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 "base/i18n/case_conversion.h"
6 #include "base/i18n/rtl.h"
7 #include "base/strings/utf_string_conversions.h"
8 #include "base/test/icu_test_util.h"
9 #include "testing/gtest/include/gtest/gtest.h"
10 #include "third_party/icu/source/i18n/unicode/usearch.h"
11
12 namespace base {
13 namespace i18n {
14
15 namespace {
16
17 const char16_t kNonASCIIMixed[] =
18 u"\xC4\xD6\xE4\xF6\x20\xCF\xEF\x20\xF7\x25"
19 u"\xA4\x23\x2A\x5E\x60\x40\xA3\x24\x2030\x201A\x7E\x20\x1F07\x1F0F"
20 u"\x20\x1E00\x1E01";
21 const char16_t kNonASCIILower[] =
22 u"\xE4\xF6\xE4\xF6\x20\xEF\xEF"
23 u"\x20\xF7\x25\xA4\x23\x2A\x5E\x60\x40\xA3\x24\x2030\x201A\x7E\x20\x1F07"
24 u"\x1F07\x20\x1E01\x1E01";
25 const char16_t kNonASCIIUpper[] =
26 u"\xC4\xD6\xC4\xD6\x20\xCF\xCF"
27 u"\x20\xF7\x25\xA4\x23\x2A\x5E\x60\x40\xA3\x24\x2030\x201A\x7E\x20\x1F0F"
28 u"\x1F0F\x20\x1E00\x1E00";
29
30 } // namespace
31
32 // Test upper and lower case string conversion.
TEST(CaseConversionTest,UpperLower)33 TEST(CaseConversionTest, UpperLower) {
34 const std::u16string mixed(u"Text with UPPer & lowER casE.");
35 const std::u16string expected_lower(u"text with upper & lower case.");
36 const std::u16string expected_upper(u"TEXT WITH UPPER & LOWER CASE.");
37
38 std::u16string result = ToLower(mixed);
39 EXPECT_EQ(expected_lower, result);
40
41 result = ToUpper(mixed);
42 EXPECT_EQ(expected_upper, result);
43 }
44
TEST(CaseConversionTest,NonASCII)45 TEST(CaseConversionTest, NonASCII) {
46 const std::u16string mixed(kNonASCIIMixed);
47 const std::u16string expected_lower(kNonASCIILower);
48 const std::u16string expected_upper(kNonASCIIUpper);
49
50 std::u16string result = ToLower(mixed);
51 EXPECT_EQ(expected_lower, result);
52
53 result = ToUpper(mixed);
54 EXPECT_EQ(expected_upper, result);
55 }
56
TEST(CaseConversionTest,TurkishLocaleConversion)57 TEST(CaseConversionTest, TurkishLocaleConversion) {
58 const std::u16string mixed(u"\x49\x131");
59 const std::u16string expected_lower(u"\x69\x131");
60 const std::u16string expected_upper(u"\x49\x49");
61
62 test::ScopedRestoreICUDefaultLocale restore_locale;
63 i18n::SetICUDefaultLocale("en_US");
64
65 std::u16string result = ToLower(mixed);
66 EXPECT_EQ(expected_lower, result);
67
68 result = ToUpper(mixed);
69 EXPECT_EQ(expected_upper, result);
70
71 i18n::SetICUDefaultLocale("tr");
72
73 const std::u16string expected_lower_turkish(u"\x131\x131");
74 const std::u16string expected_upper_turkish(u"\x49\x49");
75
76 result = ToLower(mixed);
77 EXPECT_EQ(expected_lower_turkish, result);
78
79 result = ToUpper(mixed);
80 EXPECT_EQ(expected_upper_turkish, result);
81 }
82
TEST(CaseConversionTest,FoldCase)83 TEST(CaseConversionTest, FoldCase) {
84 // Simple ASCII, should lower-case.
85 EXPECT_EQ(u"hello, world", FoldCase(u"Hello, World"));
86
87 // Non-ASCII cases from above. They should all fold to the same result.
88 EXPECT_EQ(FoldCase(kNonASCIIMixed), FoldCase(kNonASCIILower));
89 EXPECT_EQ(FoldCase(kNonASCIIMixed), FoldCase(kNonASCIIUpper));
90
91 // Turkish cases from above. This is the lower-case expected result from the
92 // US locale. It should be the same even when the current locale is Turkish.
93 const std::u16string turkish(u"\x49\x131");
94 const std::u16string turkish_expected(u"\x69\x131");
95
96 test::ScopedRestoreICUDefaultLocale restore_locale;
97 i18n::SetICUDefaultLocale("en_US");
98 EXPECT_EQ(turkish_expected, FoldCase(turkish));
99
100 i18n::SetICUDefaultLocale("tr");
101 EXPECT_EQ(turkish_expected, FoldCase(turkish));
102
103 // Test a case that gets bigger when processed.
104 // U+130 = LATIN CAPITAL LETTER I WITH DOT ABOVE gets folded to a lower case
105 // "i" followed by U+307 COMBINING DOT ABOVE.
106 EXPECT_EQ(u"i\u0307j", FoldCase(u"\u0130j"));
107
108 // U+00DF (SHARP S) and U+1E9E (CAPIRAL SHARP S) are both folded to "ss".
109 EXPECT_EQ(u"ssss", FoldCase(u"\u00DF\u1E9E"));
110 }
111
112 } // namespace i18n
113 } // namespace base
114
115
116
117