascii->wstr != data) {
+            PyObject_Free(ascii->wstr);
+            ascii->wstr = NULL;
+            compact->wstr_length = 0;
+        }
+        if (compact->utf8) {
+            PyObject_Free(compact->utf8);
+            compact->utf8 = NULL;
+            compact->utf8_length = 0;
+        }
+    }
+}
+
+
 void
 _PyUnicode_Fini(PyInterpreterState *interp)
 {
@@ -16070,6 +16099,21 @@ _PyUnicode_Fini(PyInterpreterState *interp)
     _PyUnicode_FiniEncodings(&state->fs_codec);
 
     unicode_clear_identifiers(state);
+
+    // Clear the single character singletons
+    for (int i = 0; i < 128; i++) {
+        unicode_static_dealloc((PyObject*)&_Py_SINGLETON(strings).ascii[i]);
+    }
+    for (int i = 0; i < 128; i++) {
+        unicode_static_dealloc((PyObject*)&_Py_SINGLETON(strings).latin1[i]);
+    }
+}
+
+
+void
+_PyStaticUnicode_Dealloc(PyObject *op)
+{
+    unicode_static_dealloc(op);
 }