1 /*
2 *
3 * Copyright (c) 2004
4 * John Maddock
5 *
6 * Use, modification and distribution are subject to the
7 * Boost Software License, Version 1.0. (See accompanying file
8 * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
9 *
10 */
11
12 /*
13 * LOCATION: see http://www.boost.org for most recent version.
14 * FILE test_unicode.hpp
15 * VERSION see <boost/version.hpp>
16 * DESCRIPTION: Unicode specific tests (requires ICU).
17 */
18
19 #include <boost/regex/config.hpp>
20 #ifdef BOOST_HAS_ICU
21 #include "test.hpp"
22
23 #ifdef BOOST_MSVC
24 #pragma warning(disable:4127)
25 #endif
26
27 #ifndef BOOST_NO_STD_WSTRING
28
29 #define TEST_REGEX_SEARCH_U(s, f, t, m, a)\
30 do{\
31 const wchar_t e[] = { s };\
32 std::wstring se(e, (sizeof(e) / sizeof(wchar_t)) - 1);\
33 const wchar_t st[] = { t };\
34 std::wstring sst(st, (sizeof(st) / sizeof(wchar_t)) - 1);\
35 test_info<wchar_t>::set_info(__FILE__, __LINE__, se, f, sst, m, a);\
36 test_icu(wchar_t(0), test_regex_search_tag());\
37 }while(0)
38
39 #define TEST_REGEX_CLASS_U(classname, character)\
40 TEST_REGEX_SEARCH_U(\
41 L"[[:" BOOST_JOIN(L, BOOST_STRINGIZE(classname)) L":]]",\
42 perl, \
43 BOOST_JOIN(L, \
44 BOOST_STRINGIZE(\
45 BOOST_JOIN(\x, character))), \
46 match_default, \
47 make_array(0, 1, -2, -2))
48
49 #else
50
51 #define TEST_REGEX_SEARCH_U(s, f, t, m, a)
52 #define TEST_REGEX_CLASS_U(classname, character)
53
54 #endif
55
test_unicode()56 void test_unicode()
57 {
58 using namespace boost::regex_constants;
59
60 TEST_REGEX_CLASS_U(L*, 3108);
61 TEST_REGEX_CLASS_U(Letter, 3108);
62 TEST_REGEX_CLASS_U(Lu, 2145);
63 TEST_REGEX_CLASS_U(Uppercase Letter, 2145);
64 TEST_REGEX_CLASS_U(Ll, 2146);
65 TEST_REGEX_CLASS_U(Lowercase Letter, 2146);
66 TEST_REGEX_CLASS_U(Lt, 1FFC);
67 TEST_REGEX_CLASS_U(Titlecase Letter, 1FFC);
68 TEST_REGEX_CLASS_U(Lm, 1D61);
69 TEST_REGEX_CLASS_U(Modifier Letter, 1D61);
70 TEST_REGEX_CLASS_U(Lo, 1974);
71 TEST_REGEX_CLASS_U(Other Letter, 1974);
72 TEST_REGEX_CLASS_U(M*, 20EA);
73 TEST_REGEX_CLASS_U(Mark, 20EA);
74 TEST_REGEX_CLASS_U(Mn, 20EA);
75 TEST_REGEX_CLASS_U(Non-Spacing Mark, 20EA);
76 TEST_REGEX_CLASS_U(Mc, 1938);
77 TEST_REGEX_CLASS_U(Spacing Combining Mark, 1938);
78 TEST_REGEX_CLASS_U(Me, 0488);
79 TEST_REGEX_CLASS_U(Enclosing Mark, 0488);
80 TEST_REGEX_CLASS_U(N*, 0669);
81 TEST_REGEX_CLASS_U(Number, 0669);
82 TEST_REGEX_CLASS_U(Nd, 0669);
83 TEST_REGEX_CLASS_U(Decimal Digit Number, 0669);
84 TEST_REGEX_CLASS_U(Nl, 303A);
85 TEST_REGEX_CLASS_U(Letter Number, 303A);
86 TEST_REGEX_CLASS_U(No, 2793);
87 TEST_REGEX_CLASS_U(Other Number, 2793);
88
89 TEST_REGEX_CLASS_U(S*, 2144);
90 TEST_REGEX_CLASS_U(Symbol, 2144);
91 TEST_REGEX_CLASS_U(Sm, 2144);
92 TEST_REGEX_CLASS_U(Math Symbol, 2144);
93 TEST_REGEX_CLASS_U(Sc, 20B1);
94 TEST_REGEX_CLASS_U(Currency Symbol, 20B1);
95 TEST_REGEX_CLASS_U(Sk, 1FFE);
96 TEST_REGEX_CLASS_U(Modifier Symbol, 1FFE);
97 TEST_REGEX_CLASS_U(So, 19FF);
98 TEST_REGEX_CLASS_U(Other Symbol, 19FF);
99
100 TEST_REGEX_CLASS_U(P*, 005F);
101 TEST_REGEX_CLASS_U(Punctuation, 005F);
102 TEST_REGEX_CLASS_U(Pc, 005F);
103 TEST_REGEX_CLASS_U(Connector Punctuation, 005F);
104 TEST_REGEX_CLASS_U(Pd, 002D);
105 TEST_REGEX_CLASS_U(Dash Punctuation, 002D);
106 TEST_REGEX_CLASS_U(Ps, 0028);
107 TEST_REGEX_CLASS_U(Open Punctuation, 0028);
108 TEST_REGEX_CLASS_U(Pe, FF63);
109 TEST_REGEX_CLASS_U(Close Punctuation, FF63);
110 TEST_REGEX_CLASS_U(Pi, 2039);
111 TEST_REGEX_CLASS_U(Initial Punctuation, 2039);
112 TEST_REGEX_CLASS_U(Pf, 203A);
113 TEST_REGEX_CLASS_U(Final Punctuation, 203A);
114 TEST_REGEX_CLASS_U(Po, 2038);
115 TEST_REGEX_CLASS_U(Other Punctuation, 2038);
116
117 TEST_REGEX_CLASS_U(Z*, 202F);
118 TEST_REGEX_CLASS_U(Separator, 202F);
119 TEST_REGEX_CLASS_U(Zs, 202F);
120 TEST_REGEX_CLASS_U(Space Separator, 202F);
121 TEST_REGEX_CLASS_U(Zl, 2028);
122 TEST_REGEX_CLASS_U(Line Separator, 2028);
123 TEST_REGEX_CLASS_U(Zp, 2029);
124 TEST_REGEX_CLASS_U(Paragraph Separator, 2029);
125 #if !BOOST_WORKAROUND(BOOST_MSVC, < 1300)
126 // Some tests have to be disabled for VC6 because the compiler
127 // mangles the string literals...
