]> git.ipfire.org Git - thirdparty/systemd.git/blob - src/test/test-utf8.c
Merge pull request #9185 from marckleinebudde/can
[thirdparty/systemd.git] / src / test / test-utf8.c
1 /* SPDX-License-Identifier: LGPL-2.1+ */
2 /***
3 This file is part of systemd.
4
5 Copyright 2013 Dave Reisner
6 ***/
7
8 #include "alloc-util.h"
9 #include "string-util.h"
10 #include "utf8.h"
11 #include "util.h"
12
13 static void test_utf8_is_printable(void) {
14 assert_se(utf8_is_printable("ascii is valid\tunicode", 22));
15 assert_se(utf8_is_printable("\342\204\242", 3));
16 assert_se(!utf8_is_printable("\341\204", 2));
17 assert_se(utf8_is_printable("ąę", 4));
18 }
19
20 static void test_utf8_is_valid(void) {
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"));
24 }
25
26 static void test_ascii_is_valid(void) {
27 assert_se( ascii_is_valid("alsdjf\t\vbarr\nba z"));
28 assert_se(!ascii_is_valid("\342\204\242"));
29 assert_se(!ascii_is_valid("\341\204"));
30 }
31
32 static void test_ascii_is_valid_n(void) {
33 assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 17));
34 assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 16));
35 assert_se(!ascii_is_valid_n("alsdjf\t\vbarr\nba z", 18));
36 assert_se(!ascii_is_valid_n("\342\204\242", 3));
37 assert_se(!ascii_is_valid_n("\342\204\242", 2));
38 assert_se(!ascii_is_valid_n("\342\204\242", 1));
39 assert_se( ascii_is_valid_n("\342\204\242", 0));
40 }
41
42 static void test_utf8_encoded_valid_unichar(void) {
43 assert_se(utf8_encoded_valid_unichar("\342\204\242") == 3);
44 assert_se(utf8_encoded_valid_unichar("\302\256") == 2);
45 assert_se(utf8_encoded_valid_unichar("a") == 1);
46 assert_se(utf8_encoded_valid_unichar("\341\204") < 0);
47 assert_se(utf8_encoded_valid_unichar("\341\204\341\204") < 0);
48 }
49
50 static void test_utf8_escaping(void) {
51 _cleanup_free_ char *p1, *p2, *p3;
52
53 p1 = utf8_escape_invalid("goo goo goo");
54 puts(p1);
55 assert_se(utf8_is_valid(p1));
56
57 p2 = utf8_escape_invalid("\341\204\341\204");
58 puts(p2);
59 assert_se(utf8_is_valid(p2));
60
61 p3 = utf8_escape_invalid("\341\204");
62 puts(p3);
63 assert_se(utf8_is_valid(p3));
64 }
65
66 static void test_utf8_escaping_printable(void) {
67 _cleanup_free_ char *p1, *p2, *p3, *p4, *p5, *p6;
68
69 p1 = utf8_escape_non_printable("goo goo goo");
70 puts(p1);
71 assert_se(utf8_is_valid(p1));
72
73 p2 = utf8_escape_non_printable("\341\204\341\204");
74 puts(p2);
75 assert_se(utf8_is_valid(p2));
76
77 p3 = utf8_escape_non_printable("\341\204");
78 puts(p3);
79 assert_se(utf8_is_valid(p3));
80
81 p4 = utf8_escape_non_printable("ąę\n가너도루\n1234\n\341\204\341\204\n\001 \019\20\a");
82 puts(p4);
83 assert_se(utf8_is_valid(p4));
84
85 p5 = utf8_escape_non_printable("\001 \019\20\a");
86 puts(p5);
87 assert_se(utf8_is_valid(p5));
88
89 p6 = utf8_escape_non_printable("\xef\xbf\x30\x13");
90 puts(p6);
91 assert_se(utf8_is_valid(p6));
92 }
93
94 static void test_utf16_to_utf8(void) {
95 char *a = NULL;
96 const uint16_t utf16[] = { htole16('a'), htole16(0xd800), htole16('b'), htole16(0xdc00), htole16('c'), htole16(0xd801), htole16(0xdc37) };
97 const char utf8[] = { 'a', 'b', 'c', 0xf0, 0x90, 0x90, 0xb7, 0 };
98
99 a = utf16_to_utf8(utf16, 14);
100 assert_se(a);
101 assert_se(streq(a, utf8));
102
103 free(a);
104 }
105
106 static void test_utf8_n_codepoints(void) {
107 assert_se(utf8_n_codepoints("abc") == 3);
108 assert_se(utf8_n_codepoints("zażółcić gęślą jaźń") == 19);
109 assert_se(utf8_n_codepoints("串") == 1);
110 assert_se(utf8_n_codepoints("") == 0);
111 assert_se(utf8_n_codepoints("…👊🔪💐…") == 5);
112 assert_se(utf8_n_codepoints("\xF1") == (size_t) -1);
113 }
114
115 static void test_utf8_console_width(void) {
116 assert_se(utf8_console_width("abc") == 3);
117 assert_se(utf8_console_width("zażółcić gęślą jaźń") == 19);
118 assert_se(utf8_console_width("串") == 2);
119 assert_se(utf8_console_width("") == 0);
120 assert_se(utf8_console_width("…👊🔪💐…") == 8);
121 assert_se(utf8_console_width("\xF1") == (size_t) -1);
122 }
123
124 int main(int argc, char *argv[]) {
125 test_utf8_is_valid();
126 test_utf8_is_printable();
127 test_ascii_is_valid();
128 test_ascii_is_valid_n();
129 test_utf8_encoded_valid_unichar();
130 test_utf8_escaping();
131 test_utf8_escaping_printable();
132 test_utf16_to_utf8();
133 test_utf8_n_codepoints();
134 test_utf8_console_width();
135
136 return 0;
137 }