src/test/test-utf8.c

   1 /* SPDX-License-Identifier: LGPL-2.1-or-later */
   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         log_info("/* %s */", __func__);
  11
  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         assert_se(!utf8_is_printable("\r", 1));
  17         assert_se(utf8_is_printable("\n", 1));
  18         assert_se(utf8_is_printable("\t", 1));
  19 }
  20
  21 static void test_utf8_n_is_valid(void) {
  22         log_info("/* %s */", __func__);
  23
  24         assert_se( utf8_is_valid_n("ascii is valid unicode", 21));
  25         assert_se( utf8_is_valid_n("ascii is valid unicode", 22));
  26         assert_se(!utf8_is_valid_n("ascii is valid unicode", 23));
  27         assert_se( utf8_is_valid_n("\342\204\242", 0));
  28         assert_se(!utf8_is_valid_n("\342\204\242", 1));
  29         assert_se(!utf8_is_valid_n("\342\204\242", 2));
  30         assert_se( utf8_is_valid_n("\342\204\242", 3));
  31         assert_se(!utf8_is_valid_n("\342\204\242", 4));
  32         assert_se( utf8_is_valid_n("<ZZ>", 0));
  33         assert_se( utf8_is_valid_n("<ZZ>", 1));
  34         assert_se( utf8_is_valid_n("<ZZ>", 2));
  35         assert_se( utf8_is_valid_n("<ZZ>", 3));
  36         assert_se( utf8_is_valid_n("<ZZ>", 4));
  37         assert_se(!utf8_is_valid_n("<ZZ>", 5));
  38 }
  39
  40 static void test_utf8_is_valid(void) {
  41         log_info("/* %s */", __func__);
  42
  43         assert_se(utf8_is_valid("ascii is valid unicode"));
  44         assert_se(utf8_is_valid("\342\204\242"));
  45         assert_se(!utf8_is_valid("\341\204"));
  46 }
  47
  48 static void test_ascii_is_valid(void) {
  49         log_info("/* %s */", __func__);
  50
  51         assert_se( ascii_is_valid("alsdjf\t\vbarr\nba z"));
  52         assert_se(!ascii_is_valid("\342\204\242"));
  53         assert_se(!ascii_is_valid("\341\204"));
  54 }
  55
  56 static void test_ascii_is_valid_n(void) {
  57         log_info("/* %s */", __func__);
  58
  59         assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 17));
  60         assert_se( ascii_is_valid_n("alsdjf\t\vbarr\nba z", 16));
  61         assert_se(!ascii_is_valid_n("alsdjf\t\vbarr\nba z", 18));
  62         assert_se(!ascii_is_valid_n("\342\204\242", 3));
  63         assert_se(!ascii_is_valid_n("\342\204\242", 2));
  64         assert_se(!ascii_is_valid_n("\342\204\242", 1));
  65         assert_se( ascii_is_valid_n("\342\204\242", 0));
  66 }
  67
  68 static void test_utf8_encoded_valid_unichar(void) {
  69         log_info("/* %s */", __func__);
  70
  71         assert_se(utf8_encoded_valid_unichar("\342\204\242", 1) == -EINVAL); /* truncated */
  72         assert_se(utf8_encoded_valid_unichar("\342\204\242", 2) == -EINVAL); /* truncated */
  73         assert_se(utf8_encoded_valid_unichar("\342\204\242", 3) == 3);
  74         assert_se(utf8_encoded_valid_unichar("\342\204\242", 4) == 3);
  75         assert_se(utf8_encoded_valid_unichar("\302\256", 1) == -EINVAL); /* truncated */
  76         assert_se(utf8_encoded_valid_unichar("\302\256", 2) == 2);
  77         assert_se(utf8_encoded_valid_unichar("\302\256", 3) == 2);
  78         assert_se(utf8_encoded_valid_unichar("\302\256", (size_t) -1) == 2);
  79         assert_se(utf8_encoded_valid_unichar("a", 1) == 1);
  80         assert_se(utf8_encoded_valid_unichar("a", 2) == 1);
  81         assert_se(utf8_encoded_valid_unichar("\341\204", 1) == -EINVAL); /* truncated, potentially valid */
  82         assert_se(utf8_encoded_valid_unichar("\341\204", 2) == -EINVAL); /* truncated, potentially valid */
  83         assert_se(utf8_encoded_valid_unichar("\341\204", 3) == -EINVAL);
  84         assert_se(utf8_encoded_valid_unichar("\341\204\341\204", 4) == -EINVAL);
  85         assert_se(utf8_encoded_valid_unichar("\341\204\341\204", 5) == -EINVAL);
  86 }
  87
  88 static void test_utf8_escape_invalid(void) {
  89         _cleanup_free_ char *p1, *p2, *p3;
  90
  91         log_info("/* %s */", __func__);
  92
  93         p1 = utf8_escape_invalid("goo goo goo");
  94         puts(p1);
  95         assert_se(utf8_is_valid(p1));
  96
  97         p2 = utf8_escape_invalid("\341\204\341\204");
  98         puts(p2);
  99         assert_se(utf8_is_valid(p2));
 100
 101         p3 = utf8_escape_invalid("\341\204");
 102         puts(p3);
 103         assert_se(utf8_is_valid(p3));
 104 }
 105
 106 static void test_utf8_escape_non_printable(void) {
 107         _cleanup_free_ char *p1, *p2, *p3, *p4, *p5, *p6;
 108
 109         log_info("/* %s */", __func__);
 110
 111         p1 = utf8_escape_non_printable("goo goo goo");
 112         puts(p1);
 113         assert_se(utf8_is_valid(p1));
 114
 115         p2 = utf8_escape_non_printable("\341\204\341\204");
 116         puts(p2);
 117         assert_se(utf8_is_valid(p2));
 118
 119         p3 = utf8_escape_non_printable("\341\204");
