]>
git.ipfire.org Git - thirdparty/systemd.git/blob - src/test/test-utf8.c
1 /* SPDX-License-Identifier: LGPL-2.1+ */
3 #include "alloc-util.h"
4 #include "string-util.h"
9 static void test_utf8_is_printable(void) {
10 log_info("/* %s */", __func__
);
12 assert_se(utf8_is_printable("ascii is valid\tunicode", 22));
13 assert_se(utf8_is_printable("\342\204\242", 3));
14 assert_se(!utf8_is_printable("\341\204", 2));
15 assert_se(utf8_is_printable("ąę", 4));
18 static void test_utf8_is_valid(void) {
19 log_info("/* %s */", __func__
);
21 assert_se(utf8_is_valid("ascii is valid unicode"));
22 assert_se(utf8_is_valid("\342\204\242"));
23 assert_se(!utf8_is_valid("\341\204"));
26 static void test_ascii_is_valid(void) {
27 log_info("/* %s */", __func__
);
29 assert_se( ascii_is_valid("alsdjf\t\vbarr\nba z"));
30 assert_se(!ascii_is_valid("\342\204\242"));
31 assert_se(!ascii_is_valid("\341\204"));
34 static void test_ascii_is_valid_n(void) {
35 log_info("/* %s */", __func__
);
37 assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 17));
38 assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 16));
39 assert_se(!ascii_is_valid_n("alsdjf\t\vbarr\nba z", 18));
40 assert_se(!ascii_is_valid_n("\342\204\242", 3));
41 assert_se(!ascii_is_valid_n("\342\204\242", 2));
42 assert_se(!ascii_is_valid_n("\342\204\242", 1));
43 assert_se( ascii_is_valid_n("\342\204\242", 0));
46 static void test_utf8_encoded_valid_unichar(void) {
47 log_info("/* %s */", __func__
);
49 assert_se(utf8_encoded_valid_unichar("\342\204\242", 1) == -EINVAL
); /* truncated */
50 assert_se(utf8_encoded_valid_unichar("\342\204\242", 2) == -EINVAL
); /* truncated */
51 assert_se(utf8_encoded_valid_unichar("\342\204\242", 3) == 3);
52 assert_se(utf8_encoded_valid_unichar("\342\204\242", 4) == 3);
53 assert_se(utf8_encoded_valid_unichar("\302\256", 1) == -EINVAL
); /* truncated */
54 assert_se(utf8_encoded_valid_unichar("\302\256", 2) == 2);
55 assert_se(utf8_encoded_valid_unichar("\302\256", 3) == 2);
56 assert_se(utf8_encoded_valid_unichar("\302\256", (size_t) -1) == 2);
57 assert_se(utf8_encoded_valid_unichar("a", 1) == 1);
58 assert_se(utf8_encoded_valid_unichar("a", 2) == 1);
59 assert_se(utf8_encoded_valid_unichar("\341\204", 1) == -EINVAL
); /* truncated, potentially valid */
60 assert_se(utf8_encoded_valid_unichar("\341\204", 2) == -EINVAL
); /* truncated, potentially valid */
61 assert_se(utf8_encoded_valid_unichar("\341\204", 3) == -EINVAL
);
62 assert_se(utf8_encoded_valid_unichar("\341\204\341\204", 4) == -EINVAL
);
63 assert_se(utf8_encoded_valid_unichar("\341\204\341\204", 5) == -EINVAL
);
66 static void test_utf8_escape_invalid(void) {
67 _cleanup_free_
char *p1
, *p2
, *p3
;
69 log_info("/* %s */", __func__
);
71 p1
= utf8_escape_invalid("goo goo goo");
73 assert_se(utf8_is_valid(p1
));
75 p2
= utf8_escape_invalid("\341\204\341\204");
77 assert_se(utf8_is_valid(p2
));
79 p3
= utf8_escape_invalid("\341\204");
81 assert_se(utf8_is_valid(p3
));
84 static void test_utf8_escape_non_printable(void) {
85 _cleanup_free_
char *p1
, *p2
, *p3
, *p4
, *p5
, *p6
;
87 log_info("/* %s */", __func__
);
89 p1
= utf8_escape_non_printable("goo goo goo");
91 assert_se(utf8_is_valid(p1
));
93 p2
= utf8_escape_non_printable("\341\204\341\204");
95 assert_se(utf8_is_valid(p2
));
97 p3
= utf8_escape_non_printable("\341\204");
