Reject some previuosly-escaped chars in hostnames.

url/url_canon_unittest.cc

 // Copyright 2013 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include <errno.h>
 #include <stddef.h>
 
 #include "base/macros.h"
 #include "base/strings/utf_string_conversions.h"
 #include "testing/gtest/include/gtest/gtest.h"
@@ skipping 276 matching lines @@
 
   EXPECT_EQ(std::string(":"), out_str);
   EXPECT_EQ(0, out_comp.begin);
   EXPECT_EQ(0, out_comp.len);
 }
 
 TEST(URLCanonTest, Host) {
   IPAddressCase host_cases[] = {
        // Basic canonicalization, uppercase should be converted to lowercase.
     {"GoOgLe.CoM", L"GoOgLe.CoM", "google.com", Component(0, 10), CanonHostInfo::NEUTRAL, -1, ""},
-      // Spaces and some other characters should be escaped.
-    {"Goo%20 goo%7C|.com", L"Goo%20 goo%7C|.com", "goo%20%20goo%7C%7C.com", Component(0, 22), CanonHostInfo::NEUTRAL, -1, ""},
+      // Spaces and some other characters should fail (used to be allowed).
+    {"Goo%20 goo%7C|.com", L"Goo%20 goo%7C|.com", "goo%20%20goo%7C%7C.com", Component(0, 22), CanonHostInfo::BROKEN, -1, ""},
       // Exciting different types of spaces!
-    {NULL, L"GOO\x00a0\x3000goo.com", "goo%20%20goo.com", Component(0, 16), CanonHostInfo::NEUTRAL, -1, ""},
+    {NULL, L"GOO\x00a0\x3000goo.com", "goo%20%20goo.com", Component(0, 16), CanonHostInfo::BROKEN, -1, ""},
       // Other types of space (no-break, zero-width, zero-width-no-break) are
       // name-prepped away to nothing.
     {NULL, L"GOO\x200b\x2060\xfeffgoo.com", "googoo.com", Component(0, 10), CanonHostInfo::NEUTRAL, -1, ""},
       // Ideographic full stop (full-width period for Chinese, etc.) should be
       // treated as a dot.
     {NULL, L"www.foo\x3002" L"bar.com", "www.foo.bar.com", Component(0, 15), CanonHostInfo::NEUTRAL, -1, ""},
       // Invalid unicode characters should fail...
       // ...In wide input, ICU will barf and we'll end up with the input as
       //    escaped UTF-8 (the invalid character should be replaced with the
       //    replacement character).
@@ skipping 43 matching lines @@
       // ZWNJ (U+200C) and ZWJ (U+200D) are mapped away in UTS 46 transitional
       // handling as well as in IDNA 2003.
     {"a\xe2\x80\x8c" "b\xe2\x80\x8d" "c", L"a\x200c" L"b\x200d" L"c", "abc",
       Component(0, 3), CanonHostInfo::NEUTRAL, -1, ""},
       // ZWJ between Devanagari characters is still mapped away in UTS 46
       // transitional handling. IDNA 2008 would give xn--11bo0mv54g.
     {"\xe0\xa4\x95\xe0\xa5\x8d\xe2\x80\x8d\xe0\xa4\x9c",
      L"\x915\x94d\x200d\x91c", "xn--11bo0m",
      Component(0, 10), CanonHostInfo::NEUTRAL, -1, ""},
       // Fullwidth exclamation mark is disallowed. UTS 46, table 4, row (b)
-      // However, we do allow this at the moment because we don't use
-      // STD3 rules and canonicalize full-width ASCII to ASCII.
+      // Full-width ASCII is converted to ASCII.
     {"wow\xef\xbc\x81", L"wow\xff01", "wow%21",
-      Component(0, 6), CanonHostInfo::NEUTRAL, -1, ""},
+      Component(0, 6), CanonHostInfo::BROKEN, -1, ""},
       // U+2132 (turned capital F) is disallowed. UTS 46, table 4, row (c)
       // Allowed in IDNA 2003, but the mapping changed after Unicode 3.2
     {"\xe2\x84\xb2oo", L"\x2132oo", "%E2%84%B2oo",
       Component(0, 11), CanonHostInfo::BROKEN, -1, ""},
       // U+2F868 (CJK Comp) is disallowed. UTS 46, table 4, row (d)
