[2.7] bpo-36742: Fix urlparse.urlsplit() error message for Unicode URL (GH-13937)

If urlparse.urlsplit() detects an invalid netloc according to NFKC
normalization, the error message type is now str rather than unicode,
and use repr() to format the URL, to prevent <exception str() failed>
when display the error message.

Author: vstinner

Lib/test/test_urlparse.py

@@ -656,6 +656,15 @@ def test_urlsplit_normalization(self):
                     with self.assertRaises(ValueError):
                         urlparse.urlsplit(url)
 
+        # check error message: invalid netloc must be formated with repr()
+        # to get an ASCII error message
+        with self.assertRaises(ValueError) as cm:
+            urlparse.urlsplit(u'http://example.com\uFF03@bing.com')
+        self.assertEqual(str(cm.exception),
+                         "netloc u'example.com\\[email protected]' contains invalid characters "
+                         "under NFKC normalization")
+        self.assertIsInstance(cm.exception.args[0], str)
+
 def test_main():
     test_support.run_unittest(UrlParseTestCase)
 

Lib/urlparse.py

@@ -180,8 +180,9 @@ def _checknetloc(netloc):
         return
     for c in '/?#@:':
         if c in netloc2:
-            raise ValueError(u"netloc '" + netloc + u"' contains invalid " +
-                             u"characters under NFKC normalization")
+            raise ValueError("netloc %r contains invalid characters "
+                             "under NFKC normalization"
+                             % netloc)
 
 def urlsplit(url, scheme='', allow_fragments=True):
     """Parse a URL into 5 components:

Misc/NEWS.d/next/Library/2019-06-10-12-02-45.bpo-36742.UEdHXJ.rst

@@ -0,0 +1,3 @@
+:func:`urlparse.urlsplit` error message for invalid ``netloc`` according to
+NFKC normalization is now a :class:`str` string, rather than a
+:class:`unicode` string, to prevent error when displaying the error.