xref: /llvm-project/libcxx/test/std/utilities/format/format.functions/unicode.pass.cpp (revision 68c3d66a97a0c1b7bb1b6b2b84640eece07bcc9c) 
1 //===----------------------------------------------------------------------===//
2 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
3 // See https://llvm.org/LICENSE.txt for license information.
4 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
5 //
6 //===----------------------------------------------------------------------===//
7 
8 // UNSUPPORTED: c++03, c++11, c++14, c++17
9 // UNSUPPORTED: libcpp-has-no-incomplete-format
10 
11 // This version runs the test when the platform has Unicode support.
12 // UNSUPPORTED: libcpp-has-no-unicode
13 
14 // TODO FMT Investigate Windows and AIX issues.
15 // UNSUPPORTED msvc, target={{.+}}-windows-gnu
16 // UNSUPPORTED: LIBCXX-AIX-FIXME
17 
18 // TODO FMT This test should not require std::to_chars(floating-point)
19 // XFAIL: availability-fp_to_chars-missing
20 
21 // <format>
22 
23 // Tests the Unicode width support of the standard format specifiers.
24 // It tests [format.string.std]/8 - 11:
25 // - Properly determining the estimated with of a unicode string.
26 // - Properly truncating to the wanted maximum width.
27 
28 // More specific extended grapheme cluster boundary rules are tested in
29 // test/libcxx/utilities/format/format.string/format.string.std/extended_grapheme_cluster.pass.cpp
30 // this test is based on test data provided by the Unicode Consortium.
31 
32 #include <format>
33 #include <cassert>
34 #include <vector>
35 
36 #include "make_string.h"
37 #include "test_macros.h"
38 #include "string_literal.h"
39 #include "test_format_string.h"
40 
41 #ifndef TEST_HAS_NO_LOCALIZATION
42 #  include <iostream>
43 #  include <type_traits>
44 #endif
45 
46 #define SV(S) MAKE_STRING_VIEW(CharT, S)
47 
48 template < class CharT, class... Args>
49 void check(std::basic_string_view<CharT> expected, test_format_string<CharT, Args...> fmt, Args&&... args) {
50   std::basic_string<CharT> out = std::format(fmt, std::forward<Args>(args)...);
51 #ifndef TEST_HAS_NO_LOCALIZATION
52   if constexpr (std::same_as<CharT, char>)
53     if (out != expected)
54       std::cerr << "\nFormat string   " << fmt.get() << "\nExpected output " << expected << "\nActual output   " << out
55                 << '\n';
56 #endif
57   assert(out == expected);
58 };
59 
60 template <class CharT>
61 static void test_single_code_point_fill() {
62   //*** 1-byte code points ***
63   check(SV("* *"), SV("{:*^3}"), SV(" "));
64   check(SV("*~*"), SV("{:*^3}"), SV("~"));
65 
66   //*** 2-byte code points ***
67   check(SV("*\u00a1*"), SV("{:*^3}"), SV("\u00a1")); // INVERTED EXCLAMATION MARK
68   check(SV("*\u07ff*"), SV("{:*^3}"), SV("\u07ff")); // NKO TAMAN SIGN
69 
70   //*** 3-byte code points ***
71   check(SV("*\u0800*"), SV("{:*^3}"), SV("\u0800")); // SAMARITAN LETTER ALAF
