1 // Copyright 2011 The Chromium Authors
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/strings/string_piece.h"
10 #include "base/strings/sys_string_conversions.h"
11 #include "base/strings/utf_string_conversions.h"
12 #include "base/test/scoped_locale.h"
13 #include "build/build_config.h"
14 #include "testing/gtest/include/gtest/gtest.h"
15
16 #ifdef WCHAR_T_IS_32_BIT
17 static const std::wstring kSysWideOldItalicLetterA = L"\x10300";
18 #else
19 static const std::wstring kSysWideOldItalicLetterA = L"\xd800\xdf00";
20 #endif
21
22 namespace base {
23
TEST(SysStrings,SysWideToUTF8)24 TEST(SysStrings, SysWideToUTF8) {
25 EXPECT_EQ("Hello, world", SysWideToUTF8(L"Hello, world"));
26 EXPECT_EQ("\xe4\xbd\xa0\xe5\xa5\xbd", SysWideToUTF8(L"\x4f60\x597d"));
27
28 // A value outside of the BMP and therefore not representable with one UTF-16
29 // code unit.
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
TEST(SysStrings,SysUTF8ToWide)53 TEST(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
57 // A value outside of the BMP and therefore not representable with one UTF-16
58 // code unit.
59 EXPECT_EQ(kSysWideOldItalicLetterA, SysUTF8ToWide("\xF0\x90\x8C\x80"));
60
61 // Error case. When Windows finds an invalid UTF-8 character, it just skips
62 // it. This seems weird because it's inconsistent with the reverse conversion.
63 //
64 // This is what XP does, but Vista has different behavior, so we don't bother
65 // verifying it:
66 // EXPECT_EQ(L"\x4f60zyxw", SysUTF8ToWide("\xe4\xbd\xa0\xe5\xa5zyxw"));
67
68 // Test embedded NULLs.
69 std::string utf8_null("a");
70 utf8_null.push_back(0);
71 utf8_null.push_back('b');
72
73 std::wstring expected_null(L"a");
74 expected_null.push_back(0);
75 expected_null.push_back('b');
76
77 EXPECT_EQ(expected_null, SysUTF8ToWide(utf8_null));
78 }
79
80 // Tests depend on setting a specific Linux locale.
81 #if BUILDFLAG(IS_LINUX) || BUILDFLAG(IS_CHROMEOS)
TEST(SysStrings,SysWideToNativeMB)82 TEST(SysStrings, SysWideToNativeMB) {
83 #if !defined(SYSTEM_NATIVE_UTF8)
84 ScopedLocale locale("en_US.UTF-8");
85 #endif
86 EXPECT_EQ("Hello, world", SysWideToNativeMB(L"Hello, world"));
87 EXPECT_EQ("\xe4\xbd\xa0\xe5\xa5\xbd", SysWideToNativeMB(L"\x4f60\x597d"));
88
89 // A value outside of the BMP and therefore not representable with one UTF-16
90 // code unit.
91 EXPECT_EQ("\xF0\x90\x8C\x80", SysWideToNativeMB(kSysWideOldItalicLetterA));
92
93 // Error case. When Windows finds a UTF-16 character going off the end of
94 // a string, it just converts that literal value to UTF-8, even though this
95 // is invalid.
96 //
97 // This is what XP does, but Vista has different behavior, so we don't bother
98 // verifying it:
99 // EXPECT_EQ("\xE4\xBD\xA0\xED\xA0\x80zyxw",
100 // SysWideToNativeMB(L"\x4f60\xd800zyxw"));
101
102 // Test embedded NULLs.
103 std::wstring wide_null(L"a");
104 wide_null.push_back(0);
105 wide_null.push_back('b');
106
107 std::string expected_null("a");
108 expected_null.push_back(0);
109 expected_null.push_back('b');
110
111 EXPECT_EQ(expected_null, SysWideToNativeMB(wide_null));
112 }
113
114 // We assume the test is running in a UTF8 locale.
TEST(SysStrings,SysNativeMBToWide)115 TEST(SysStrings, SysNativeMBToWide) {
116 #if !defined(SYSTEM_NATIVE_UTF8)
117 ScopedLocale locale("en_US.UTF-8");
118 #endif
119 EXPECT_EQ(L"Hello, world", SysNativeMBToWide("Hello, world"));
120 EXPECT_EQ(L"\x4f60\x597d", SysNativeMBToWide("\xe4\xbd\xa0\xe5\xa5\xbd"));
121
122 // A value outside of the BMP and therefore not representable with one UTF-16
123 // code unit.
