From: Donghee Na <donghee.na@python.org>
Date: Fri, 2 Jan 2026 14:05:08 +0000 (+0900)
Subject: gh-139757: Fix unintended bytecode specialization for non-ascii string (gh-143352)
X-Git-Tag: v3.15.0a5~11^2~149
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=61f2ad9a3a78597bb1983c3392a758ed02709574;p=thirdparty%2FPython%2Fcpython.git

gh-139757: Fix unintended bytecode specialization for non-ascii string (gh-143352)
---

diff --git a/Lib/test/test_opcache.py b/Lib/test/test_opcache.py
index 4113b79ef5c8..4df88d0f1688 100644
--- a/Lib/test/test_opcache.py
+++ b/Lib/test/test_opcache.py
@@ -1785,6 +1785,15 @@ class TestSpecializer(TestBase):
         self.assert_specialized(binary_subscr_str_int, "BINARY_OP_SUBSCR_STR_INT")
         self.assert_no_opcode(binary_subscr_str_int, "BINARY_OP")
 
+        def binary_subscr_str_int_non_compact():
+            for _ in range(_testinternalcapi.SPECIALIZATION_THRESHOLD):
+                a = "바이트코드_특수화"
+                for idx, expected in enumerate(a):
+                    self.assertEqual(a[idx], expected)
+
+        binary_subscr_str_int_non_compact()
+        self.assert_no_opcode(binary_subscr_str_int_non_compact, "BINARY_OP_SUBSCR_STR_INT")
+
         def binary_subscr_getitems():
             class C:
                 def __init__(self, val):
diff --git a/Misc/NEWS.d/next/Core_and_Builtins/2026-01-02-22-35-12.gh-issue-139757.v5LRew.rst b/Misc/NEWS.d/next/Core_and_Builtins/2026-01-02-22-35-12.gh-issue-139757.v5LRew.rst
new file mode 100644
index 000000000000..ef1a4e4b3af2
--- /dev/null
+++ b/Misc/NEWS.d/next/Core_and_Builtins/2026-01-02-22-35-12.gh-issue-139757.v5LRew.rst
@@ -0,0 +1,2 @@
+Fix unintended bytecode specialization for non-ascii string.
+Patch by Donghee Na, Ken Jin and Chris Eibl.
diff --git a/Python/specialize.c b/Python/specialize.c
index e67078afdd9d..e9302157e778 100644
--- a/Python/specialize.c
+++ b/Python/specialize.c
@@ -2240,7 +2240,7 @@ _Py_Specialize_BinaryOp(_PyStackRef lhs_st, _PyStackRef rhs_st, _Py_CODEUNIT *in
                     specialize(instr, BINARY_OP_SUBSCR_TUPLE_INT);
                     return;
                 }
-                if (PyUnicode_CheckExact(lhs)) {
+                if (PyUnicode_CheckExact(lhs) && PyUnicode_IS_COMPACT_ASCII(lhs)) {
                     specialize(instr, BINARY_OP_SUBSCR_STR_INT);
                     return;
                 }