base/sys_string_conversions_unittest.cc

   1 // Copyright (c) 2006-2008 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include <locale.h>
   6
   7 #include <string>
   8
   9 #include "base/basictypes.h"
  10 #include "base/string_piece.h"
  11 #include "base/utf_string_conversions.h"
  12 #include "base/sys_string_conversions.h"
  13 #include "testing/gtest/include/gtest/gtest.h"
  14
  15 #ifdef WCHAR_T_IS_UTF32
  16 static const std::wstring kSysWideOldItalicLetterA = L"\x10300";
  17 #else
  18 static const std::wstring kSysWideOldItalicLetterA = L"\xd800\xdf00";
  19 #endif
  20
  21 TEST(SysStrings, SysWideToUTF8) {
  22   using base::SysWideToUTF8;
  23   EXPECT_EQ("Hello, world", SysWideToUTF8(L"Hello, world"));
  24   EXPECT_EQ("\xe4\xbd\xa0\xe5\xa5\xbd", SysWideToUTF8(L"\x4f60\x597d"));
  25
  26   // >16 bits
  27   EXPECT_EQ("\xF0\x90\x8C\x80", SysWideToUTF8(kSysWideOldItalicLetterA));
  28
  29   // Error case. When Windows finds a UTF-16 character going off the end of
  30   // a string, it just converts that literal value to UTF-8, even though this
  31   // is invalid.
  32   //
  33   // This is what XP does, but Vista has different behavior, so we don't bother
  34   // verifying it:
  35   // EXPECT_EQ("\xE4\xBD\xA0\xED\xA0\x80zyxw",
  36   //           SysWideToUTF8(L"\x4f60\xd800zyxw"));
  37
  38   // Test embedded NULLs.
  39   std::wstring wide_null(L"a");
  40   wide_null.push_back(0);
  41   wide_null.push_back('b');
  42
  43   std::string expected_null("a");
  44   expected_null.push_back(0);
  45   expected_null.push_back('b');
  46
  47   EXPECT_EQ(expected_null, SysWideToUTF8(wide_null));
  48 }
  49
  50 TEST(SysStrings, SysUTF8ToWide) {
  51   using base::SysUTF8ToWide;
  52   EXPECT_EQ(L"Hello, world", SysUTF8ToWide("Hello, world"));
  53   EXPECT_EQ(L"\x4f60\x597d", SysUTF8ToWide("\xe4\xbd\xa0\xe5\xa5\xbd"));
  54   // >16 bits
  55   EXPECT_EQ(kSysWideOldItalicLetterA, SysUTF8ToWide("\xF0\x90\x8C\x80"));
  56
  57   // Error case. When Windows finds an invalid UTF-8 character, it just skips
  58   // it. This seems weird because it's inconsistent with the reverse conversion.
  59   //
  60   // This is what XP does, but Vista has different behavior, so we don't bother
  61   // verifying it:
  62   // EXPECT_EQ(L"\x4f60zyxw", SysUTF8ToWide("\xe4\xbd\xa0\xe5\xa5zyxw"));
  63
  64   // Test embedded NULLs.
  65   std::string utf8_null("a");
  66   utf8_null.push_back(0);
  67   utf8_null.push_back('b');
  68
  69   std::wstring expected_null(L"a");
  70   expected_null.push_back(0);
  71   expected_null.push_back('b');
  72
  73   EXPECT_EQ(expected_null, SysUTF8ToWide(utf8_null));
  74 }
  75
  76 #if defined(OS_LINUX)  // Tests depend on setting a specific Linux locale.
  77 namespace {
  78
  79 class ScopedSetLocale {
  80  public:
  81   explicit ScopedSetLocale(const char* locale) {
  82     old_locale_ = setlocale(LC_ALL, NULL);
  83     setlocale(LC_ALL, locale);
  84   }
  85   ~ScopedSetLocale() {
  86     setlocale(LC_ALL, old_locale_.c_str());
  87   }
  88
  89  private:
  90   std::string old_locale_;
  91 };
  92
  93 }  // namespace
  94
  95 TEST(SysStrings, SysWideToNativeMB) {
  96   using base::SysWideToNativeMB;
  97   ScopedSetLocale locale("en_US.utf-8");
  98   EXPECT_EQ("Hello, world", SysWideToNativeMB(L"Hello, world"));
  99   EXPECT_EQ("\xe4\xbd\xa0\xe5\xa5\xbd", SysWideToNativeMB(L"\x4f60\x597d"));
 100
 101   // >16 bits
 102   EXPECT_EQ("\xF0\x90\x8C\x80", SysWideToNativeMB(kSysWideOldItalicLetterA));
 103
 104   // Error case. When Windows finds a UTF-16 character going off the end of
 105   // a string, it just converts that literal value to UTF-8, even though this
 106   // is invalid.
 107   //
 108   // This is what XP does, but Vista has different behavior, so we don't bother
 109   // verifying it:
 110   // EXPECT_EQ("\xE4\xBD\xA0\xED\xA0\x80zyxw",
 111   //           SysWideToNativeMB(L"\x4f60\xd800zyxw"));
 112
 113   // Test embedded NULLs.
