gh-131798: Narrow the return type of _FORMAT_SIMPLE and _FORMAT_WITH_SPEC to str for built-in types

Author: NekoAsakura

Include/internal/pycore_optimizer.h

@@ -393,6 +393,7 @@ extern JitOptRef _Py_uop_sym_new_type(
 
 extern JitOptRef _Py_uop_sym_new_const(JitOptContext *ctx, PyObject *const_val);
 extern JitOptRef _Py_uop_sym_new_const_steal(JitOptContext *ctx, PyObject *const_val);
+extern bool _Py_uop_sym_is_safe_type(JitOptRef sym);
 bool _Py_uop_sym_is_safe_const(JitOptContext *ctx, JitOptRef sym);
 _PyStackRef _Py_uop_sym_get_const_as_stackref(JitOptContext *ctx, JitOptRef sym);
 extern JitOptRef _Py_uop_sym_new_null(JitOptContext *ctx);

Misc/NEWS.d/next/Core_and_Builtins/2026-03-30-17-01-34.gh-issue-131798.WSefcr.rst

@@ -1,2 +1,2 @@
 Allow the JIT to remove unicode guards after ``_FORMAT_SIMPLE`` and
-``_FORMAT_WITH_SPEC`` by setting the return type to string.
+``_FORMAT_WITH_SPEC`` when the input type is a known built-in type.

Python/optimizer_analysis.c

@@ -250,6 +250,7 @@ add_op(JitOptContext *ctx, _PyUOpInstruction *this_instr,
 /* Shortened forms for convenience, used in optimizer_bytecodes.c */
 #define sym_is_not_null _Py_uop_sym_is_not_null
 #define sym_is_const _Py_uop_sym_is_const
+#define sym_is_safe_type _Py_uop_sym_is_safe_type
 #define sym_is_safe_const _Py_uop_sym_is_safe_const
 #define sym_get_const _Py_uop_sym_get_const
 #define sym_new_const_steal _Py_uop_sym_new_const_steal

Python/optimizer_bytecodes.c

@@ -1552,11 +1552,19 @@ dummy_func(void) {
     }
 
     op(_FORMAT_SIMPLE, (value -- res)) {
-        res = sym_new_type(ctx, &PyUnicode_Type);
+        if (sym_is_safe_type(value)) {
+            res = sym_new_type(ctx, &PyUnicode_Type);
+        } else {
+            res = sym_new_not_null(ctx);
+        }
     }
 
     op(_FORMAT_WITH_SPEC, (value, fmt_spec -- res)) {
-        res = sym_new_type(ctx, &PyUnicode_Type);
+        if (sym_is_safe_type(value)) {
+            res = sym_new_type(ctx, &PyUnicode_Type);
+        } else {
+            res = sym_new_not_null(ctx);
+        }
     }
 
     op(_SET_UPDATE, (set, unused[oparg-1], iterable -- set, unused[oparg-1], i)) {

Python/optimizer_cases.c.h

@@ -264,6 +264,25 @@ _Py_uop_sym_get_const_as_stackref(JitOptContext *ctx, JitOptRef sym)
     return PyStackRef_FromPyObjectBorrow(const_val);
 }
 
+/*
+ Indicates whether the type is a known built-in type
+ that is safe to narrow.
+ */
+bool
+_Py_uop_sym_is_safe_type(JitOptRef sym)
+{
+    PyTypeObject *typ = _Py_uop_sym_get_type(sym);
+    if (typ == NULL) {
+        return false;
+    }
+    return (typ == &PyLong_Type) ||
+           (typ == &PyUnicode_Type) ||
+           (typ == &PyFloat_Type) ||
+           (typ == &_PyNone_Type) ||
+           (typ == &PyBool_Type) ||
+           (typ == &PyFrozenDict_Type);
+}
+
 /*
  Indicates whether the constant is safe to constant evaluate
  (without side effects).