Tweaks handling of U+30F[C-E]

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 169 matching lines @@
   // Devanagari + Bengali
   {"xn--11b0x.in", L"\x0915\x0995.in", 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)
   {"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--o-ybcg0cu0cq.com",
     L"o\x0585\x0580\x0574\x0578\x0582\x0566\x0568.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
   {"xn--weka801xo02a.com", L"\x30fc\x52d5\x753b\x30fc.com", false},
   // Han + U+30FC + Han
   {"xn--wekz60nb2ay85atj0b.jp", L"\x65e5\x672c\x30fc\x91ce\x7403.jp", false},
+  // U+30FC at the beginning
+  {"xn--wek060nb2a.jp", L"\x30fc\x65e5\x672c", false},
   // Latin + U+30FC + Latin
   {"xn--abcdef-r64e.jp", L"abc\x30fc" L"def.jp", false},
+
+  // U+30FB (・) is not allowed next to Latin, but allowed otherwise.
+  // U+30FB + Han
+  {"xn--vekt920a.jp", L"\x30fb\x91ce.jp", true},
+  // Han + U+30FB + Han
+  {"xn--vek160nb2ay85atj0b.jp", L"\x65e5\x672c\x30fb\x91ce\x7403.jp", true},
   // Latin + U+30FB + Latin
   {"xn--abcdef-k64e.jp", L"abc\x30fb" L"def.jp", false},
   // U+30FB + Latin
   {"xn--abc-os4b.jp", L"\x30fb" L"abc.jp", false},
 
+  // U+30FD (ヽ) is allowed only after Katakana.
+  // Katakana + U+30FD
+  {"xn--lck2i.jp", L"\x30ab\x30fd.jp", true},
+  // Hiragana + U+30FD
+  {"xn--u8j7t.jp", L"\x304b\x30fd.jp", false},
+  // Han + U+30FD
+  {"xn--xek368f.jp", L"\x4e00\x30fd.jp", false},
+  {"xn--aa-mju.jp", L"a\x30fd.jp", false},
+  {"xn--a1-bo4a.jp", L"a1\x30fd.jp", false},
+
+  // U+30FE (ヾ) is allowed only after Katakana.
+  // Katakana + U+30FE
+  {"xn--lck4i.jp", L"\x30ab\x30fe.jp", true},
+  // Hiragana + U+30FE
+  {"xn--u8j9t.jp", L"\x304b\x30fe.jp", false},
+  // Han + U+30FE
+  {"xn--yek168f.jp", L"\x4e00\x30fe.jp", false},
+  {"xn--a-oju.jp", L"a\x30fe.jp", false},
+  {"xn--a1-eo4a.jp", L"a1\x30fe.jp", false},
+
   // Cyrillic labels made of Latin-look-alike Cyrillic letters.
   // ѕсоре.com with ѕсоре in Cyrillic
   {"xn--e1argc3h.com", L"\x0455\x0441\x043e\x0440\x0435.com", false},
   // ѕсоре123.com with ѕсоре in Cyrillic.
   {"xn--123-qdd8bmf3n.com",
       L"\x0455\x0441\x043e\x0440\x0435" L"123.com", false},
   // ѕсоре-рау.com with ѕсоре and рау in Cyrillic.
   {"xn----8sbn9akccw8m.com",
       L"\x0455\x0441\x043e\x0440\x0435-\x0440\x0430\x0443.com", false},
   // ѕсоре·рау.com with scope and pay in Cyrillic and U+00B7 between them.
@@ skipping 728 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