72   check(SV("*\ufffd*"), SV("{:*^3}"), SV("\ufffd")); // REPLACEMENT CHARACTER
73 
74   // 2 column ranges
75   check(SV("*\u1100*"), SV("{:*^4}"), SV("\u1100")); // HANGUL CHOSEONG KIYEOK
76   check(SV("*\u115f*"), SV("{:*^4}"), SV("\u115f")); // HANGUL CHOSEONG FILLER
77 
78   check(SV("*\u2329*"), SV("{:*^4}"), SV("\u2329")); // LEFT-POINTING ANGLE BRACKET
79   check(SV("*\u232a*"), SV("{:*^4}"), SV("\u232a")); // RIGHT-POINTING ANGLE BRACKET
80 
81   check(SV("*\u2e80*"), SV("{:*^4}"), SV("\u2e80")); // CJK RADICAL REPEAT
82   check(SV("*\u303e*"), SV("{:*^4}"), SV("\u303e")); // IDEOGRAPHIC VARIATION INDICATOR
83 
84   check(SV("*\u3041*"), SV("{:*^4}"), SV("\u3041")); // U+3041 HIRAGANA LETTER SMALL A
85   check(SV("*\ua4d0*"), SV("{:*^3}"), SV("\ua4d0")); // U+A4D0 LISU LETTER BA
86 
87   check(SV("*\uac00*"), SV("{:*^4}"), SV("\uac00")); // <Hangul Syllable, First>
88   check(SV("*\ud7a3*"), SV("{:*^4}"), SV("\ud7a3")); // Hangul Syllable Hih
89 
90   check(SV("*\uf900*"), SV("{:*^4}"), SV("\uf900")); // CJK COMPATIBILITY IDEOGRAPH-F900
91   check(SV("*\ufaff*"), SV("{:*^4}"), SV("\ufaff")); // U+FB00 LATIN SMALL LIGATURE FF
92 
93   check(SV("*\ufe10*"), SV("{:*^4}"), SV("\ufe10")); // PRESENTATION FORM FOR VERTICAL COMMA
94   check(SV("*\ufe19*"), SV("{:*^4}"), SV("\ufe19")); // PRESENTATION FORM FOR VERTICAL HORIZONTAL ELLIPSIS
95 
96   check(SV("*\ufe30*"), SV("{:*^4}"), SV("\ufe30")); // PRESENTATION FORM FOR VERTICAL TWO DOT LEADER
97   check(SV("*\ufe70*"), SV("{:*^3}"), SV("\ufe70")); // U+FE70 ARABIC FATHATAN ISOLATED FORM
98 
99   check(SV("*\uff01*"), SV("{:*^4}"), SV("\uff01")); // U+FF01 FULLWIDTH EXCLAMATION MARK
100   check(SV("*\uff60*"), SV("{:*^4}"), SV("\uff60")); // FULLWIDTH RIGHT WHITE PARENTHESIS
101 
102   check(SV("*\uffe0*"), SV("{:*^4}"), SV("\uffe0")); // FULLWIDTH CENT SIGN
103   check(SV("*\uffe6*"), SV("{:*^4}"), SV("\uffe6")); // FULLWIDTH WON SIGN
104 
105   //*** 4-byte code points ***
106   check(SV("*\U00010000*"), SV("{:*^3}"), SV("\U00010000")); // LINEAR B SYLLABLE B008 A
107   check(SV("*\U0010FFFF*"), SV("{:*^3}"), SV("\U0010FFFF")); // Undefined Character
108 
109   // 2 column ranges
110   check(SV("*\U0001f300*"), SV("{:*^4}"), SV("\U0001f300")); // CYCLONE
111   check(SV("*\U0001f64f*"), SV("{:*^4}"), SV("\U0001f64f")); // PERSON WITH FOLDED HANDS
112   check(SV("*\U0001f900*"), SV("{:*^4}"), SV("\U0001f900")); // CIRCLED CROSS FORMEE WITH FOUR DOTS
113   check(SV("*\U0001f9ff*"), SV("{:*^4}"), SV("\U0001f9ff")); // NAZAR AMULET
114   check(SV("*\U00020000*"), SV("{:*^4}"), SV("\U00020000")); // <CJK Ideograph Extension B, First>
115   check(SV("*\U0002fffd*"), SV("{:*^4}"), SV("\U0002fffd")); // Undefined Character
116   check(SV("*\U00030000*"), SV("{:*^4}"), SV("\U00030000")); // <CJK Ideograph Extension G, First>
117   check(SV("*\U0003fffd*"), SV("{:*^4}"), SV("\U0003fffd")); // Undefined Character
118 }
119 
120 // One column output is unaffected.