 120         puts(p3);
 121         assert_se(utf8_is_valid(p3));
 122
 123         p4 = utf8_escape_non_printable("ąę\n가너도루\n1234\n\341\204\341\204\n\001 \019\20\a");
 124         puts(p4);
 125         assert_se(utf8_is_valid(p4));
 126
 127         p5 = utf8_escape_non_printable("\001 \019\20\a");
 128         puts(p5);
 129         assert_se(utf8_is_valid(p5));
 130
 131         p6 = utf8_escape_non_printable("\xef\xbf\x30\x13");
 132         puts(p6);
 133         assert_se(utf8_is_valid(p6));
 134 }
 135
 136 static void test_utf8_escape_non_printable_full(void) {
 137         log_info("/* %s */", __func__);
 138
 139         for (size_t i = 0; i < 20; i++) {
 140                 _cleanup_free_ char *p;
 141
 142                 p = utf8_escape_non_printable_full("goo goo goo", i);
 143                 puts(p);
 144                 assert_se(utf8_is_valid(p));
 145                 assert_se(utf8_console_width(p) <= i);
 146         }
 147
 148         for (size_t i = 0; i < 20; i++) {
 149                 _cleanup_free_ char *p;
 150
 151                 p = utf8_escape_non_printable_full("\001 \019\20\a", i);
 152                 puts(p);
 153                 assert_se(utf8_is_valid(p));
 154                 assert_se(utf8_console_width(p) <= i);
 155         }
 156
 157         for (size_t i = 0; i < 20; i++) {
 158                 _cleanup_free_ char *p;
 159
 160                 p = utf8_escape_non_printable_full("\xef\xbf\x30\x13", i);
 161                 puts(p);
 162                 assert_se(utf8_is_valid(p));
 163                 assert_se(utf8_console_width(p) <= i);
 164         }
 165 }
 166
 167 static void test_utf16_to_utf8(void) {
 168         const char16_t utf16[] = { htole16('a'), htole16(0xd800), htole16('b'), htole16(0xdc00), htole16('c'), htole16(0xd801), htole16(0xdc37) };
 169         static const char utf8[] = { 'a', 'b', 'c', 0xf0, 0x90, 0x90, 0xb7 };
 170         _cleanup_free_ char16_t *b = NULL;
 171         _cleanup_free_ char *a = NULL;
 172
 173         log_info("/* %s */", __func__);
 174
 175         /* Convert UTF-16 to UTF-8, filtering embedded bad chars */
 176         a = utf16_to_utf8(utf16, sizeof(utf16));
 177         assert_se(a);
 178         assert_se(memcmp(a, utf8, sizeof(utf8)) == 0);
 179
 180         /* Convert UTF-8 to UTF-16, and back */
 181         b = utf8_to_utf16(utf8, sizeof(utf8));
 182         assert_se(b);
 183
 184         free(a);
 185         a = utf16_to_utf8(b, char16_strlen(b) * 2);
 186         assert_se(a);
 187         assert_se(strlen(a) == sizeof(utf8));
 188         assert_se(memcmp(a, utf8, sizeof(utf8)) == 0);
 189 }
 190
 191 static void test_utf8_n_codepoints(void) {
 192         log_info("/* %s */", __func__);
 193
 194         assert_se(utf8_n_codepoints("abc") == 3);
 195         assert_se(utf8_n_codepoints("zażółcić gęślą jaźń") == 19);
 196         assert_se(utf8_n_codepoints("串") == 1);
 197         assert_se(utf8_n_codepoints("") == 0);
 198         assert_se(utf8_n_codepoints("…👊🔪💐…") == 5);
 199         assert_se(utf8_n_codepoints("\xF1") == (size_t) -1);
 200 }
 201
 202 static void test_utf8_console_width(void) {
 203         log_info("/* %s */", __func__);
 204
 205         assert_se(utf8_console_width("abc") == 3);
 206         assert_se(utf8_console_width("zażółcić gęślą jaźń") == 19);
 207         assert_se(utf8_console_width("串") == 2);
 208         assert_se(utf8_console_width("") == 0);
 209         assert_se(utf8_console_width("…👊🔪💐…") == 8);
 210         assert_se(utf8_console_width("\xF1") == (size_t) -1);
 211 }
 212
 213 static void test_utf8_to_utf16(void) {
 214         const char *p;
 215
 216         log_info("/* %s */", __func__);
 217
 218         FOREACH_STRING(p,
 219                        "abc",
 220                        "zażółcić gęślą jaźń",
 221                        "串",
 222                        "",
 223                        "…👊🔪💐…") {
 224
 225                 _cleanup_free_ char16_t *a = NULL;
 226                 _cleanup_free_ char *b = NULL;
 227
 228                 a = utf8_to_utf16(p, strlen(p));
 229                 assert_se(a);
 230
 231                 b = utf16_to_utf8(a, char16_strlen(a) * 2);
 232                 assert_se(b);
 233                 assert_se(streq(p, b));
 234         }
 235 }
 236
 237 int main(int argc, char *argv[]) {
 238         test_utf8_n_is_valid();
 239         test_utf8_is_valid();
 240         test_utf8_is_printable();
 241         test_ascii_is_valid();
 242         test_ascii_is_valid_n();
 243         test_utf8_encoded_valid_unichar();
 244         test_utf8_escape_invalid();
 245         test_utf8_escape_non_printable();
 246         test_utf8_escape_non_printable_full();
 247         test_utf16_to_utf8();
 248         test_utf8_n_codepoints();
 249         test_utf8_console_width();
 250         test_utf8_to_utf16();
 251
 252         return 0;
 253 }