url_formatter: Remove U+2010 from the displayable character set.

components/url_formatter/url_formatter_unittest.cc

 // Copyright 2015 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 "components/url_formatter/url_formatter.h"
 
 #include <stddef.h>
 #include <string.h>
 
 #include <vector>
@@ skipping 272 matching lines @@
   {"google.xn--comabc-k8d", L"google.com\x0338" L"abc", false},
   // Hyphenation Point instead of Katakana Middle dot
   {"xn--svgy16dha.jp", L"\x30a1\x2027\x30a3.jp", false},
   // Gershayim with other Hebrew characters is allowed.
   {"xn--5db6bh9b.il", L"\x05e9\x05d1\x05f4\x05e6.il", true},
   // Hebrew Gershayim with Latin is disallowed.
   {"xn--ab-yod.com", L"a\x05f4" L"b.com", false},
   // Hebrew Gershayim with Arabic is disallowed.
   {"xn--5eb7h.eg", L"\x0628\x05f4.eg", false},
 
+  // Hyphens (http://unicode.org/cldr/utility/confusables.jsp?a=-)
+  // Hyphen-Minus (the only hyphen allowed)
+  // abc-def
+  {"abc-def.com", L"abc-def.com", true},
+  // Modifier Letter Minus Sign
+  {"xn--abcdef-5od.com", L"abc\x02d7" L"def.com", false},
+  // Hyphen
+  {"xn--abcdef-dg0c.com", L"abc\x2010" L"def.com", false},
+  // Non-Breaking Hyphen
+  // This is actually an invalid IDNA domain (U+2011 normalizes to U+2010), but
+  // it is included to ensure that we do not inadvertently allow this character
+  // to be displayed as Unicode.
+  {"xn--abcdef-kg0c.com", L"abc\x2011" L"def.com", false},
+  // Figure Dash
+  {"xn--abcdef-rg0c.com", L"abc\x2012" L"def.com", false},
+  // En Dash
+  {"xn--abcdef-yg0c.com", L"abc\x2013" L"def.com", false},
+  // Hyphen Bullet
+  {"xn--abcdef-kq0c.com", L"abc\x2043" L"def.com", false},
+  // Minus Sign
+  {"xn--abcdef-5d3c.com", L"abc\x2212" L"def.com", false},
+  // Heavy Minus Sign
+  {"xn--abcdef-kg1d.com", L"abc\x2796" L"def.com", false},
+  // Coptic Capital Letter Dialect-P Ni
+  {"xn--abcdef-yy8d.com", L"abc\x2cba" L"def.com", false},
+  // Small Em Dash
+  {"xn--abcdef-5g0c.com", L"abc\xfe58" L"def.com", false},
+
   // Custom dangerous patterns
   // Two Katakana-Hiragana combining mark in a row
   {"google.xn--com-oh4ba.evil.jp", L"google.com\x309a\x309a.evil.jp", false},
   // Katakana Letter No not enclosed by {Han,Hiragana,Katakana}.
   {"google.xn--comevil-v04f.jp", L"google.com\x30ce" L"evil.jp", false},
   // TODO(jshin): Review the danger of allowing the following two.
   // Hiragana 'No' by itself is allowed.
   {"xn--ldk.jp", L"\x30ce.jp", true},
   // Hebrew Gershayim used by itself is allowed.
   {"xn--5eb.il", L"\x05f4.il", true},
@@ skipping 590 matching lines @@
     0, kNpos, kNpos, kNpos, kNpos, kNpos, kNpos, 0, kNpos, kNpos, kNpos, kNpos,
     0, 1, 2, 3, 4, 5, 6, 7
   };
   CheckAdjustedOffsets("http://user@foo.com/", kFormatUrlOmitAll,
                        net::UnescapeRule::NORMAL, omit_all_offsets);
 }
 
 }  // namespace
 
 }  // namespace url_formatter