       // Allowed in IDNA 2003, but the mapping changed after Unicode 3.2
     {"\xf0\xaf\xa1\xa8\xe5\xa7\xbb.cn", L"\xd87e\xdc68\x59fb.cn",
       "%F0%AF%A1%A8%E5%A7%BB.cn",
       Component(0, 24), CanonHostInfo::BROKEN, -1, ""},
       // Maps uppercase letters to lower case letters. UTS 46 table 4 row (e)
@@ skipping 93 matching lines @@
     {"%30%78%63%30%2e%30%32%35%30.01", L"%30%78%63%30%2e%30%32%35%30.01",
       "192.168.0.1", Component(0, 11), CanonHostInfo::IPV4, 3,
       "C0A80001"},
     {"%30%78%63%30%2e%30%32%35%30.01%2e", L"%30%78%63%30%2e%30%32%35%30.01%2e",
       "192.168.0.1", Component(0, 11), CanonHostInfo::IPV4, 3,
       "C0A80001"},
       // Invalid escaping should trigger the regular host error handling.
     {"%3g%78%63%30%2e%30%32%35%30%2E.01", L"%3g%78%63%30%2e%30%32%35%30%2E.01", "%253gxc0.0250..01", Component(0, 17), CanonHostInfo::BROKEN, -1, ""},
       // Something that isn't exactly an IP should get treated as a host and
       // spaces escaped.
-    {"192.168.0.1 hello", L"192.168.0.1 hello", "192.168.0.1%20hello", Component(0, 19), CanonHostInfo::NEUTRAL, -1, ""},
+    {"192.168.0.1 hello", L"192.168.0.1 hello", "192.168.0.1%20hello", Component(0, 19), CanonHostInfo::BROKEN, -1, ""},
       // Fullwidth and escaped UTF-8 fullwidth should still be treated as IP.
       // These are "0Xc0.0250.01" in fullwidth.
     {"\xef\xbc\x90%Ef%bc\xb8%ef%Bd%83\xef\xbc\x90%EF%BC%8E\xef\xbc\x90\xef\xbc\x92\xef\xbc\x95\xef\xbc\x90\xef\xbc%8E\xef\xbc\x90\xef\xbc\x91", L"\xff10\xff38\xff43\xff10\xff0e\xff10\xff12\xff15\xff10\xff0e\xff10\xff11", "192.168.0.1", Component(0, 11), CanonHostInfo::IPV4, 3, "C0A80001"},
       // Broken IP addresses get marked as such.
     {"192.168.0.257", L"192.168.0.257", "192.168.0.257", Component(0, 13), CanonHostInfo::BROKEN, -1, ""},
     {"[google.com]", L"[google.com]", "[google.com]", Component(0, 12), CanonHostInfo::BROKEN, -1, ""},
       // Cyrillic letter followed by '(' should return punycode for '(' escaped
       // before punycode string was created. I.e.
       // if '(' is escaped after punycode is created we would get xn--%28-8tb
       // (incorrect).
     {"\xd1\x82(", L"\x0442(", "xn--%28-7ed", Component(0, 11),
-      CanonHostInfo::NEUTRAL, -1, ""},
+      CanonHostInfo::BROKEN, -1, ""},
       // Address with all hexidecimal characters with leading number of 1<<32
       // or greater and should return NEUTRAL rather than BROKEN if not all
       // components are numbers.
     {"12345678912345.de", L"12345678912345.de", "12345678912345.de", Component(0, 17), CanonHostInfo::NEUTRAL, -1, ""},
     {"1.12345678912345.de", L"1.12345678912345.de", "1.12345678912345.de", Component(0, 19), CanonHostInfo::NEUTRAL, -1, ""},
     {"12345678912345.12345678912345.de", L"12345678912345.12345678912345.de", "12345678912345.12345678912345.de", Component(0, 32), CanonHostInfo::NEUTRAL, -1, ""},
     {"1.2.0xB3A73CE5B59.de", L"1.2.0xB3A73CE5B59.de", "1.2.0xb3a73ce5b59.de", Component(0, 20), CanonHostInfo::NEUTRAL, -1, ""},
     {"12345678912345.0xde", L"12345678912345.0xde", "12345678912345.0xde", Component(0, 19), CanonHostInfo::BROKEN, -1, ""},
   };
 
@@ skipping 1746 matching lines @@
   };
 
   for (auto& test_case : cases) {
     SCOPED_TRACE(test_case.scheme);
     EXPECT_EQ(test_case.expected_port,
               DefaultPortForScheme(test_case.scheme, strlen(test_case.scheme)));
   }
 }
 
 }  // namespace url