121 // Two column output is removed, thus the result is only the fill character.
122 template <class CharT>
123 static void test_single_code_point_truncate() {
124   //*** 1-byte code points ***
125   check(SV("* *"), SV("{:*^3.1}"), SV(" "));
126   check(SV("*~*"), SV("{:*^3.1}"), SV("~"));
127 
128   //*** 2-byte code points ***
129   check(SV("*\u00a1*"), SV("{:*^3.1}"), SV("\u00a1")); // INVERTED EXCLAMATION MARK
130   check(SV("*\u07ff*"), SV("{:*^3.1}"), SV("\u07ff")); // NKO TAMAN SIGN
131 
132   //*** 3.1-byte code points ***
133   check(SV("*\u0800*"), SV("{:*^3.1}"), SV("\u0800")); // SAMARITAN LETTER ALAF
134   check(SV("*\ufffd*"), SV("{:*^3.1}"), SV("\ufffd")); // REPLACEMENT CHARACTER
135 
136   // 2 column ranges
137   check(SV("***"), SV("{:*^3.1}"), SV("\u1100")); // HANGUL CHOSEONG KIYEOK
138   check(SV("***"), SV("{:*^3.1}"), SV("\u115f")); // HANGUL CHOSEONG FILLER
139 
140   check(SV("***"), SV("{:*^3.1}"), SV("\u2329")); // LEFT-POINTING ANGLE BRACKET
141   check(SV("***"), SV("{:*^3.1}"), SV("\u232a")); // RIGHT-POINTING ANGLE BRACKET
142 
143   check(SV("***"), SV("{:*^3.1}"), SV("\u2e80")); // CJK RADICAL REPEAT
144   check(SV("***"), SV("{:*^3.1}"), SV("\u303e")); // IDEOGRAPHIC VARIATION INDICATOR
145 
146   check(SV("***"), SV("{:*^3.1}"), SV("\u3041")); // U+3041 HIRAGANA LETTER SMALL A
147   check(SV("*\ua4d0*"), SV("{:*^3.1}"), SV("\ua4d0")); // U+A4D0 LISU LETTER BA
148 
149   check(SV("***"), SV("{:*^3.1}"), SV("\uac00")); // <Hangul Syllable, First>
150   check(SV("***"), SV("{:*^3.1}"), SV("\ud7a3")); // Hangul Syllable Hih
151 
152   check(SV("***"), SV("{:*^3.1}"), SV("\uf900")); // CJK COMPATIBILITY IDEOGRAPH-F900
153   check(SV("***"), SV("{:*^3.1}"), SV("\ufaff")); // U+FB00 LATIN SMALL LIGATURE FF
154 
155   check(SV("***"), SV("{:*^3.1}"), SV("\ufe10")); // PRESENTATION FORM FOR VERTICAL COMMA
156   check(SV("***"), SV("{:*^3.1}"), SV("\ufe19")); // PRESENTATION FORM FOR VERTICAL HORIZONTAL ELLIPSIS
157 
158   check(SV("***"), SV("{:*^3.1}"), SV("\ufe30")); // PRESENTATION FORM FOR VERTICAL TWO DOT LEADER
159   check(SV("*\ufe70*"), SV("{:*^3.1}"), SV("\ufe70")); // U+FE70 ARABIC FATHATAN ISOLATED FORM
160 
161   check(SV("***"), SV("{:*^3.1}"), SV("\uff01"));      // U+FF01 FULLWIDTH EXCLAMATION MARK
162   check(SV("***"), SV("{:*^3.1}"), SV("\uff60")); // FULLWIDTH RIGHT WHITE PARENTHESIS
163 
164   check(SV("***"), SV("{:*^3.1}"), SV("\uffe0")); // FULLWIDTH CENT SIGN
165   check(SV("***"), SV("{:*^3.1}"), SV("\uffe6")); // FULLWIDTH WON SIGN
