bpo-30566: Fix IndexError when using punycode codec (GH-18632)

Trying to decode an invalid string with the punycode codec
shoud raise UnicodeError.

Author: berkerpeksag

Lib/encodings/punycode.py

@@ -143,7 +143,7 @@ def decode_generalized_number(extended, extpos, bias, errors):
             digit = char - 22 # 0x30-26
         elif errors == "strict":
             raise UnicodeError("Invalid extended code point '%s'"
-                               % extended[extpos])
+                               % extended[extpos-1])
         else:
             return extpos, None
         t = T(j, bias)

Lib/test/test_codecs.py

@@ -1343,6 +1343,18 @@ def test_decode(self):
             puny = puny.decode("ascii").encode("ascii")
             self.assertEqual(uni, puny.decode("punycode"))
 
+    def test_decode_invalid(self):
+        testcases = [
+            (b"xn--w&", "strict", UnicodeError()),
+            (b"xn--w&", "ignore", "xn-"),
+        ]
+        for puny, errors, expected in testcases:
+            with self.subTest(puny=puny, errors=errors):
+                if isinstance(expected, Exception):
+                    self.assertRaises(UnicodeError, puny.decode, "punycode", errors)
+                else:
+                    self.assertEqual(puny.decode("punycode", errors), expected)
+
 
 # From http://www.gnu.org/software/libidn/draft-josefsson-idn-test-vectors.html
 nameprep_tests = [

Misc/NEWS.d/next/Library/2020-02-24-03-45-28.bpo-30566.qROxty.rst

@@ -0,0 +1,2 @@
+Fix :exc:`IndexError` when trying to decode an invalid string with punycode
+codec.