99 assert_se(utf8_is_valid(p3
));
101 p4
= utf8_escape_non_printable("ąę\n가너도루\n1234\n\341\204\341\204\n\001 \019\20\a");
103 assert_se(utf8_is_valid(p4
));
105 p5
= utf8_escape_non_printable("\001 \019\20\a");
107 assert_se(utf8_is_valid(p5
));
109 p6
= utf8_escape_non_printable("\xef\xbf\x30\x13");
111 assert_se(utf8_is_valid(p6
));
114 static void test_utf8_escape_non_printable_full(void) {
115 log_info("/* %s */", __func__
);
117 for (size_t i
= 0; i
< 20; i
++) {
118 _cleanup_free_
char *p
;
120 p
= utf8_escape_non_printable_full("goo goo goo", i
);
122 assert_se(utf8_is_valid(p
));
123 assert_se(utf8_console_width(p
) <= i
);
126 for (size_t i
= 0; i
< 20; i
++) {
127 _cleanup_free_
char *p
;
129 p
= utf8_escape_non_printable_full("\001 \019\20\a", i
);
131 assert_se(utf8_is_valid(p
));
132 assert_se(utf8_console_width(p
) <= i
);
135 for (size_t i
= 0; i
< 20; i
++) {
136 _cleanup_free_
char *p
;
138 p
= utf8_escape_non_printable_full("\xef\xbf\x30\x13", i
);
140 assert_se(utf8_is_valid(p
));
141 assert_se(utf8_console_width(p
) <= i
);
145 static void test_utf16_to_utf8(void) {
146 const char16_t utf16
[] = { htole16('a'), htole16(0xd800), htole16('b'), htole16(0xdc00), htole16('c'), htole16(0xd801), htole16(0xdc37) };
147 static const char utf8
[] = { 'a', 'b', 'c', 0xf0, 0x90, 0x90, 0xb7 };
148 _cleanup_free_ char16_t
*b
= NULL
;
149 _cleanup_free_
char *a
= NULL
;
151 log_info("/* %s */", __func__
);
153 /* Convert UTF-16 to UTF-8, filtering embedded bad chars */
154 a
= utf16_to_utf8(utf16
, sizeof(utf16
));
156 assert_se(memcmp(a
, utf8
, sizeof(utf8
)) == 0);
158 /* Convert UTF-8 to UTF-16, and back */
159 b
= utf8_to_utf16(utf8
, sizeof(utf8
));
163 a
= utf16_to_utf8(b
, char16_strlen(b
) * 2);
165 assert_se(strlen(a
) == sizeof(utf8
));
166 assert_se(memcmp(a
, utf8
, sizeof(utf8
)) == 0);
169 static void test_utf8_n_codepoints(void) {
170 log_info("/* %s */", __func__
);
172 assert_se(utf8_n_codepoints("abc") == 3);
173 assert_se(utf8_n_codepoints("zażółcić gęślą jaźń") == 19);
174 assert_se(utf8_n_codepoints("串") == 1);
175 assert_se(utf8_n_codepoints("") == 0);
176 assert_se(utf8_n_codepoints("…👊🔪💐…") == 5);
177 assert_se(utf8_n_codepoints("\xF1") == (size_t) -1);
180 static void test_utf8_console_width(void) {
181 log_info("/* %s */", __func__
);
183 assert_se(utf8_console_width("abc") == 3);
184 assert_se(utf8_console_width("zażółcić gęślą jaźń") == 19);
185 assert_se(utf8_console_width("串") == 2);
186 assert_se(utf8_console_width("") == 0);
187 assert_se(utf8_console_width("…👊🔪💐…") == 8);
188 assert_se(utf8_console_width("\xF1") == (size_t) -1);
191 static void test_utf8_to_utf16(void) {
194 log_info("/* %s */", __func__
);
198 "zażółcić gęślą jaźń",
203 _cleanup_free_ char16_t
*a
= NULL
;
204 _cleanup_free_
char *b
= NULL
;
206 a
= utf8_to_utf16(p
, strlen(p
));
209 b
= utf16_to_utf8(a
, char16_strlen(a
) * 2);
211 assert_se(streq(p
, b
));
215 int main(int argc
, char *argv
[]) {
216 test_utf8_is_valid();
217 test_utf8_is_printable();
218 test_ascii_is_valid();
219 test_ascii_is_valid_n();
220 test_utf8_encoded_valid_unichar();
221 test_utf8_escape_invalid();
222 test_utf8_escape_non_printable();
223 test_utf8_escape_non_printable_full();
224 test_utf16_to_utf8();
225 test_utf8_n_codepoints();
226 test_utf8_console_width();
227 test_utf8_to_utf16();