rtl_unittest.cc revision ddb351dbec246cf1fab5ec20d2d5520909041de1 
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 "base/i18n/rtl.h"
6
7#include <algorithm>
8
9#include "base/file_path.h"
10#include "base/string_util.h"
11#include "base/utf_string_conversions.h"
12#include "base/sys_string_conversions.h"
13#include "testing/gtest/include/gtest/gtest.h"
14#include "testing/platform_test.h"
15
16namespace {
17base::i18n::TextDirection GetTextDirection(const char* locale_name) {
18  return base::i18n::GetTextDirectionForLocale(locale_name);
19}
20}
21
22class RTLTest : public PlatformTest {
23};
24
25TEST_F(RTLTest, GetFirstStrongCharacterDirection) {
26  // Test pure LTR string.
27  std::wstring string(L"foo bar");
28  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
29            base::i18n::GetFirstStrongCharacterDirection(string));
30
31  // Test bidi string in which the first character with strong directionality
32  // is a character with type L.
33  string.assign(L"foo \x05d0 bar");
34  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
35            base::i18n::GetFirstStrongCharacterDirection(string));
36
37  // Test bidi string in which the first character with strong directionality
38  // is a character with type R.
39  string.assign(L"\x05d0 foo bar");
40  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT,
41            base::i18n::GetFirstStrongCharacterDirection(string));
42
43  // Test bidi string which starts with a character with weak directionality
44  // and in which the first character with strong directionality is a character
45  // with type L.
46  string.assign(L"!foo \x05d0 bar");
47  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
48            base::i18n::GetFirstStrongCharacterDirection(string));
49
50  // Test bidi string which starts with a character with weak directionality
51  // and in which the first character with strong directionality is a character
52  // with type R.
53  string.assign(L",\x05d0 foo bar");
54  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT,
55            base::i18n::GetFirstStrongCharacterDirection(string));
56
57  // Test bidi string in which the first character with strong directionality
58  // is a character with type LRE.
59  string.assign(L"\x202a \x05d0 foo  bar");
60  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
61            base::i18n::GetFirstStrongCharacterDirection(string));
62
63  // Test bidi string in which the first character with strong directionality
64  // is a character with type LRO.
65  string.assign(L"\x202d \x05d0 foo  bar");
66  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
67            base::i18n::GetFirstStrongCharacterDirection(string));
68
69  // Test bidi string in which the first character with strong directionality
70  // is a character with type RLE.
71  string.assign(L"\x202b foo \x05d0 bar");
72  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT,
73            base::i18n::GetFirstStrongCharacterDirection(string));
74
75  // Test bidi string in which the first character with strong directionality
76  // is a character with type RLO.
77  string.assign(L"\x202e foo \x05d0 bar");
78  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT,
79            base::i18n::GetFirstStrongCharacterDirection(string));
80
81  // Test bidi string in which the first character with strong directionality
82  // is a character with type AL.
83  string.assign(L"\x0622 foo \x05d0 bar");
84  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT,
85            base::i18n::GetFirstStrongCharacterDirection(string));
86
87  // Test a string without strong directionality characters.
88  string.assign(L",!.{}");
89  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
90            base::i18n::GetFirstStrongCharacterDirection(string));
91
92  // Test empty string.
93  string.assign(L"");
94  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
95            base::i18n::GetFirstStrongCharacterDirection(string));
96
97  // Test characters in non-BMP (e.g. Phoenician letters. Please refer to
98  // http://demo.icu-project.org/icu-bin/ubrowse?scr=151&b=10910 for more
99  // information).
100#if defined(WCHAR_T_IS_UTF32)
101  string.assign(L" ! \x10910" L"abc 123");
102#elif defined(WCHAR_T_IS_UTF16)
103  string.assign(L" ! \xd802\xdd10" L"abc 123");
104#else
105#error wchar_t should be either UTF-16 or UTF-32
106#endif
107  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT,
108            base::i18n::GetFirstStrongCharacterDirection(string));
109
110#if defined(WCHAR_T_IS_UTF32)
111  string.assign(L" ! \x10401" L"abc 123");
112#elif defined(WCHAR_T_IS_UTF16)
113  string.assign(L" ! \xd801\xdc01" L"abc 123");
114#else
115#error wchar_t should be either UTF-16 or UTF-32
116#endif
117  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT,
118            base::i18n::GetFirstStrongCharacterDirection(string));
119}
120
121TEST_F(RTLTest, WrapPathWithLTRFormatting) {
122  const wchar_t* kTestData[] = {
123    // Test common path, such as "c:\foo\bar".
124    L"c:/foo/bar",
125    // Test path with file name, such as "c:\foo\bar\test.jpg".
126    L"c:/foo/bar/test.jpg",
127    // Test path ending with punctuation, such as "c:\(foo)\bar.".
