src/test/test-utf8.c

   1 /* SPDX-License-Identifier: LGPL-2.1+ */
   2
   3 #include "alloc-util.h"
   4 #include "string-util.h"
   5 #include "strv.h"
   6 #include "utf8.h"
   7 #include "util.h"
   8
   9 static void test_utf8_is_printable(void) {
  10         assert_se(utf8_is_printable("ascii is valid\tunicode", 22));
  11         assert_se(utf8_is_printable("\342\204\242", 3));
  12         assert_se(!utf8_is_printable("\341\204", 2));
  13         assert_se(utf8_is_printable("ąę", 4));
  14 }
  15
  16 static void test_utf8_is_valid(void) {
  17         assert_se(utf8_is_valid("ascii is valid unicode"));
  18         assert_se(utf8_is_valid("\342\204\242"));
  19         assert_se(!utf8_is_valid("\341\204"));
  20 }
  21
  22 static void test_ascii_is_valid(void) {
  23         assert_se( ascii_is_valid("alsdjf\t\vbarr\nba z"));
  24         assert_se(!ascii_is_valid("\342\204\242"));
  25         assert_se(!ascii_is_valid("\341\204"));
  26 }
  27
  28 static void test_ascii_is_valid_n(void) {
  29         assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 17));
  30         assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 16));
  31         assert_se(!ascii_is_valid_n("alsdjf\t\vbarr\nba z", 18));
  32         assert_se(!ascii_is_valid_n("\342\204\242", 3));
  33         assert_se(!ascii_is_valid_n("\342\204\242", 2));
  34         assert_se(!ascii_is_valid_n("\342\204\242", 1));
  35         assert_se( ascii_is_valid_n("\342\204\242", 0));
  36 }
  37
  38 static void test_utf8_encoded_valid_unichar(void) {
  39         assert_se(utf8_encoded_valid_unichar("\342\204\242", 1) == -EINVAL); /* truncated */
  40         assert_se(utf8_encoded_valid_unichar("\342\204\242", 2) == -EINVAL); /* truncated */
  41         assert_se(utf8_encoded_valid_unichar("\342\204\242", 3) == 3);
  42         assert_se(utf8_encoded_valid_unichar("\342\204\242", 4) == 3);
  43         assert_se(utf8_encoded_valid_unichar("\302\256", 1) == -EINVAL); /* truncated */
  44         assert_se(utf8_encoded_valid_unichar("\302\256", 2) == 2);
  45         assert_se(utf8_encoded_valid_unichar("\302\256", 3) == 2);
  46         assert_se(utf8_encoded_valid_unichar("\302\256", (size_t) -1) == 2);
  47         assert_se(utf8_encoded_valid_unichar("a", 1) == 1);
  48         assert_se(utf8_encoded_valid_unichar("a", 2) == 1);
  49         assert_se(utf8_encoded_valid_unichar("\341\204", 1) == -EINVAL); /* truncated, potentially valid */
  50         assert_se(utf8_encoded_valid_unichar("\341\204", 2) == -EINVAL); /* truncated, potentially valid */
  51         assert_se(utf8_encoded_valid_unichar("\341\204", 3) == -EINVAL);
  52         assert_se(utf8_encoded_valid_unichar("\341\204\341\204", 4) == -EINVAL);
  53         assert_se(utf8_encoded_valid_unichar("\341\204\341\204", 5) == -EINVAL);
  54 }
  55
  56 static void test_utf8_escaping(void) {
  57         _cleanup_free_ char *p1, *p2, *p3;
  58
  59         p1 = utf8_escape_invalid("goo goo goo");
  60         puts(p1);
  61         assert_se(utf8_is_valid(p1));
  62
  63         p2 = utf8_escape_invalid("\341\204\341\204");
  64         puts(p2);
  65         assert_se(utf8_is_valid(p2));
  66
  67         p3 = utf8_escape_invalid("\341\204");
  68         puts(p3);
  69         assert_se(utf8_is_valid(p3));
  70 }
  71
  72 static void test_utf8_escaping_printable(void) {
  73         _cleanup_free_ char *p1, *p2, *p3, *p4, *p5, *p6;
  74
  75         p1 = utf8_escape_non_printable("goo goo goo");
  76         puts(p1);
  77         assert_se(utf8_is_valid(p1));
  78
  79         p2 = utf8_escape_non_printable("\341\204\341\204");
