test that static strings (_PyRuntime.static_objects.singletons.strings, not strings allocated by deepfreeze) are reused.

Ref: python/cpython#103876

Author: oraluben

src/_cds/_cdsmodule.c

@@ -442,6 +442,20 @@ PyCDS_InitMoveIn()
         PyCDS_Table_New();
     cds_status.move_in_ctx->in_heap_str_to_string_ref_list_map =
         PyCDS_Table_New();
+    cds_status.move_in_ctx->static_strings = PyDict_New();
+
+#define HANDLE_LITERAL(lit) \
+    PyDict_SetDefault(cds_status.move_in_ctx->static_strings, (lit), (lit));
+#define HANDLE_ASCII(c) \
+    PyDict_SetDefault(cds_status.move_in_ctx->static_strings, (c), (c));
+#define HANDLE_LATIN1(c) \
+    PyDict_SetDefault(cds_status.move_in_ctx->static_strings, (c), (c));
+
+#include "string_singletons.h"
+
+#undef HANDLE_LITERAL
+#undef HANDLE_ASCII
+#undef HANDLE_LATIN1
 }
 
 void
@@ -453,6 +467,9 @@ PyCDS_FinalizeMoveIn()
         cds_status.move_in_ctx->orig_pyobject_to_in_heap_pyobject_map);
     PyCDS_Table_Destroy(
         cds_status.move_in_ctx->in_heap_str_to_string_ref_list_map);
+
+    Py_DecRef(cds_status.move_in_ctx->static_strings);
+
     free(cds_status.move_in_ctx);
 }
 
@@ -578,6 +595,16 @@ _Py_COMP_DIAG_POP
             assert(PyCDS_STR_INTERNED(*target) == SSTATE_INTERNED_IMMORTAL);
         }
         else { /* String is not in archive yet. */
+            if ((*target = PyDict_GetItem(
+                     cds_status.move_in_ctx->static_strings, op)) != NULL) {
+                // `_Py_ID` are static, and interned.
+                // But `_Py_STR`, `_Py_SINGLETON(strings).{ascii,latin1}` are
+                // not interned.
+                // So we manually construct the `static_strings` dict to make
+                // archive to refer to those static strings.
+                return;
+            }
+
             PyUnicode_InternInPlace(&op);
 
             *target = _PyCDS_PyUnicode_Copy(op);
@@ -857,8 +884,7 @@ _PyCDS_PyUnicode_Copy(PyObject *op)
         data = unicode + 1;
     (((PyASCIIObject *)(unicode))->length) = size;
     (((PyASCIIObject *)(unicode))->hash) = -1;
-    // immortal bit will be set later by interning
-    PyCDS_STR_INTERNED(unicode) = SSTATE_NOT_INTERNED;
+    PyCDS_STR_INTERNED(unicode) = SSTATE_INTERNED_IMMORTAL;
     (((PyASCIIObject *)(unicode))->state).kind = kind;
     (((PyASCIIObject *)(unicode))->state).compact = 1;
     (((PyASCIIObject *)(unicode))->state).ascii = is_ascii;

src/_cds/_cdsmodule.h

@@ -76,6 +76,8 @@ struct MoveInContext {
 
     table *orig_pyobject_to_in_heap_pyobject_map;
     table *in_heap_str_to_string_ref_list_map;
+
+    PyObject *static_strings;
 };
 
 /*

src/_cds/string_singletons.h

@@ -0,0 +1,38 @@
+#ifndef PYCDS_STRING_SINGLETONS_H
+#define PYCDS_STRING_SINGLETONS_H
+
+#define LIT(lit_name)                                                   \
+    (PyObject *)(&_PyRuntime.static_objects.singletons.strings.literals \
+                      ._py_##lit_name)
+
+HANDLE_LITERAL(LIT(anon_dictcomp))
+HANDLE_LITERAL(LIT(anon_genexpr))
+HANDLE_LITERAL(LIT(anon_lambda))
+HANDLE_LITERAL(LIT(anon_listcomp))
+HANDLE_LITERAL(LIT(anon_module))
+HANDLE_LITERAL(LIT(anon_setcomp))
+HANDLE_LITERAL(LIT(anon_string))
+HANDLE_LITERAL(LIT(anon_unknown))
+HANDLE_LITERAL(LIT(close_br))
+HANDLE_LITERAL(LIT(dbl_close_br))
+HANDLE_LITERAL(LIT(dbl_open_br))
+HANDLE_LITERAL(LIT(dbl_percent))
+HANDLE_LITERAL(LIT(dot))
+HANDLE_LITERAL(LIT(dot_locals))
+HANDLE_LITERAL(LIT(empty))
+HANDLE_LITERAL(LIT(json_decoder))
+HANDLE_LITERAL(LIT(list_err))
+HANDLE_LITERAL(LIT(newline))
+HANDLE_LITERAL(LIT(open_br))
+HANDLE_LITERAL(LIT(percent))
+HANDLE_LITERAL(LIT(shim_name))
+HANDLE_LITERAL(LIT(utf_8))
+
+for (size_t i = 0; i < 128; ++i) {
+    HANDLE_ASCII(
+        (PyObject *)&_PyRuntime.static_objects.singletons.strings.ascii[i]);
+    HANDLE_LATIN1(
+        (PyObject *)&_PyRuntime.static_objects.singletons.strings.latin1[i]);
+}
+
+#endif

tests/test_cds_extension/test_share_object.py

@@ -268,8 +268,8 @@ def test_share_dynamic(self):
             self.run_check(s + random.choice(string.punctuation.replace('_', '')), False, True, True)
 
     @assert_archive_created
-    def test_share_ascii(self):
-        for i in random.sample(range(0, 128 + 1), 10):
+    def test_share_static_ascii_latin1(self):
+        for i in random.sample(range(0, 128), 10):
             c = chr(i)
             l = chr(128 + i)
             self.run_check(c, True, False, False)