python
diff --git a/‎Lib/test/test_unicodedata.py
Lines changed: 23 additions & 0 deletions b/‎Lib/test/test_unicodedata.py
Lines changed: 23 additions & 0 deletions
diff --git a/‎Misc/NEWS.d/next/Library/2025-02-02-16-30-27.gh-issue-129569.i0kPOG.rst
Lines changed: 1 addition & 0 deletions b/‎Misc/NEWS.d/next/Library/2025-02-02-16-30-27.gh-issue-129569.i0kPOG.rst
Lines changed: 1 addition & 0 deletions
diff --git a/‎Modules/unicodedata.c
Lines changed: 5 additions & 5 deletions b/‎Modules/unicodedata.c
Lines changed: 5 additions & 5 deletions
@@ -467,6 +467,29 @@ def test_bug_834676(self):
         # Check for bug 834676
         unicodedata.normalize('NFC', '\ud55c\uae00')
 
+    def test_normalize_return_type(self):
+        # gh-129569: normalize() return type must always be str
+        normalize = unicodedata.normalize
+
+        class MyStr(str):
+            pass
+
+        normalization_forms = ("NFC", "NFKC", "NFD", "NFKD")
+        input_strings = (
+            # normalized strings
+            "",
+            "ascii",
+            # unnormalized strings
+            "\u1e0b\u0323",
+            "\u0071\u0307\u0323",
+        )
+
+        for form in normalization_forms:
+            for input_str in input_strings:
+                with self.subTest(form=form, input_str=input_str):
+                    self.assertIs(type(normalize(form, input_str)), str)
+                    self.assertIs(type(normalize(form, MyStr(input_str))), str)
+
 
 if __name__ == "__main__":
     unittest.main()
@@ -0,0 +1 @@
+Fix :func:`unicodedata.normalize` to always return a built-in :class:`str` object when given an input of a :class:`str` subclass, regardless of whether the string is already normalized.
@@ -933,34 +933,34 @@ unicodedata_UCD_normalize_impl(PyObject *self, PyObject *form,
     if (PyUnicode_GET_LENGTH(input) == 0) {
         /* Special case empty input strings, since resizing
            them  later would cause internal errors. */
-        return Py_NewRef(input);
+        return PyUnicode_FromObject(input);
     }
 
     if (PyUnicode_CompareWithASCIIString(form, "NFC") == 0) {
         if (is_normalized_quickcheck(self, input,
                                      true,  false, true) == YES) {
-            return Py_NewRef(input);
+            return PyUnicode_FromObject(input);
         }
         return nfc_nfkc(self, input, 0);
     }
     if (PyUnicode_CompareWithASCIIString(form, "NFKC") == 0) {
         if (is_normalized_quickcheck(self, input,
                                      true,  true,  true) == YES) {
-            return Py_NewRef(input);
+            return PyUnicode_FromObject(input);
         }
         return nfc_nfkc(self, input, 1);
     }
     if (PyUnicode_CompareWithASCIIString(form, "NFD") == 0) {
         if (is_normalized_quickcheck(self, input,
                                      false, false, true) == YES) {
-            return Py_NewRef(input);
+            return PyUnicode_FromObject(input);
         }
         return nfd_nfkd(self, input, 0);
     }
     if (PyUnicode_CompareWithASCIIString(form, "NFKD") == 0) {
         if (is_normalized_quickcheck(self, input,
                                      false, true,  true) == YES) {
-            return Py_NewRef(input);
+            return PyUnicode_FromObject(input);
         }
         return nfd_nfkd(self, input, 1);
     }
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	+Fix :func:`unicodedata.normalize` to always return a built-in :class:`str` object when given an input of a :class:`str` subclass, regardless of whether the string is already normalized.
Original file line number	Diff line number	Diff line change
`@@ -933,34 +933,34 @@ unicodedata_UCD_normalize_impl(PyObject self, PyObject form,`
`933`	`933`	`if (PyUnicode_GET_LENGTH(input) == 0) {`
`934`	`934`	`/* Special case empty input strings, since resizing`
`935`	`935`	`them later would cause internal errors. */`
`936`		`- return Py_NewRef(input);`
	`936`	`+ return PyUnicode_FromObject(input);`
`937`	`937`	`}`
`938`	`938`
`939`	`939`	`if (PyUnicode_CompareWithASCIIString(form, "NFC") == 0) {`
`940`	`940`	`if (is_normalized_quickcheck(self, input,`
`941`	`941`	`true, false, true) == YES) {`
`942`		`- return Py_NewRef(input);`
	`942`	`+ return PyUnicode_FromObject(input);`
`943`	`943`	`}`
`944`	`944`	`return nfc_nfkc(self, input, 0);`
`945`	`945`	`}`
`946`	`946`	`if (PyUnicode_CompareWithASCIIString(form, "NFKC") == 0) {`
`947`	`947`	`if (is_normalized_quickcheck(self, input,`
`948`	`948`	`true, true, true) == YES) {`
`949`		`- return Py_NewRef(input);`
	`949`	`+ return PyUnicode_FromObject(input);`
`950`	`950`	`}`
`951`	`951`	`return nfc_nfkc(self, input, 1);`
`952`	`952`	`}`
`953`	`953`	`if (PyUnicode_CompareWithASCIIString(form, "NFD") == 0) {`
`954`	`954`	`if (is_normalized_quickcheck(self, input,`
`955`	`955`	`false, false, true) == YES) {`
`956`		`- return Py_NewRef(input);`
	`956`	`+ return PyUnicode_FromObject(input);`
`957`	`957`	`}`
`958`	`958`	`return nfd_nfkd(self, input, 0);`
`959`	`959`	`}`
`960`	`960`	`if (PyUnicode_CompareWithASCIIString(form, "NFKD") == 0) {`
`961`	`961`	`if (is_normalized_quickcheck(self, input,`
`962`	`962`	`false, true, true) == YES) {`
`963`		`- return Py_NewRef(input);`
	`963`	`+ return PyUnicode_FromObject(input);`
`964`	`964`	`}`
`965`	`965`	`return nfd_nfkd(self, input, 1);`
`966`	`966`	`}`