128 TEST_REGEX_CLASS_U(C*, 009F);
129 TEST_REGEX_CLASS_U(Other, 009F);
130 TEST_REGEX_CLASS_U(Cc, 009F);
131 TEST_REGEX_CLASS_U(Control, 009F);
132 #endif
133 TEST_REGEX_CLASS_U(Cf, FFFB);
134 TEST_REGEX_CLASS_U(Format, FFFB);
135 //TEST_REGEX_CLASS_U(Cs, DC00);
136 //TEST_REGEX_CLASS_U(Surrogate, DC00);
137 TEST_REGEX_CLASS_U(Co, F8FF);
138 TEST_REGEX_CLASS_U(Private Use, F8FF);
139 TEST_REGEX_CLASS_U(Cn, FFFF);
140 TEST_REGEX_CLASS_U(Not Assigned, FFFF);
141 TEST_REGEX_CLASS_U(Any, 2038);
142 TEST_REGEX_CLASS_U(Assigned, 2038);
143 TEST_REGEX_CLASS_U(ASCII, 7f);
144 TEST_REGEX_SEARCH_U(L"[[:Assigned:]]", perl, L"\xffff", match_default, make_array(-2, -2));
145 TEST_REGEX_SEARCH_U(L"[[:ASCII:]]", perl, L"\x80", match_default, make_array(-2, -2));
146
147 TEST_REGEX_SEARCH_U(L"\\N{KHMER DIGIT SIX}", perl, L"\x17E6", match_default, make_array(0, 1, -2, -2));
148 TEST_REGEX_SEARCH_U(L"\\N{MODIFIER LETTER LOW ACUTE ACCENT}", perl, L"\x02CF", match_default, make_array(0, 1, -2, -2));
149 TEST_REGEX_SEARCH_U(L"\\N{SUPERSCRIPT ONE}", perl, L"\x00B9", match_default, make_array(0, 1, -2, -2));
150 TEST_REGEX_SEARCH_U(L"[\\N{KHMER DIGIT SIX}]", perl, L"\x17E6", match_default, make_array(0, 1, -2, -2));
151 TEST_REGEX_SEARCH_U(L"[\\N{MODIFIER LETTER LOW ACUTE ACCENT}]", perl, L"\x02CF", match_default, make_array(0, 1, -2, -2));
152 TEST_REGEX_SEARCH_U(L"[\\N{SUPERSCRIPT ONE}]", perl, L"\x00B9", match_default, make_array(0, 1, -2, -2));
153 TEST_REGEX_SEARCH_U(L"\\N{CJK UNIFIED IDEOGRAPH-7FED}", perl, L"\x7FED", match_default, make_array(0, 1, -2, -2));
154 #if !BOOST_WORKAROUND(BOOST_MSVC, < 1300)
155 // Some tests have to be disabled for VC6 because the compiler
156 // mangles the string literals...
157 TEST_REGEX_SEARCH_U(L"\\w+", perl, L" e\x301" L"coute ", match_default, make_array(1, 8, -2, -2));
158
159 TEST_REGEX_SEARCH_U(L"^", perl, L" \x2028 \x2029 \x000D\x000A \x000A \x000C \x000D \x0085 ",
160 match_default | match_not_bol, make_array(2, 2, -2, 4, 4, -2, 7, 7, -2, 9, 9, -2, 11, 11, -2, 13, 13, -2, 15, 15, -2, -2));
161 TEST_REGEX_SEARCH_U(L"$", perl, L" \x2028 \x2029 \x000D\x000A \x000A \x000C \x000D \x0085 ",
162 match_default | match_not_eol, make_array(1, 1, -2, 3, 3, -2, 5, 5, -2, 8, 8, -2, 10, 10, -2, 12, 12, -2, 14, 14, -2, -2));
163 TEST_REGEX_SEARCH_U(L".", perl, L" \x2028\x2029\x000D\x000A\x000A\x000C\x000D\x0085 ",
164 match_default | match_not_dot_newline, make_array(0, 1, -2, 9, 10, -2, -2));
165 #endif
166 }
167
168 #else
test_unicode()169 void test_unicode(){}
170 #endif
171