166 
167   //*** 3.1-byte code points ***
168   check(SV("*\U00010000*"), SV("{:*^3.1}"), SV("\U00010000")); // LINEAR B SYLLABLE B008 A
169   check(SV("*\U0010FFFF*"), SV("{:*^3.1}"), SV("\U0010FFFF")); // Undefined Character
170 
171   // 2 column ranges
172   check(SV("***"), SV("{:*^3.1}"), SV("\U0001f300")); // CYCLONE
173   check(SV("***"), SV("{:*^3.1}"), SV("\U0001f64f")); // PERSON WITH FOLDED HANDS
174   check(SV("***"), SV("{:*^3.1}"), SV("\U0001f900")); // CIRCLED CROSS FORMEE WITH FOUR DOTS
175   check(SV("***"), SV("{:*^3.1}"), SV("\U0001f9ff")); // NAZAR AMULET
176   check(SV("***"), SV("{:*^3.1}"), SV("\U00020000")); // <CJK Ideograph Extension B, First>
177   check(SV("***"), SV("{:*^3.1}"), SV("\U0002fffd")); // Undefined Character
178   check(SV("***"), SV("{:*^3.1}"), SV("\U00030000")); // <CJK Ideograph Extension G, First>
179   check(SV("***"), SV("{:*^3.1}"), SV("\U0003fffd")); // Undefined Character
180 }
181 
182 // The examples used in that paper.
183 template <class CharT>
184 static void test_P1868() {
185   // Fill
186   check(SV("*\u0041*"), SV("{:*^3}"), SV("\u0041")); // { LATIN CAPITAL LETTER A }
187   check(SV("*\u00c1*"), SV("{:*^3}"), SV("\u00c1")); // { LATIN CAPITAL LETTER A WITH ACUTE }
188   check(SV("*\u0041\u0301*"),
189         SV("{:*^3}"),
190         SV("\u0041\u0301"));                         // { LATIN CAPITAL LETTER A } { COMBINING ACUTE ACCENT }
191   check(SV("*\u0132*"), SV("{:*^3}"), SV("\u0132")); // { LATIN CAPITAL LIGATURE IJ }
192   check(SV("*\u0394*"), SV("{:*^3}"), SV("\u0394")); // { GREEK CAPITAL LETTER DELTA }
193 
194   check(SV("*\u0429*"), SV("{:*^3}"), SV("\u0429"));         // { CYRILLIC CAPITAL LETTER SHCHA }
195   check(SV("*\u05d0*"), SV("{:*^3}"), SV("\u05d0"));         // { HEBREW LETTER ALEF }
196   check(SV("*\u0634*"), SV("{:*^3}"), SV("\u0634"));         // { ARABIC LETTER SHEEN }
197   check(SV("*\u3009*"), SV("{:*^4}"), SV("\u3009"));         // { RIGHT-POINTING ANGLE BRACKET }
198   check(SV("*\u754c*"), SV("{:*^4}"), SV("\u754c"));         // { CJK Unified Ideograph-754C }
199   check(SV("*\U0001f921*"), SV("{:*^4}"), SV("\U0001f921")); // { UNICORN FACE }
200   check(SV("*\U0001f468\u200d\U0001F469\u200d\U0001F467\u200d\U0001F466*"),
201         SV("{:*^4}"),
202         SV("\U0001f468\u200d\U0001F469\u200d\U0001F467\u200d\U0001F466")); // { Family: Man, Woman, Girl, Boy }
203 
204   // Truncate to 1 column: 1 column grapheme clusters are kept together.