128    L"c:/(foo)/bar.",
129    // Test path ending with separator, such as "c:\foo\bar\".
130    L"c:/foo/bar/",
131    // Test path with RTL character.
132    L"c:/\x05d0",
133    // Test path with 2 level RTL directory names.
134    L"c:/\x05d0/\x0622",
135    // Test path with mixed RTL/LTR directory names and ending with punctuation.
136    L"c:/\x05d0/\x0622/(foo)/b.a.r.",
137    // Test path without driver name, such as "/foo/bar/test/jpg".
138    L"/foo/bar/test.jpg",
139    // Test path start with current directory, such as "./foo".
140    L"./foo",
141    // Test path start with parent directory, such as "../foo/bar.jpg".
142    L"../foo/bar.jpg",
143    // Test absolute path, such as "//foo/bar.jpg".
144    L"//foo/bar.jpg",
145    // Test path with mixed RTL/LTR directory names.
146    L"c:/foo/\x05d0/\x0622/\x05d1.jpg",
147    // Test empty path.
148    L""
149  };
150  for (unsigned int i = 0; i < arraysize(kTestData); ++i) {
151    FilePath path;
152#if defined(OS_WIN)
153    std::wstring win_path(kTestData[i]);
154    std::replace(win_path.begin(), win_path.end(), '/', '\\');
155    path = FilePath(win_path);
156    std::wstring wrapped_expected =
157        std::wstring(L"\x202a") + win_path + L"\x202c";
158#else
159    path = FilePath(base::SysWideToNativeMB(kTestData[i]));
160    std::wstring wrapped_expected =
161        std::wstring(L"\x202a") + kTestData[i] + L"\x202c";
162#endif
163    string16 localized_file_path_string;
164    base::i18n::WrapPathWithLTRFormatting(path, &localized_file_path_string);
165
166    std::wstring wrapped_actual = UTF16ToWide(localized_file_path_string);
167    EXPECT_EQ(wrapped_expected, wrapped_actual);
168  }
169}
170
171typedef struct  {
172    std::wstring raw_filename;
173    std::wstring display_string;
174} StringAndLTRString;
175
176TEST_F(RTLTest, GetDisplayStringInLTRDirectionality) {
177  const StringAndLTRString test_data[] = {
178    { L"test", L"\x202atest\x202c" },
179    { L"test.html", L"\x202atest.html\x202c" },
180    { L"\x05d0\x05d1\x05d2", L"\x202a\x05d0\x05d1\x05d2\x202c" },
181    { L"\x05d0\x05d1\x05d2.txt", L"\x202a\x05d0\x05d1\x05d2.txt\x202c" },
182    { L"\x05d0"L"abc", L"\x202a\x05d0"L"abc\x202c" },
183    { L"\x05d0"L"abc.txt", L"\x202a\x05d0"L"abc.txt\x202c" },
184    { L"abc\x05d0\x05d1", L"\x202a"L"abc\x05d0\x05d1\x202c" },
185    { L"abc\x05d0\x05d1.jpg", L"\x202a"L"abc\x05d0\x05d1.jpg\x202c" },
186  };
187  for (unsigned int i = 0; i < arraysize(test_data); ++i) {
188    string16 input = WideToUTF16(test_data[i].raw_filename);
189    string16 expected = base::i18n::GetDisplayStringInLTRDirectionality(input);
190    if (base::i18n::IsRTL())
191      EXPECT_EQ(expected, WideToUTF16(test_data[i].display_string));
192    else
193      EXPECT_EQ(expected, input);
194  }
195}
196
197TEST_F(RTLTest, GetTextDirection) {
198  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("ar"));
199  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("ar_EG"));
200  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("he"));
201  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("he_IL"));
202  // iw is an obsolete code for Hebrew.
203  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("iw"));
204  // Although we're not yet localized to Farsi and Urdu, we
205  // do have the text layout direction information for them.
206  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("fa"));
207  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("ur"));
208#if 0
209  // Enable these when we include the minimal locale data for Azerbaijani
210  // written in Arabic and Dhivehi. At the moment, our copy of
211  // ICU data does not have entries for them.
212  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("az_Arab"));
213  // Dhivehi that uses Thaana script.
214  EXPECT_EQ(base::i18n::RIGHT_TO_LEFT, GetTextDirection("dv"));
215#endif
216  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT, GetTextDirection("en"));
217  // Chinese in China with '-'.
218  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT, GetTextDirection("zh-CN"));
219  // Filipino : 3-letter code
220  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT, GetTextDirection("fil"));
221  // Russian
222  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT, GetTextDirection("ru"));
223  // Japanese that uses multiple scripts
224  EXPECT_EQ(base::i18n::LEFT_TO_RIGHT, GetTextDirection("ja"));
225}
226
227