Add URL filter to trigger punycode for Japanese homographic sequences....

net/base/net_util.cc

 // Copyright (c) 2010 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 <algorithm>
 #include <map>
+#include <unicode/regex.h>
 #include <unicode/ucnv.h>
 #include <unicode/uidna.h>
 #include <unicode/ulocdata.h>
 #include <unicode/uniset.h>
 #include <unicode/uscript.h>
 #include <unicode/uset.h>
 
 #include "build/build_config.h"
 
 #if defined(OS_WIN)
@@ skipping 595 matching lines @@
 #ifdef U_WCHAR_IS_UTF16
   icu::UnicodeSet dangerous_characters(icu::UnicodeString(
       L"[[\\ \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);
+  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\\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);
+  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;
-  component_characters.addAll(icu::UnicodeString(str, str_len));
+  icu::UnicodeString component_string(str, str_len);
+  component_characters.addAll(component_string);
   if (dangerous_characters.containsSome(component_characters))
     return false;
 
+  DCHECK(U_SUCCESS(status));
+  dangerous_patterns.reset(component_string);
+  if (dangerous_patterns.find())
+    return false;
+
   // If the language list is empty, the result is completely determined
   // by whether a component is a single script or not. This will block
   // even "safe" script mixing cases like <Chinese, Latin-ASCII> that are
   // allowed with |languages| (while it blocks Chinese + Latin letters with
   // an accent as should be the case), but we want to err on the safe side
   // when |languages| is empty.
   if (languages.empty())
     return IsIDNComponentInSingleScript(str, str_len);
 
   // |common_characters| is made up of  ASCII numbers, hyphen, plus and
@@ skipping 1265 matching lines @@
 }
 
 int GetPortFromAddrinfo(const struct addrinfo* info) {
   uint16* port_field = GetPortFieldFromAddrinfo(info);
   if (!port_field)
     return -1;
   return ntohs(*port_field);
 }
 
 }  // namespace net