 114   std::wstring wide_null(L"a");
 115   wide_null.push_back(0);
 116   wide_null.push_back('b');
 117
 118   std::string expected_null("a");
 119   expected_null.push_back(0);
 120   expected_null.push_back('b');
 121
 122   EXPECT_EQ(expected_null, SysWideToNativeMB(wide_null));
 123 }
 124
 125 // We assume the test is running in a UTF8 locale.
 126 TEST(SysStrings, SysNativeMBToWide) {
 127   using base::SysNativeMBToWide;
 128   ScopedSetLocale locale("en_US.utf-8");
 129   EXPECT_EQ(L"Hello, world", SysNativeMBToWide("Hello, world"));
 130   EXPECT_EQ(L"\x4f60\x597d", SysNativeMBToWide("\xe4\xbd\xa0\xe5\xa5\xbd"));
 131   // >16 bits
 132   EXPECT_EQ(kSysWideOldItalicLetterA, SysNativeMBToWide("\xF0\x90\x8C\x80"));
 133
 134   // Error case. When Windows finds an invalid UTF-8 character, it just skips
 135   // it. This seems weird because it's inconsistent with the reverse conversion.
 136   //
 137   // This is what XP does, but Vista has different behavior, so we don't bother
 138   // verifying it:
 139   // EXPECT_EQ(L"\x4f60zyxw", SysNativeMBToWide("\xe4\xbd\xa0\xe5\xa5zyxw"));
 140
 141   // Test embedded NULLs.
 142   std::string utf8_null("a");
 143   utf8_null.push_back(0);
 144   utf8_null.push_back('b');
 145
 146   std::wstring expected_null(L"a");
 147   expected_null.push_back(0);
 148   expected_null.push_back('b');
 149
 150   EXPECT_EQ(expected_null, SysNativeMBToWide(utf8_null));
 151 }
 152
 153 static const wchar_t* const kConvertRoundtripCases[] = {
 154   L"Google Video",
 155   // "网页 图片 资讯更多 »"
 156   L"\x7f51\x9875\x0020\x56fe\x7247\x0020\x8d44\x8baf\x66f4\x591a\x0020\x00bb",
 157   //  "Παγκόσμιος Ιστός"
 158   L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9"
 159   L"\x03bf\x03c2\x0020\x0399\x03c3\x03c4\x03cc\x03c2",
 160   // "Поиск страниц на русском"
 161   L"\x041f\x043e\x0438\x0441\x043a\x0020\x0441\x0442"
 162   L"\x0440\x0430\x043d\x0438\x0446\x0020\x043d\x0430"
 163   L"\x0020\x0440\x0443\x0441\x0441\x043a\x043e\x043c",
 164   // "전체서비스"
 165   L"\xc804\xccb4\xc11c\xbe44\xc2a4",
 166
 167   // Test characters that take more than 16 bits. This will depend on whether
 168   // wchar_t is 16 or 32 bits.
 169 #if defined(WCHAR_T_IS_UTF16)
 170   L"\xd800\xdf00",
 171   // ?????  (Mathematical Alphanumeric Symbols (U+011d40 - U+011d44 : A,B,C,D,E)
 172   L"\xd807\xdd40\xd807\xdd41\xd807\xdd42\xd807\xdd43\xd807\xdd44",
 173 #elif defined(WCHAR_T_IS_UTF32)
 174   L"\x10300",
 175   // ?????  (Mathematical Alphanumeric Symbols (U+011d40 - U+011d44 : A,B,C,D,E)
 176   L"\x11d40\x11d41\x11d42\x11d43\x11d44",
 177 #endif
 178 };
 179
 180
 181 TEST(SysStrings, SysNativeMBAndWide) {
 182   ScopedSetLocale locale("en_US.utf-8");
 183   for (size_t i = 0; i < arraysize(kConvertRoundtripCases); ++i) {
 184     std::wstring wide = kConvertRoundtripCases[i];
 185     std::wstring trip = base::SysNativeMBToWide(base::SysWideToNativeMB(wide));
 186     EXPECT_EQ(wide.size(), trip.size());
 187     EXPECT_EQ(wide, trip);
 188   }
 189
 190   // We assume our test is running in UTF-8, so double check through ICU.
 191   for (size_t i = 0; i < arraysize(kConvertRoundtripCases); ++i) {
 192     std::wstring wide = kConvertRoundtripCases[i];
 193     std::wstring trip = base::SysNativeMBToWide(WideToUTF8(wide));
 194     EXPECT_EQ(wide.size(), trip.size());
 195     EXPECT_EQ(wide, trip);
 196   }
 197
 198   for (size_t i = 0; i < arraysize(kConvertRoundtripCases); ++i) {
 199     std::wstring wide = kConvertRoundtripCases[i];
 200     std::wstring trip = UTF8ToWide(base::SysWideToNativeMB(wide));
 201     EXPECT_EQ(wide.size(), trip.size());
 202     EXPECT_EQ(wide, trip);
 203   }
 204 }
 205 #endif  // OS_LINUX