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") == 3);
  40         assert_se(utf8_encoded_valid_unichar("\302\256") == 2);
  41         assert_se(utf8_encoded_valid_unichar("a") == 1);
  42         assert_se(utf8_encoded_valid_unichar("\341\204") < 0);
  43         assert_se(utf8_encoded_valid_unichar("\341\204\341\204") < 0);
  44 }
  45
  46 static void test_utf8_escaping(void) {
  47         _cleanup_free_ char *p1, *p2, *p3;
  48
  49         p1 = utf8_escape_invalid("goo goo goo");
  50         puts(p1);
  51         assert_se(utf8_is_valid(p1));
  52
  53         p2 = utf8_escape_invalid("\341\204\341\204");
  54         puts(p2);
  55         assert_se(utf8_is_valid(p2));
  56
  57         p3 = utf8_escape_invalid("\341\204");
  58         puts(p3);
  59         assert_se(utf8_is_valid(p3));
  60 }
  61
  62 static void test_utf8_escaping_printable(void) {
  63         _cleanup_free_ char *p1, *p2, *p3, *p4, *p5, *p6;
  64
  65         p1 = utf8_escape_non_printable("goo goo goo");
  66         puts(p1);
  67         assert_se(utf8_is_valid(p1));
  68
  69         p2 = utf8_escape_non_printable("\341\204\341\204");
  70         puts(p2);
  71         assert_se(utf8_is_valid(p2));
  72
  73         p3 = utf8_escape_non_printable("\341\204");
  74         puts(p3);
  75         assert_se(utf8_is_valid(p3));
  76
  77         p4 = utf8_escape_non_printable("ąę\n가너도루\n1234\n\341\204\341\204\n\001 \019\20\a");
  78         puts(p4);
  79         assert_se(utf8_is_valid(p4));
  80
  81         p5 = utf8_escape_non_printable("\001 \019\20\a");
  82         puts(p5);
  83         assert_se(utf8_is_valid(p5));
  84
  85         p6 = utf8_escape_non_printable("\xef\xbf\x30\x13");
  86         puts(p6);
  87         assert_se(utf8_is_valid(p6));
  88 }
  89
  90 static void test_utf16_to_utf8(void) {
  91         const char16_t utf16[] = { htole16('a'), htole16(0xd800), htole16('b'), htole16(0xdc00), htole16('c'), htole16(0xd801), htole16(0xdc37) };
  92         static const char utf8[] = { 'a', 'b', 'c', 0xf0, 0x90, 0x90, 0xb7 };
  93         _cleanup_free_ char16_t *b = NULL;
  94         _cleanup_free_ char *a = NULL;
  95
  96         /* Convert UTF-16 to UTF-8, filtering embedded bad chars */
  97         a = utf16_to_utf8(utf16, sizeof(utf16));
  98         assert_se(a);
  99         assert_se(memcmp(a, utf8, sizeof(utf8)) == 0);
 100
 101         /* Convert UTF-8 to UTF-16, and back */
 102         b = utf8_to_utf16(utf8, sizeof(utf8));
 103         assert_se(b);
 104
 105         free(a);
 106         a = utf16_to_utf8(b, char16_strlen(b) * 2);
 107         assert_se(a);
 108         assert_se(strlen(a) == sizeof(utf8));
 109         assert_se(memcmp(a, utf8, sizeof(utf8)) == 0);
 110 }
 111
 112 static void test_utf8_n_codepoints(void) {
 113         assert_se(utf8_n_codepoints("abc") == 3);
 114         assert_se(utf8_n_codepoints("zażółcić gęślą jaźń") == 19);
 115         assert_se(utf8_n_codepoints("串") == 1);
 116         assert_se(utf8_n_codepoints("") == 0);
 117         assert_se(utf8_n_codepoints("…👊🔪💐…") == 5);
 118         assert_se(utf8_n_codepoints("\xF1") == (size_t) -1);
 119 }
 120
 121 static void test_utf8_console_width(void) {
 122         assert_se(utf8_console_width("abc") == 3);
 123         assert_se(utf8_console_width("zażółcić gęślą jaźń") == 19);
 124         assert_se(utf8_console_width("串") == 2);
 125         assert_se(utf8_console_width("") == 0);
 126         assert_se(utf8_console_width("…👊🔪💐…") == 8);
 127         assert_se(utf8_console_width("\xF1") == (size_t) -1);
 128 }
 129
 130 static void test_utf8_to_utf16(void) {
 131         const char *p;
 132
 133         FOREACH_STRING(p,
 134                        "abc",
 135                        "zażółcić gęślą jaźń",
 136                        "串",
 137                        "",
 138                        "…👊🔪💐…") {
 139
 140                 _cleanup_free_ char16_t *a = NULL;
 141                 _cleanup_free_ char *b = NULL;
 142
 143                 a = utf8_to_utf16(p, strlen(p));
 144                 assert_se(a);
 145
 146                 b = utf16_to_utf8(a, char16_strlen(a) * 2);
 147                 assert_se(b);
 148                 assert_se(streq(p, b));
 149         }
 150 }
 151
 152 int main(int argc, char *argv[]) {
 153         test_utf8_is_valid();
 154         test_utf8_is_printable();
 155         test_ascii_is_valid();
 156         test_ascii_is_valid_n();
 157         test_utf8_encoded_valid_unichar();
 158         test_utf8_escaping();
 159         test_utf8_escaping_printable();
 160         test_utf16_to_utf8();
 161         test_utf8_n_codepoints();
 162         test_utf8_console_width();
 163         test_utf8_to_utf16();
 164
 165         return 0;
 166 }