1// Copyright (c) 2011 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 <stddef.h>
6
7#include <string>
8
9#include "base/macros.h"
10#include "base/strings/string_piece.h"
11#include "base/strings/sys_string_conversions.h"
12#include "base/strings/utf_string_conversions.h"
13#include "base/test/scoped_locale.h"
14#include "build/build_config.h"
15#include "testing/gtest/include/gtest/gtest.h"
16
17#ifdef WCHAR_T_IS_UTF32
18static const std::wstring kSysWideOldItalicLetterA = L"\x10300";
19#else
20static const std::wstring kSysWideOldItalicLetterA = L"\xd800\xdf00";
21#endif
22
23namespace base {
24
25TEST(SysStrings, SysWideToUTF8) {
26  EXPECT_EQ("Hello, world", SysWideToUTF8(L"Hello, world"));
27  EXPECT_EQ("\xe4\xbd\xa0\xe5\xa5\xbd", SysWideToUTF8(L"\x4f60\x597d"));
28
29  // >16 bits
30  EXPECT_EQ("\xF0\x90\x8C\x80", SysWideToUTF8(kSysWideOldItalicLetterA));
31
32  // Error case. When Windows finds a UTF-16 character going off the end of
33  // a string, it just converts that literal value to UTF-8, even though this
34  // is invalid.
35  //
36  // This is what XP does, but Vista has different behavior, so we don't bother
37  // verifying it:
38  // EXPECT_EQ("\xE4\xBD\xA0\xED\xA0\x80zyxw",
39  //           SysWideToUTF8(L"\x4f60\xd800zyxw"));
40
41  // Test embedded NULLs.
42  std::wstring wide_null(L"a");
43  wide_null.push_back(0);
44  wide_null.push_back('b');
45
46  std::string expected_null("a");
47  expected_null.push_back(0);
48  expected_null.push_back('b');
49
50  EXPECT_EQ(expected_null, SysWideToUTF8(wide_null));
51}
52
53TEST(SysStrings, SysUTF8ToWide) {
54  EXPECT_EQ(L"Hello, world", SysUTF8ToWide("Hello, world"));
55  EXPECT_EQ(L"\x4f60\x597d", SysUTF8ToWide("\xe4\xbd\xa0\xe5\xa5\xbd"));
56  // >16 bits
57  EXPECT_EQ(kSysWideOldItalicLetterA, SysUTF8ToWide("\xF0\x90\x8C\x80"));
58
59  // Error case. When Windows finds an invalid UTF-8 character, it just skips
60  // it. This seems weird because it's inconsistent with the reverse conversion.
61  //
62  // This is what XP does, but Vista has different behavior, so we don't bother
63  // verifying it:
64  // EXPECT_EQ(L"\x4f60zyxw", SysUTF8ToWide("\xe4\xbd\xa0\xe5\xa5zyxw"));
65
66  // Test embedded NULLs.
67  std::string utf8_null("a");
68  utf8_null.push_back(0);
69  utf8_null.push_back('b');
70
71  std::wstring expected_null(L"a");
72  expected_null.push_back(0);
73  expected_null.push_back('b');
74
75  EXPECT_EQ(expected_null, SysUTF8ToWide(utf8_null));
76}
77
78#if defined(OS_LINUX)  // Tests depend on setting a specific Linux locale.
79
80TEST(SysStrings, SysWideToNativeMB) {
81#if !defined(SYSTEM_NATIVE_UTF8)
82  ScopedLocale locale("en_US.UTF-8");
83#endif
84  EXPECT_EQ("Hello, world", SysWideToNativeMB(L"Hello, world"));
85  EXPECT_EQ("\xe4\xbd\xa0\xe5\xa5\xbd", SysWideToNativeMB(L"\x4f60\x597d"));
86
87  // >16 bits
88  EXPECT_EQ("\xF0\x90\x8C\x80", SysWideToNativeMB(kSysWideOldItalicLetterA));
89
90  // Error case. When Windows finds a UTF-16 character going off the end of
91  // a string, it just converts that literal value to UTF-8, even though this
92  // is invalid.
93  //
94  // This is what XP does, but Vista has different behavior, so we don't bother
95  // verifying it:
96  // EXPECT_EQ("\xE4\xBD\xA0\xED\xA0\x80zyxw",
97  //           SysWideToNativeMB(L"\x4f60\xd800zyxw"));
98
99  // Test embedded NULLs.
100  std::wstring wide_null(L"a");
101  wide_null.push_back(0);
102  wide_null.push_back('b');
103
104  std::string expected_null("a");
105  expected_null.push_back(0);
106  expected_null.push_back('b');
107
108  EXPECT_EQ(expected_null, SysWideToNativeMB(wide_null));
109}
110
111// We assume the test is running in a UTF8 locale.