205   check(SV("*\u0041*"), SV("{:*^3.1}"), SV("\u0041")); // { LATIN CAPITAL LETTER A }
206   check(SV("*\u00c1*"), SV("{:*^3.1}"), SV("\u00c1")); // { LATIN CAPITAL LETTER A WITH ACUTE }
207   check(SV("*\u0041\u0301*"),
208         SV("{:*^3.1}"),
209         SV("\u0041\u0301"));                           // { LATIN CAPITAL LETTER A } { COMBINING ACUTE ACCENT }
210   check(SV("*\u0132*"), SV("{:*^3.1}"), SV("\u0132")); // { LATIN CAPITAL LIGATURE IJ }
211   check(SV("*\u0394*"), SV("{:*^3.1}"), SV("\u0394")); // { GREEK CAPITAL LETTER DELTA }
212 
213   check(SV("*\u0429*"), SV("{:*^3.1}"), SV("\u0429")); // { CYRILLIC CAPITAL LETTER SHCHA }
214   check(SV("*\u05d0*"), SV("{:*^3.1}"), SV("\u05d0")); // { HEBREW LETTER ALEF }
215   check(SV("*\u0634*"), SV("{:*^3.1}"), SV("\u0634")); // { ARABIC LETTER SHEEN }
216   check(SV("***"), SV("{:*^3.1}"), SV("\u3009"));      // { RIGHT-POINTING ANGLE BRACKET }
217   check(SV("***"), SV("{:*^3.1}"), SV("\u754c"));      // { CJK Unified Ideograph-754C }
218   check(SV("***"), SV("{:*^3.1}"), SV("\U0001f921"));  // { UNICORN FACE }
219   check(SV("***"),
220         SV("{:*^3.1}"),
221         SV("\U0001f468\u200d\U0001F469\u200d\U0001F467\u200d\U0001F466")); // { Family: Man, Woman, Girl, Boy }
222 
223   // Truncate to 2 column: 2 column grapheme clusters are kept together.
224   check(SV("*\u0041*"), SV("{:*^3.2}"), SV("\u0041")); // { LATIN CAPITAL LETTER A }
225   check(SV("*\u00c1*"), SV("{:*^3.2}"), SV("\u00c1")); // { LATIN CAPITAL LETTER A WITH ACUTE }
226   check(SV("*\u0041\u0301*"),
227         SV("{:*^3.2}"),
228         SV("\u0041\u0301"));                           // { LATIN CAPITAL LETTER A } { COMBINING ACUTE ACCENT }
229   check(SV("*\u0132*"), SV("{:*^3.2}"), SV("\u0132")); // { LATIN CAPITAL LIGATURE IJ }
230   check(SV("*\u0394*"), SV("{:*^3.2}"), SV("\u0394")); // { GREEK CAPITAL LETTER DELTA }
231 
232   check(SV("*\u0429*"), SV("{:*^3.2}"), SV("\u0429"));         // { CYRILLIC CAPITAL LETTER SHCHA }
233   check(SV("*\u05d0*"), SV("{:*^3.2}"), SV("\u05d0"));         // { HEBREW LETTER ALEF }
234   check(SV("*\u0634*"), SV("{:*^3.2}"), SV("\u0634"));         // { ARABIC LETTER SHEEN }
235   check(SV("*\u3009*"), SV("{:*^4.2}"), SV("\u3009"));         // { RIGHT-POINTING ANGLE BRACKET }
236   check(SV("*\u754c*"), SV("{:*^4.2}"), SV("\u754c"));         // { CJK Unified Ideograph-754C }
237   check(SV("*\U0001f921*"), SV("{:*^4.2}"), SV("\U0001f921")); // { UNICORN FACE }
238   check(SV("*\U0001f468\u200d\U0001F469\u200d\U0001F467\u200d\U0001F466*"),
239         SV("{:*^4.2}"),
240         SV("\U0001f468\u200d\U0001F469\u200d\U0001F467\u200d\U0001F466")); // { Family: Man, Woman, Girl, Boy }
241 }
242 
243 #ifdef _LIBCPP_VERSION
244 // Tests the libc++ specific behaviour for malformed UTF-sequences. The
245 // Standard doesn't specify how to handle this.
246 template <class CharT>
247 static void test_malformed_code_point() {
248   if constexpr (sizeof(CharT) == 1) {
249     // Malformed at end.
