Address Inada review + refactor test location

Author: StanFromIreland

Lib/test/multibytecodec_support.py

@@ -282,6 +282,22 @@ def test_incrementalencoder_del_segfault(self):
         with self.assertRaises(AttributeError):
             del e.errors
 
+    def test_null_terminator(self):
+        # see gh-101828
+        if any(enc in self.encoding for enc in ('shift', 'euc_jis')):
+            text = "バルーンフルーツ"
+        else:
+            text = "Spam"
+        encode_w_null = (text + "\0").encode(self.encoding)
+        encode_plus_null = text.encode(self.encoding) + "\0".encode(self.encoding)
+        self.assertTrue(encode_w_null.endswith(b'\x00'))
+        self.assertEqual(encode_w_null, encode_plus_null)
+
+        encode_w_null_2 = (text + "\0" + text + "\0").encode(self.encoding)
+        encode_plus_null_2 = encode_plus_null + encode_plus_null
+        self.assertEqual(encode_w_null_2.count(b'\x00'), 2)
+        self.assertEqual(encode_w_null_2, encode_plus_null_2)
+
 
 class TestBase_Mapping(unittest.TestCase):
     pass_enctest = []

Lib/test/test_codecencodings_jp.py

@@ -106,26 +106,6 @@ class Test_SJIS_2004(multibytecodec_support.TestBase, unittest.TestCase):
         b"\x85Gℜ\x85Q = ⟨ሴ⟩"
     )
 
-    def test_null_terminator(self):
-        # see gh-101828
-        cases = (
-            "バルーンフルーツ",
-            "ライフアップキノコ",
-            "テスト",
-            "'Tis but a scratch!"
-        )
-        for case in cases:
-            with self.subTest(case=case):
-                encode_w_null = (case + "\0").encode(self.encoding)
-                encode_plus_null = case.encode(self.encoding) + "\0".encode(self.encoding)
-                self.assertTrue(encode_w_null.endswith(b'\x00'))
-                self.assertEqual(encode_w_null, encode_plus_null)
-
-                encode_w_null_2 = encode_w_null + encode_w_null
-                encode_plus_null_2 = encode_plus_null + encode_plus_null
-                self.assertEqual(encode_w_null_2.count(b'\x00'), 2)
-                self.assertEqual(encode_w_null_2, encode_plus_null_2)
-
 class Test_SJISX0213(multibytecodec_support.TestBase, unittest.TestCase):
     encoding = 'shift_jisx0213'
     tstring = multibytecodec_support.load_teststring('shift_jisx0213')
@@ -142,25 +122,5 @@ class Test_SJISX0213(multibytecodec_support.TestBase, unittest.TestCase):
         b"\x85Gℜ\x85Q = ⟨ሴ⟩"
     )
 
-    def test_null_terminator(self):
-        # see gh-101828
-        cases = (
-            "バルーンフルーツ",
-            "ライフアップキノコ",
-            "テスト",
-            "'Tis but a scratch!"
-        )
-        for case in cases:
-            with self.subTest(case=case):
-                encode_w_null = (case + "\0").encode(self.encoding)
-                encode_plus_null = case.encode(self.encoding) + "\0".encode(self.encoding)
-                self.assertTrue(encode_w_null.endswith(b'\x00'))
-                self.assertEqual(encode_w_null, encode_plus_null)
-
-                encode_w_null_2 = encode_w_null + encode_w_null
-                encode_plus_null_2 = encode_plus_null + encode_plus_null
-                self.assertEqual(encode_w_null_2.count(b'\x00'), 2)
-                self.assertEqual(encode_w_null_2, encode_plus_null_2)
-
 if __name__ == "__main__":
     unittest.main()

Misc/NEWS.d/next/Library/2025-09-25-20-16-10.gh-issue-101828.yTxJlJ.rst

@@ -1,2 +1,3 @@
-Fix ``'shift_jisx0213'`` and ``'shift_jis_2004'`` codecs truncating null chars
+Fix ``'shift_jisx0213'``, ``'shift_jis_2004'``, ``'euc_jisx0213'`` and
+``'euc_jis_2004'`` codecs truncating null chars
 as they were treated as part of multi-character sequences.

Modules/cjkcodecs/_codecs_jp.c

@@ -192,8 +192,11 @@ ENCODER(euc_jis_2004)
                                 JISX0213_ENCPAIRS);
                             if (code == DBCINV)
                                 return 1;
-                        } else
+                        }
+                        else if (c2 != 0) {
+                            /* Don't consume null char as part of pair */
                             insize = 2;
+                        }
                     }
                 }
             }
@@ -612,17 +615,8 @@ ENCODER(shift_jis_2004)
                                 return 1;
                             }
                             else if (ch2 != 0) {
-                                insize = 2;
-                            }
-                            else {
                                 /* Don't consume null char as part of pair */
-                                code = find_pairencmap(
-                                    (ucs2_t)c, 0,
-                                    jisx0213_pair_encmap,
-                                    JISX0213_ENCPAIRS);
-                                if (code == DBCINV) {
-                                    return 1;
-                                }
+                                insize = 2;
                             }
                         }
                     }