Added characters that look like padlocks to net IDN character blacklist.

net/base/net_util_icu.cc

 // Copyright (c) 2012 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 "net/base/net_util.h"
 
 #include <map>
 #include <vector>
 
 #include "base/i18n/time_formatting.h"
@@ skipping 176 matching lines @@
   // For now, we borrow the list from Mozilla and tweaked it slightly.
   // (e.g. Characters like U+00A0, U+3000, U+3002 are omitted because
   //  they're gonna be canonicalized to U+0020 and full stop before
   //  reaching here.)
   // The original list is available at
   // http://kb.mozillazine.org/Network.IDN.blacklist_chars and
   // at http://mxr.mozilla.org/seamonkey/source/modules/libpref/src/init/all.js#703
 
   UErrorCode status = U_ZERO_ERROR;
 #ifdef U_WCHAR_IS_UTF16
-  icu::UnicodeSet dangerous_characters(icu::UnicodeString(
-      L"[[\\ \u00ad\u00bc\u00bd\u01c3\u0337\u0338"
-      L"\u05c3\u05f4\u06d4\u0702\u115f\u1160][\u2000-\u200b]"
-      L"[\u2024\u2027\u2028\u2029\u2039\u203a\u2044\u205f]"
-      L"[\u2154-\u2156][\u2159-\u215b][\u215f\u2215\u23ae"
-      L"\u29f6\u29f8\u2afb\u2afd][\u2ff0-\u2ffb][\u3014"
-      L"\u3015\u3033\u3164\u321d\u321e\u33ae\u33af\u33c6\u33df\ufe14"
-      L"\ufe15\ufe3f\ufe5d\ufe5e\ufeff\uff0e\uff06\uff61\uffa0\ufff9]"
-      L"[\ufffa-\ufffd]]"), status);
+  icu::UnicodeSet dangerous_characters(
+      icu::UnicodeString(
+          L"[[\\ \u00ad\u00bc\u00bd\u01c3\u0337\u0338"
+          L"\u05c3\u05f4\u06d4\u0702\u115f\u1160][\u2000-\u200b]"
+          L"[\u2024\u2027\u2028\u2029\u2039\u203a\u2044\u205f]"
+          L"[\u2154-\u2156][\u2159-\u215b][\u215f\u2215\u23ae"
+          L"\u29f6\u29f8\u2afb\u2afd][\u2ff0-\u2ffb][\u3014"
+          L"\u3015\u3033\u3164\u321d\u321e\u33ae\u33af\u33c6\u33df\ufe14"
+          L"\ufe15\ufe3f\ufe5d\ufe5e\ufeff\uff0e\uff06\uff61\uffa0\ufff9]"
+          L"[\ufffa-\ufffd]\U0001f50f\U0001f510\U0001f512\U0001f513]"),
+      status);
   DCHECK(U_SUCCESS(status));
   icu::RegexMatcher dangerous_patterns(icu::UnicodeString(
       // Lone katakana no, so, or n
       L"[^\\p{Katakana}][\u30ce\u30f3\u30bd][^\\p{Katakana}]"
       // Repeating Japanese accent characters
       L"|[\u3099\u309a\u309b\u309c][\u3099\u309a\u309b\u309c]"),
       0, status);
 #else
   icu::UnicodeSet dangerous_characters(icu::UnicodeString(
       "[[\\u0020\\u00ad\\u00bc\\u00bd\\u01c3\\u0337\\u0338"
       "\\u05c3\\u05f4\\u06d4\\u0702\\u115f\\u1160][\\u2000-\\u200b]"
       "[\\u2024\\u2027\\u2028\\u2029\\u2039\\u203a\\u2044\\u205f]"
       "[\\u2154-\\u2156][\\u2159-\\u215b][\\u215f\\u2215\\u23ae"
       "\\u29f6\\u29f8\\u2afb\\u2afd][\\u2ff0-\\u2ffb][\\u3014"
       "\\u3015\\u3033\\u3164\\u321d\\u321e\\u33ae\\u33af\\u33c6\\u33df\\ufe14"
       "\\ufe15\\ufe3f\\ufe5d\\ufe5e\\ufeff\\uff0e\\uff06\\uff61\\uffa0\\ufff9]"
-      "[\\ufffa-\\ufffd]]", -1, US_INV), status);
+      "[\\ufffa-\\ufffd]\\U0001f50f\\U0001f510\\U0001f512\\U0001f513]", -1,
+      US_INV), status);
   DCHECK(U_SUCCESS(status));
   icu::RegexMatcher dangerous_patterns(icu::UnicodeString(
       // Lone katakana no, so, or n
       "[^\\p{Katakana}][\\u30ce\\u30f3\\u30bd][^\\p{Katakana}]"
       // Repeating Japanese accent characters
       "|[\\u3099\\u309a\\u309b\\u309c][\\u3099\\u309a\\u309b\\u309c]"),
       0, status);
 #endif
   DCHECK(U_SUCCESS(status));
   icu::UnicodeSet component_characters;
@@ skipping 590 matching lines @@
   if (offset_for_adjustment)
     offsets.push_back(*offset_for_adjustment);
   base::string16 result = FormatUrlWithOffsets(url, languages, format_types,
       unescape_rules, new_parsed, prefix_end, &offsets);
   if (offset_for_adjustment)
     *offset_for_adjustment = offsets[0];
   return result;
 }
 
 }  // namespace net