250     check(SV("*ZZZZ\x8f*"), SV("{:*^7}"), SV("ZZZZ\x8f"));
251     check(SV("*ZZZZ\xcf*"), SV("{:*^7}"), SV("ZZZZ\xcf"));
252     check(SV("*ZZZZ\xef*"), SV("{:*^7}"), SV("ZZZZ\xef"));
253     check(SV("*ZZZZ\xff*"), SV("{:*^7}"), SV("ZZZZ\xff"));
254 
255     // Malformed in middle, no continuation
256     check(SV("*ZZZZ\x8fZ*"), SV("{:*^8}"), SV("ZZZZ\x8fZ"));
257     check(SV("*ZZZZ\xcfZ*"), SV("{:*^8}"), SV("ZZZZ\xcfZ"));
258     check(SV("*ZZZZ\xefZ*"), SV("{:*^8}"), SV("ZZZZ\xefZ"));
259     check(SV("*ZZZZ\xffZ*"), SV("{:*^8}"), SV("ZZZZ\xffZ"));
260 
261     check(SV("*ZZZZ\x8fZZ*"), SV("{:*^9}"), SV("ZZZZ\x8fZZ"));
262     check(SV("*ZZZZ\xcfZZ*"), SV("{:*^9}"), SV("ZZZZ\xcfZZ"));
263     check(SV("*ZZZZ\xefZZ*"), SV("{:*^9}"), SV("ZZZZ\xefZZ"));
264     check(SV("*ZZZZ\xffZZ*"), SV("{:*^9}"), SV("ZZZZ\xffZZ"));
265 
266     check(SV("*ZZZZ\x8fZZZ*"), SV("{:*^10}"), SV("ZZZZ\x8fZZZ"));
267     check(SV("*ZZZZ\xcfZZZ*"), SV("{:*^10}"), SV("ZZZZ\xcfZZZ"));
268     check(SV("*ZZZZ\xefZZZ*"), SV("{:*^10}"), SV("ZZZZ\xefZZZ"));
269     check(SV("*ZZZZ\xffZZZ*"), SV("{:*^10}"), SV("ZZZZ\xffZZZ"));
270 
271     check(SV("*ZZZZ\x8fZZZZ*"), SV("{:*^11}"), SV("ZZZZ\x8fZZZZ"));
272     check(SV("*ZZZZ\xcfZZZZ*"), SV("{:*^11}"), SV("ZZZZ\xcfZZZZ"));
273     check(SV("*ZZZZ\xefZZZZ*"), SV("{:*^11}"), SV("ZZZZ\xefZZZZ"));
274     check(SV("*ZZZZ\xffZZZZ*"), SV("{:*^11}"), SV("ZZZZ\xffZZZZ"));
275 
276     // Premature end.
277     check(SV("*ZZZZ\xef\xf5*"), SV("{:*^8}"), SV("ZZZZ\xef\xf5"));
278     check(SV("*ZZZZ\xef\xf5ZZZZ*"), SV("{:*^12}"), SV("ZZZZ\xef\xf5ZZZZ"));
279     check(SV("*ZZZZ\xff\xf5\xf5*"), SV("{:*^9}"), SV("ZZZZ\xff\xf5\xf5"));
280     check(SV("*ZZZZ\xff\xf5\xf5ZZZZ*"), SV("{:*^13}"), SV("ZZZZ\xff\xf5\xf5ZZZZ"));
281 
282   } else if constexpr (sizeof(CharT) == 2) {
283     // TODO FMT Add these tests.
284   }
285   // UTF-32 doesn't combine characters, thus no corruption tests.
286 }
287 #endif
288 
289 template <class CharT>
290 static void test() {
291   test_single_code_point_fill<CharT>();
292   test_single_code_point_truncate<CharT>();
293   test_P1868<CharT>();
294 
295 #ifdef _LIBCPP_VERSION
296   test_malformed_code_point<CharT>();
297 #endif
298 }
299 
300 int main(int, char**) {
301   test<char>();
302 
303 #ifndef TEST_HAS_NO_WIDE_CHARACTERS
304   test<wchar_t>();
305 #endif
306 
307   return 0;
308 }
309