gh-126004: Fix positions handling in `codecs.replace_errors` (#127674)

This fixes how `PyCodec_ReplaceErrors` handles the `start` and `end` attributes
of `UnicodeError` objects via the `_PyUnicodeError_GetParams` helper.

diff --git a/Lib/test/test_capi/test_codecs.py b/Lib/test/test_capi/test_codecs.py
index 3e79dd2f7ca2fab4b6e5686e06b8ef8859923803..f57191ddcdbeb41ce0c69887ce7f194c36e386bd 100644 (file)
--- a/Lib/test/test_capi/test_codecs.py
+++ b/Lib/test/test_capi/test_codecs.py
@@ -839,7 +839,8 @@ class CAPICodecErrors(unittest.TestCase):
 
     def test_codec_replace_errors_handler(self):
         handler = _testcapi.codec_replace_errors
-        self.do_test_codec_errors_handler(handler, self.all_unicode_errors)
+        self.do_test_codec_errors_handler(handler, self.all_unicode_errors,
+                                          safe=True)
 
     def test_codec_xmlcharrefreplace_errors_handler(self):
         handler = _testcapi.codec_xmlcharrefreplace_errors

diff --git a/Misc/NEWS.d/next/Core_and_Builtins/2024-12-06-11-32-58.gh-issue-126004.CYAwTB.rst b/Misc/NEWS.d/next/Core_and_Builtins/2024-12-06-11-32-58.gh-issue-126004.CYAwTB.rst
new file mode 100644 (file)
index 0000000..de70c59
--- /dev/null
+++ b/Misc/NEWS.d/next/Core_and_Builtins/2024-12-06-11-32-58.gh-issue-126004.CYAwTB.rst
@@ -0,0 +1,3 @@
+Fix handling of :attr:`UnicodeError.start` and :attr:`UnicodeError.end`
+values in the :func:`codecs.replace_errors` error handler. Patch by Bénédikt
+Tran.

diff --git a/Python/codecs.c b/Python/codecs.c
index 11eaca175abf138c80df2dde1734b3faaaab922f..b657dd134a668e0afd49968a698334156eee0330 100644 (file)
--- a/Python/codecs.c
+++ b/Python/codecs.c
@@ -702,48 +702,46 @@ PyObject *PyCodec_IgnoreErrors(PyObject *exc)
 
 PyObject *PyCodec_ReplaceErrors(PyObject *exc)
 {
-    Py_ssize_t start, end, i, len;
+    Py_ssize_t start, end, slen;
 
     if (PyObject_TypeCheck(exc, (PyTypeObject *)PyExc_UnicodeEncodeError)) {
-        PyObject *res;
-        Py_UCS1 *outp;
-        if (PyUnicodeEncodeError_GetStart(exc, &start))
+        if (_PyUnicodeError_GetParams(exc, NULL, NULL,
+                                      &start, &end, &slen, false) < 0) {
             return NULL;
-        if (PyUnicodeEncodeError_GetEnd(exc, &end))
-            return NULL;
-        len = end - start;
-        res = PyUnicode_New(len, '?');
-        if (res == NULL)
+        }
+        PyObject *res = PyUnicode_New(slen, '?');
+        if (res == NULL) {
             return NULL;
+        }
         assert(PyUnicode_KIND(res) == PyUnicode_1BYTE_KIND);
-        outp = PyUnicode_1BYTE_DATA(res);
-        for (i = 0; i < len; ++i)
-            outp[i] = '?';
+        Py_UCS1 *outp = PyUnicode_1BYTE_DATA(res);
+        memset(outp, '?', sizeof(Py_UCS1) * slen);
         assert(_PyUnicode_CheckConsistency(res, 1));
         return Py_BuildValue("(Nn)", res, end);
     }
     else if (PyObject_TypeCheck(exc, (PyTypeObject *)PyExc_UnicodeDecodeError)) {
-        if (PyUnicodeDecodeError_GetEnd(exc, &end))
+        if (_PyUnicodeError_GetParams(exc, NULL, NULL,
+                                      NULL, &end, NULL, true) < 0) {
             return NULL;
+        }
         return Py_BuildValue("(Cn)",
                              (int)Py_UNICODE_REPLACEMENT_CHARACTER,
                              end);
     }
     else if (PyObject_TypeCheck(exc, (PyTypeObject *)PyExc_UnicodeTranslateError)) {
-        PyObject *res;
-        Py_UCS2 *outp;
-        if (PyUnicodeTranslateError_GetStart(exc, &start))
-            return NULL;
-        if (PyUnicodeTranslateError_GetEnd(exc, &end))
+        if (_PyUnicodeError_GetParams(exc, NULL, NULL,
+                                      &start, &end, &slen, false) < 0) {
             return NULL;
-        len = end - start;
-        res = PyUnicode_New(len, Py_UNICODE_REPLACEMENT_CHARACTER);
-        if (res == NULL)
+        }
+        PyObject *res = PyUnicode_New(slen, Py_UNICODE_REPLACEMENT_CHARACTER);
+        if (res == NULL) {
             return NULL;
-        assert(PyUnicode_KIND(res) == PyUnicode_2BYTE_KIND);
-        outp = PyUnicode_2BYTE_DATA(res);
-        for (i = 0; i < len; i++)
+        }
+        assert(slen == 0 || PyUnicode_KIND(res) == PyUnicode_2BYTE_KIND);
+        Py_UCS2 *outp = PyUnicode_2BYTE_DATA(res);
+        for (Py_ssize_t i = 0; i < slen; ++i) {
             outp[i] = Py_UNICODE_REPLACEMENT_CHARACTER;
+        }
         assert(_PyUnicode_CheckConsistency(res, 1));
         return Py_BuildValue("(Nn)", res, end);
     }