  80         puts(p2);
  81         assert_se(utf8_is_valid(p2));
  82
  83         p3 = utf8_escape_non_printable("\341\204");
  84         puts(p3);
  85         assert_se(utf8_is_valid(p3));
  86
  87         p4 = utf8_escape_non_printable("ąę\n가너도루\n1234\n\341\204\341\204\n\001 \019\20\a");
  88         puts(p4);
  89         assert_se(utf8_is_valid(p4));
  90
  91         p5 = utf8_escape_non_printable("\001 \019\20\a");
  92         puts(p5);
  93         assert_se(utf8_is_valid(p5));
  94
  95         p6 = utf8_escape_non_printable("\xef\xbf\x30\x13");
  96         puts(p6);
  97         assert_se(utf8_is_valid(p6));
  98 }
  99
 100 static void test_utf16_to_utf8(void) {
 101         const char16_t utf16[] = { htole16('a'), htole16(0xd800), htole16('b'), htole16(0xdc00), htole16('c'), htole16(0xd801), htole16(0xdc37) };
 102         static const char utf8[] = { 'a', 'b', 'c', 0xf0, 0x90, 0x90, 0xb7 };
 103         _cleanup_free_ char16_t *b = NULL;
 104         _cleanup_free_ char *a = NULL;
 105
 106         /* Convert UTF-16 to UTF-8, filtering embedded bad chars */
 107         a = utf16_to_utf8(utf16, sizeof(utf16));
 108         assert_se(a);
 109         assert_se(memcmp(a, utf8, sizeof(utf8)) == 0);
 110
 111         /* Convert UTF-8 to UTF-16, and back */
 112         b = utf8_to_utf16(utf8, sizeof(utf8));
 113         assert_se(b);
 114
 115         free(a);
 116         a = utf16_to_utf8(b, char16_strlen(b) * 2);
 117         assert_se(a);
 118         assert_se(strlen(a) == sizeof(utf8));
 119         assert_se(memcmp(a, utf8, sizeof(utf8)) == 0);
 120 }
 121
 122 static void test_utf8_n_codepoints(void) {
 123         assert_se(utf8_n_codepoints("abc") == 3);
 124         assert_se(utf8_n_codepoints("zażółcić gęślą jaźń") == 19);
 125         assert_se(utf8_n_codepoints("串") == 1);
 126         assert_se(utf8_n_codepoints("") == 0);
 127         assert_se(utf8_n_codepoints("…👊🔪💐…") == 5);
 128         assert_se(utf8_n_codepoints("\xF1") == (size_t) -1);
 129 }
 130
 131 static void test_utf8_console_width(void) {
 132         assert_se(utf8_console_width("abc") == 3);
 133         assert_se(utf8_console_width("zażółcić gęślą jaźń") == 19);
 134         assert_se(utf8_console_width("串") == 2);
 135         assert_se(utf8_console_width("") == 0);
 136         assert_se(utf8_console_width("…👊🔪💐…") == 8);
 137         assert_se(utf8_console_width("\xF1") == (size_t) -1);
 138 }
 139
 140 static void test_utf8_to_utf16(void) {
 141         const char *p;
 142
 143         FOREACH_STRING(p,
 144                        "abc",
 145                        "zażółcić gęślą jaźń",
 146                        "串",
 147                        "",
 148                        "…👊🔪💐…") {
 149
 150                 _cleanup_free_ char16_t *a = NULL;
 151                 _cleanup_free_ char *b = NULL;
 152
 153                 a = utf8_to_utf16(p, strlen(p));
 154                 assert_se(a);
 155
 156                 b = utf16_to_utf8(a, char16_strlen(a) * 2);
 157                 assert_se(b);
 158                 assert_se(streq(p, b));
 159         }
 160 }
 161
 162 int main(int argc, char *argv[]) {
 163         test_utf8_is_valid();
 164         test_utf8_is_printable();
 165         test_ascii_is_valid();
 166         test_ascii_is_valid_n();
 167         test_utf8_encoded_valid_unichar();
 168         test_utf8_escaping();
 169         test_utf8_escaping_printable();
 170         test_utf16_to_utf8();
 171         test_utf8_n_codepoints();
 172         test_utf8_console_width();
 173         test_utf8_to_utf16();
 174
 175         return 0;
 176 }