]> git.ipfire.org Git - thirdparty/systemd.git/blob - src/test/test-utf8.c
tree-wide: beautify remaining copyright statements
[thirdparty/systemd.git] / src / test / test-utf8.c
1 /* SPDX-License-Identifier: LGPL-2.1+ */
2 /***
3 Copyright © 2013 Dave Reisner
4 ***/
5
6 #include "alloc-util.h"
7 #include "string-util.h"
8 #include "utf8.h"
9 #include "util.h"
10
11 static void test_utf8_is_printable(void) {
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));
16 }
17
18 static void test_utf8_is_valid(void) {
19 assert_se(utf8_is_valid("ascii is valid unicode"));
20 assert_se(utf8_is_valid("\342\204\242"));
21 assert_se(!utf8_is_valid("\341\204"));
22 }
23
24 static void test_ascii_is_valid(void) {
25 assert_se( ascii_is_valid("alsdjf\t\vbarr\nba z"));
26 assert_se(!ascii_is_valid("\342\204\242"));
27 assert_se(!ascii_is_valid("\341\204"));
28 }
29
30 static void test_ascii_is_valid_n(void) {
31 assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 17));
32 assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 16));
33 assert_se(!ascii_is_valid_n("alsdjf\t\vbarr\nba z", 18));
34 assert_se(!ascii_is_valid_n("\342\204\242", 3));
35 assert_se(!ascii_is_valid_n("\342\204\242", 2));
36 assert_se(!ascii_is_valid_n("\342\204\242", 1));
37 assert_se( ascii_is_valid_n("\342\204\242", 0));
38 }
39
40 static void test_utf8_encoded_valid_unichar(void) {
41 assert_se(utf8_encoded_valid_unichar("\342\204\242") == 3);
42 assert_se(utf8_encoded_valid_unichar("\302\256") == 2);
43 assert_se(utf8_encoded_valid_unichar("a") == 1);
44 assert_se(utf8_encoded_valid_unichar("\341\204") < 0);
45 assert_se(utf8_encoded_valid_unichar("\341\204\341\204") < 0);
46 }
47
48 static void test_utf8_escaping(void) {
49 _cleanup_free_ char *p1, *p2, *p3;
50
51 p1 = utf8_escape_invalid("goo goo goo");
52 puts(p1);
53 assert_se(utf8_is_valid(p1));
54
55 p2 = utf8_escape_invalid("\341\204\341\204");
56 puts(p2);
57 assert_se(utf8_is_valid(p2));
58
59 p3 = utf8_escape_invalid("\341\204");
60 puts(p3);
61 assert_se(utf8_is_valid(p3));
62 }
63
64 static void test_utf8_escaping_printable(void) {
65 _cleanup_free_ char *p1, *p2, *p3, *p4, *p5, *p6;
66
67 p1 = utf8_escape_non_printable("goo goo goo");
68 puts(p1);
69 assert_se(utf8_is_valid(p1));
70
71 p2 = utf8_escape_non_printable("\341\204\341\204");
72 puts(p2);
73 assert_se(utf8_is_valid(p2));
74
75 p3 = utf8_escape_non_printable("\341\204");
76 puts(p3);
77 assert_se(utf8_is_valid(p3));
78
79 p4 = utf8_escape_non_printable("ąę\n가너도루\n1234\n\341\204\341\204\n\001 \019\20\a");
80 puts(p4);
81 assert_se(utf8_is_valid(p4));
82
83 p5 = utf8_escape_non_printable("\001 \019\20\a");
84 puts(p5);
85 assert_se(utf8_is_valid(p5));
86
87 p6 = utf8_escape_non_printable("\xef\xbf\x30\x13");
88 puts(p6);
89 assert_se(utf8_is_valid(p6));
90 }
91
92 static void test_utf16_to_utf8(void) {
93 char *a = NULL;
94 const uint16_t utf16[] = { htole16('a'), htole16(0xd800), htole16('b'), htole16(0xdc00), htole16('c'), htole16(0xd801), htole16(0xdc37) };
95 const char utf8[] = { 'a', 'b', 'c', 0xf0, 0x90, 0x90, 0xb7, 0 };
96
97 a = utf16_to_utf8(utf16, 14);
98 assert_se(a);
99 assert_se(streq(a, utf8));
100
101 free(a);
102 }
103
104 static void test_utf8_n_codepoints(void) {
105 assert_se(utf8_n_codepoints("abc") == 3);
106 assert_se(utf8_n_codepoints("zażółcić gęślą jaźń") == 19);
107 assert_se(utf8_n_codepoints("串") == 1);
108 assert_se(utf8_n_codepoints("") == 0);
109 assert_se(utf8_n_codepoints("…👊🔪💐…") == 5);
110 assert_se(utf8_n_codepoints("\xF1") == (size_t) -1);
111 }
112
113 static void test_utf8_console_width(void) {
114 assert_se(utf8_console_width("abc") == 3);
115 assert_se(utf8_console_width("zażółcić gęślą jaźń") == 19);
116 assert_se(utf8_console_width("串") == 2);
117 assert_se(utf8_console_width("") == 0);
118 assert_se(utf8_console_width("…👊🔪💐…") == 8);
119 assert_se(utf8_console_width("\xF1") == (size_t) -1);
120 }
121
122 int main(int argc, char *argv[]) {
123 test_utf8_is_valid();
124 test_utf8_is_printable();
125 test_ascii_is_valid();
126 test_ascii_is_valid_n();
127 test_utf8_encoded_valid_unichar();
128 test_utf8_escaping();
129 test_utf8_escaping_printable();
130 test_utf16_to_utf8();
131 test_utf8_n_codepoints();
132 test_utf8_console_width();
133
134 return 0;
135 }