112TEST(SysStrings, SysNativeMBToWide) {
113#if !defined(SYSTEM_NATIVE_UTF8)
114  ScopedLocale locale("en_US.UTF-8");
115#endif
116  EXPECT_EQ(L"Hello, world", SysNativeMBToWide("Hello, world"));
117  EXPECT_EQ(L"\x4f60\x597d", SysNativeMBToWide("\xe4\xbd\xa0\xe5\xa5\xbd"));
118  // >16 bits
119  EXPECT_EQ(kSysWideOldItalicLetterA, SysNativeMBToWide("\xF0\x90\x8C\x80"));
120
121  // Error case. When Windows finds an invalid UTF-8 character, it just skips
122  // it. This seems weird because it's inconsistent with the reverse conversion.
123  //
124  // This is what XP does, but Vista has different behavior, so we don't bother
125  // verifying it:
126  // EXPECT_EQ(L"\x4f60zyxw", SysNativeMBToWide("\xe4\xbd\xa0\xe5\xa5zyxw"));
127
128  // Test embedded NULLs.
129  std::string utf8_null("a");
130  utf8_null.push_back(0);
131  utf8_null.push_back('b');
132
133  std::wstring expected_null(L"a");
134  expected_null.push_back(0);
135  expected_null.push_back('b');
136
137  EXPECT_EQ(expected_null, SysNativeMBToWide(utf8_null));
138}
139
140static const wchar_t* const kConvertRoundtripCases[] = {
141  L"Google Video",
142  // "网页 图片 资讯更多 »"
143  L"\x7f51\x9875\x0020\x56fe\x7247\x0020\x8d44\x8baf\x66f4\x591a\x0020\x00bb",
144  //  "Παγκόσμιος Ιστός"
145  L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9"
146  L"\x03bf\x03c2\x0020\x0399\x03c3\x03c4\x03cc\x03c2",
147  // "Поиск страниц на русском"
148  L"\x041f\x043e\x0438\x0441\x043a\x0020\x0441\x0442"
149  L"\x0440\x0430\x043d\x0438\x0446\x0020\x043d\x0430"
150  L"\x0020\x0440\x0443\x0441\x0441\x043a\x043e\x043c",
151  // "전체서비스"
152  L"\xc804\xccb4\xc11c\xbe44\xc2a4",
153
154  // Test characters that take more than 16 bits. This will depend on whether
155  // wchar_t is 16 or 32 bits.
156#if defined(WCHAR_T_IS_UTF16)
157  L"\xd800\xdf00",
158  // ?????  (Mathematical Alphanumeric Symbols (U+011d40 - U+011d44 : A,B,C,D,E)
159  L"\xd807\xdd40\xd807\xdd41\xd807\xdd42\xd807\xdd43\xd807\xdd44",
160#elif defined(WCHAR_T_IS_UTF32)
161  L"\x10300",
162  // ?????  (Mathematical Alphanumeric Symbols (U+011d40 - U+011d44 : A,B,C,D,E)
163  L"\x11d40\x11d41\x11d42\x11d43\x11d44",
164#endif
165};
166
167
168TEST(SysStrings, SysNativeMBAndWide) {
169#if !defined(SYSTEM_NATIVE_UTF8)
170  ScopedLocale locale("en_US.UTF-8");
171#endif
172  for (size_t i = 0; i < arraysize(kConvertRoundtripCases); ++i) {
173    std::wstring wide = kConvertRoundtripCases[i];
174    std::wstring trip = SysNativeMBToWide(SysWideToNativeMB(wide));
175    EXPECT_EQ(wide.size(), trip.size());
176    EXPECT_EQ(wide, trip);
177  }
178
179  // We assume our test is running in UTF-8, so double check through ICU.
180  for (size_t i = 0; i < arraysize(kConvertRoundtripCases); ++i) {
181    std::wstring wide = kConvertRoundtripCases[i];
182    std::wstring trip = SysNativeMBToWide(WideToUTF8(wide));
183    EXPECT_EQ(wide.size(), trip.size());
184    EXPECT_EQ(wide, trip);
185  }
186
187  for (size_t i = 0; i < arraysize(kConvertRoundtripCases); ++i) {
188    std::wstring wide = kConvertRoundtripCases[i];
189    std::wstring trip = UTF8ToWide(SysWideToNativeMB(wide));
190    EXPECT_EQ(wide.size(), trip.size());
191    EXPECT_EQ(wide, trip);
192  }
193}
194#endif  // OS_LINUX
195
196}  // namespace base
197