Fixed #24013 -- Fixed escaping of reverse() prefix.

Prefix was treated as a part of the url pattern, which it is not.
Improved tests to conform with RFC 3986 which allows certain
characters in path segments without being escaped.

django/core/urlresolvers.py

@@ -453,16 +453,15 @@ class RegexURLResolver(LocaleRegexProvider):
                 )
         possibilities = self.reverse_dict.getlist(lookup_view)
 
-        prefix_norm, prefix_args = normalize(urlquote(_prefix))[0]
         for possibility, pattern, defaults in possibilities:
             for result, params in possibility:
                 if args:
-                    if len(args) != len(params) + len(prefix_args):
+                    if len(args) != len(params):
                         continue
-                    candidate_subs = dict(zip(prefix_args + params, text_args))
+                    candidate_subs = dict(zip(params, text_args))
                 else:
                     if (set(kwargs.keys()) | set(defaults.keys()) != set(params) |
-                            set(defaults.keys()) | set(prefix_args)):
+                            set(defaults.keys())):
                         continue
                     matches = True
                     for k, v in defaults.items():
@@ -477,12 +476,10 @@ class RegexURLResolver(LocaleRegexProvider):
                 # without quoting to build a decoded URL and look for a match.
                 # Then, if we have a match, redo the substitution with quoted
                 # arguments in order to return a properly encoded URL.
-                candidate_pat = prefix_norm.replace('%', '%%') + result
-                if re.search('^%s%s' % (prefix_norm, pattern), candidate_pat % candidate_subs, re.UNICODE):
+                candidate_pat = _prefix.replace('%', '%%') + result
+                if re.search('^%s%s' % (re.escape(_prefix), pattern), candidate_pat % candidate_subs, re.UNICODE):
                     # safe characters from `pchar` definition of RFC 3986
-                    candidate_subs = dict((k, urlquote(v, safe=RFC3986_SUBDELIMS + str('/~:@')))
-                                          for (k, v) in candidate_subs.items())
-                    url = candidate_pat % candidate_subs
+                    url = urlquote(candidate_pat % candidate_subs, safe=RFC3986_SUBDELIMS + str('/~:@'))
                     # Don't allow construction of scheme relative urls.
                     if url.startswith('//'):
                         url = '/%%2F%s' % url[2:]

tests/urlpatterns_reverse/tests.py

@@ -202,17 +202,24 @@ class URLPatternReverse(TestCase):
                reverse('non_path_include', prefix='/{{invalid}}/'))
 
     def test_prefix_parenthesis(self):
-        self.assertEqual('/bogus%29/includes/non_path_include/',
-               reverse('non_path_include', prefix='/bogus)/'))
+        # Parentheses are allowed and should not cause errors or be escaped
+        self.assertEqual(
+            '/bogus)/includes/non_path_include/',
+            reverse('non_path_include', prefix='/bogus)/')
+        )
+        self.assertEqual(
+            '/(bogus)/includes/non_path_include/',
+            reverse('non_path_include', prefix='/(bogus)/')
+        )
 
     def test_prefix_format_char(self):
         self.assertEqual('/bump%2520map/includes/non_path_include/',
                reverse('non_path_include', prefix='/bump%20map/'))
 
     def test_non_urlsafe_prefix_with_args(self):
-        # Regression for #20022
-        self.assertEqual('/%7Eme/places/1/',
-                reverse('places', args=[1], prefix='/~me/'))
+        # Regression for #20022, adjusted for #24013 because ~ is an unreserved
+        # character. Tests whether % is escaped.
+        self.assertEqual('/%257Eme/places/1/', reverse('places', args=[1], prefix='/%7Eme/'))
 
     def test_patterns_reported(self):
         # Regression for #17076
@@ -227,6 +234,12 @@ class URLPatternReverse(TestCase):
             # exception
             self.fail("Expected a NoReverseMatch, but none occurred.")
 
+    def test_script_name_escaping(self):
+        self.assertEqual(
+            reverse('optional', args=['foo:bar'], prefix='/script:name/'),
+            '/script:name/optional/foo:bar/'
+        )
+
     def test_reverse_returns_unicode(self):
         name, expected, args, kwargs = test_data[0]
         self.assertIsInstance(