124 EXPECT_EQ(kSysWideOldItalicLetterA, SysNativeMBToWide("\xF0\x90\x8C\x80"));
125
126 // Error case. When Windows finds an invalid UTF-8 character, it just skips
127 // it. This seems weird because it's inconsistent with the reverse conversion.
128 //
129 // This is what XP does, but Vista has different behavior, so we don't bother
130 // verifying it:
131 // EXPECT_EQ(L"\x4f60zyxw", SysNativeMBToWide("\xe4\xbd\xa0\xe5\xa5zyxw"));
132
133 // Test embedded NULLs.
134 std::string utf8_null("a");
135 utf8_null.push_back(0);
136 utf8_null.push_back('b');
137
138 std::wstring expected_null(L"a");
139 expected_null.push_back(0);
140 expected_null.push_back('b');
141
142 EXPECT_EQ(expected_null, SysNativeMBToWide(utf8_null));
143 }
144
145 static const wchar_t* const kConvertRoundtripCases[] = {
146 L"Google Video",
147 // "网页 图片 资讯更多 »"
148 L"\x7f51\x9875\x0020\x56fe\x7247\x0020\x8d44\x8baf\x66f4\x591a\x0020\x00bb",
149 // "Παγκόσμιος Ιστός"
150 L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9"
151 L"\x03bf\x03c2\x0020\x0399\x03c3\x03c4\x03cc\x03c2",
152 // "Поиск страниц на русском"
153 L"\x041f\x043e\x0438\x0441\x043a\x0020\x0441\x0442"
154 L"\x0440\x0430\x043d\x0438\x0446\x0020\x043d\x0430"
155 L"\x0020\x0440\x0443\x0441\x0441\x043a\x043e\x043c",
156 // "전체서비스"
157 L"\xc804\xccb4\xc11c\xbe44\xc2a4",
158
159 // Test characters that take more than 16 bits. This will depend on whether
160 // wchar_t is 16 or 32 bits.
161 #if defined(WCHAR_T_IS_16_BIT)
162 L"\xd800\xdf00",
163 // ????? (Mathematical Alphanumeric Symbols (U+011d40 - U+011d44 :
164 // A,B,C,D,E)
165 L"\xd807\xdd40\xd807\xdd41\xd807\xdd42\xd807\xdd43\xd807\xdd44",
166 #elif defined(WCHAR_T_IS_32_BIT)
167 L"\x10300",
168 // ????? (Mathematical Alphanumeric Symbols (U+011d40 - U+011d44 :
169 // A,B,C,D,E)
170 L"\x11d40\x11d41\x11d42\x11d43\x11d44",
171 #endif
172 };
173
174
TEST(SysStrings,SysNativeMBAndWide)175 TEST(SysStrings, SysNativeMBAndWide) {
176 #if !defined(SYSTEM_NATIVE_UTF8)
177 ScopedLocale locale("en_US.UTF-8");
178 #endif
179 for (auto* i : kConvertRoundtripCases) {
180 std::wstring wide = i;
181 std::wstring trip = SysNativeMBToWide(SysWideToNativeMB(wide));
182 EXPECT_EQ(wide.size(), trip.size());
183 EXPECT_EQ(wide, trip);
184 }
185
186 // We assume our test is running in UTF-8, so double check through ICU.
187 for (auto* i : kConvertRoundtripCases) {
188 std::wstring wide = i;
189 std::wstring trip = SysNativeMBToWide(WideToUTF8(wide));
190 EXPECT_EQ(wide.size(), trip.size());
191 EXPECT_EQ(wide, trip);
192 }
193
194 for (auto* i : kConvertRoundtripCases) {
195 std::wstring wide = i;
196 std::wstring trip = UTF8ToWide(SysWideToNativeMB(wide));
197 EXPECT_EQ(wide.size(), trip.size());
198 EXPECT_EQ(wide, trip);
199 }
200 }
201 #endif // BUILDFLAG(IS_LINUX) || BUILDFLAG(IS_CHROMEOS)
202
203 } // namespace base
204