Drop Mongolian from the IDN script list and tighten up the policy on Armenian-Latin mixing

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 103 matching lines @@
    L"\x0e2a\x0e32\x0e22\x0e01\x0e32\x0e23\x0e1a\x0e34\x0e19.th", true},
   // Thai + Common
   {"xn---123-9goxcp8c9db2r.th",
     L"\x0e20\x0e32\x0e29\x0e32\x0e44\x0e17\x0e22-123.th", true},
   // Devangari (Hindi)
   {"www.xn--l1b6a9e1b7c.in", L"www.\x0905\x0915\x094b\x0932\x093e.in", true},
   // Devanagari + Common
   {"xn---123-kbjl2j0bl2k.in",
     L"\x0939\x093f\x0928\x094d\x0926\x0940-123.in", true},
 
-  // 5 Aspirational scripts
+  // 4 Aspirational scripts
   // Unifieid Canadian Syllabary
   {"xn--dfe0tte.ca", L"\x1456\x14c2\x14ef.ca", true},
   // Tifinagh
   {"xn--4ljxa2bb4a6bxb.ma",
     L"\x2d5c\x2d49\x2d3c\x2d49\x2d4f\x2d30\x2d56.ma", true},
   // Tifinagh with a disallowed character(U+2D6F)
   {"xn--hmjzaby5d5f.ma", L"\x2d5c\x2d49\x2d3c\x2d6f\x2d49\x2d4f.ma", false},
   // Yi
   {"xn--4o7a6e1x64c.cn", L"\xa188\xa320\xa071\xa0b7.cn", true},
-  // Mongolian - 'ordu' (place, camp)
-  {"xn--56ec8bp.cn", L"\x1823\x1837\x1833\x1824.cn", true},
-  // Mongolian with a disallowed character
-  {"xn--95e5de3ds.cn", L"\x1823\x1837\x1804\x1833\x1824.cn", false},
   // Miao/Pollad
   {"xn--2u0fpf0a.cn", L"\U00016f04\U00016f62\U00016f59.cn", true},
 
+  // Mongolian is disallowed because it's written vertically.
+  // Mongolian - 'ordu' (place, camp)
+  {"xn--56ec8bp.cn", L"\x1823\x1837\x1833\x1824.cn", false},
+  // Mongolian with a disallowed character
+  {"xn--95e5de3ds.cn", L"\x1823\x1837\x1804\x1833\x1824.cn", false},
+
   // Script mixing tests
   // The following script combinations are allowed.
   // MODERATELY_RESTRICTIVE with Latin limited to ASCII-Latin.
   // ASCII-Latin + Japn (Kana + Han)
   // ASCII-Latin + Kore (Hangul + Han)
   // ASCII-Latin + Han + Bopomofo
   // ASCII-Latin + any allowed script other than Cyrillic, Greek, Cherokee
   // and Unified Canadian Syllabary
   // "payp<alpha>l.com"
   {"www.xn--paypl-g9d.com", L"payp\x03b1l.com", false},
@@ skipping 51 matching lines @@
   // Canadian Syllabary + Latin
   {"xn--ab-lym.com", L"ab\x14BF.com", false},
   {"xn--ab1-p6q.com", L"ab1\x14BF.com", false},
   {"xn--1ab-m6qd.com", L"\x14BF" L"1ab.com", false},
   {"xn--ab-jymc.com", L"\x14BF" L"ab.com", false},
   // Tifinagh + Latin
   {"xn--liy-go4a.com", L"li\u24dfy.com", false},
   {"xn--rol-ho4a.com", L"rol\u24df.com", false},
   {"xn--ily-eo4a.com", L"\u24dfily.com", false},
   {"xn--1ly-eo4a.com", L"\u24df1ly.com", false},
+  // Miao + Latin
+  {"xn--liy-rc12a.com", L"li\U00016FD8y.com", false},
+  {"xn--rol-sc12a.com", L"roll\U00016FD8.com", false},
+  {"xn--ily-pc12a.com", L"\U00016FD8ily.com", false},
+  {"xn--1ly-pc12a.com", L"\U00016FD81ly.com", false},
 
   // Invisibility check
   // Thai tone mark malek(U+0E48) repeated
   {"xn--03c0b3ca.th", L"\x0e23\x0e35\x0e48\x0e48.th", false},
   // Accute accent repeated
   {"xn--a-xbba.com", L"a\x0301\x0301.com", false},
   // 'a' with acuted accent + another acute accent
   {"xn--1ca20i.com", L"\x00e1\x0301.com", false},
   // Combining mark at the beginning
   {"xn--abc-fdc.jp", L"\x0300" L"abc.jp", false},
 
   // Mixed script confusable
-  // google with Armenian Small Letter Oh(U+0585)
+  // Armenian օ, ո, հ, and զ mixed with Latin
   {"xn--gogle-lkg.com", L"g\x0585ogle.com", false},
   {"xn--range-kkg.com", L"\x0585range.com", false},
   {"xn--cucko-pkg.com", L"cucko\x0585.com", false},
-  // Latin 'o' in Armenian.
+  {"xn--an-bed.com", L"\x0578" L"an.com", false},
+  {"xn--hig-tee.com", L"hig\x0570.com", false},
+  {"xn---ray-fef.com", L"\x0566-ray.com", false},
+  // Latin 'o', 'h' and 'n' in Armenian
   {"xn--o-ybcg0cu0cq.com",
     L"o\x0585\x0580\x0574\x0578\x0582\x0566\x0568.com", false},
+  {"xn--h-qccm4a.com", L"\x0580\x0574\x0578h.com", false},
+  {"xn--n-rccm3a.com", L"\x0580n\x0574\x0578.com", false},
+  {"xn--n1-0ddq0b.com", L"\x0580n1\x0574\x0578.com", false},
   // Hiragana HE(U+3078) mixed with Katakana
   {"xn--49jxi3as0d0fpc.com",
     L"\x30e2\x30d2\x30fc\x30c8\x3078\x30d6\x30f3.com", false},
 
   // U+30FC should be preceded by a Hiragana/Katakana.
   // Katakana + U+30FC + Han
   {"xn--lck0ip02qw5ya.jp", L"\x30ab\x30fc\x91ce\x7403.jp", true},
   // Hiragana + U+30FC + Han
   {"xn--u8j5tr47nw5ya.jp", L"\x304b\x30fc\x91ce\x7403.jp", true},
   // U+30FC + Han
@@ skipping 808 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