$n_passed codepoints passed, " . count ($failures) . " failed.\n";
if (!empty($failures))
{
echo "\n";
echo "\n";
echo '';
echo '| Codepoint | ';
echo 'mbstring | mrbs | Summary | ';
echo "
\n";
echo "\n";
echo "\n";
foreach ($failures as $failure)
{
echo '';
if ($intl_loaded)
{
echo '| ' . IntlChar::charName($failure[0]) . ' | ';
}
foreach ($failure as $value)
{
echo "$value | ";
}
echo 'Fail | ' . "\n";
echo "
\n";
}
echo "\n";
echo "
\n";
}
}
function test_ord() : void
{
global $color_fail, $intl_loaded, $max_codepoint;
$n_passed = 0;
$failures = [];
for ($i =0; $i<=$max_codepoint; $i++)
{
$str = mb_chr($i, 'UTF-8');
if (($str !== false) && mb_check_encoding($str, 'UTF-8'))
{
$mrbs_ord = Mbstring::mb_ord($str);
if ($mrbs_ord === mb_ord($str))
{
$n_passed++;
}
else
{
$failures[] = [$str, $i, $mrbs_ord];
}
}
}
echo "$n_passed codepoints passed, " . count ($failures) . " failed.
\n";
if (!empty($failures))
{
echo "\n";
echo "\n";
echo '';
echo '| Char | ';
echo 'mbstring | mrbs | Summary | ';
echo "
\n";
echo "\n";
echo "\n";
foreach ($failures as $failure)
{
echo '';
if ($intl_loaded)
{
echo '| ' . IntlChar::charName(mb_ord($failure[0])) . ' | ';
}
foreach ($failure as $value)
{
echo "$value | ";
}
echo 'Fail | ' . "\n";
echo "
\n";
}
echo "\n";
echo "
\n";
}
}
function codepoint_notation(int $codepoint) : string
{
// OK to user strtoupper here instead of mb_ because we're only looking at the hex characters
return 'U+' . str_pad(strtoupper(dechex($codepoint)), 4, '0', STR_PAD_LEFT);
}
function test(string $function, $args) : void
{
global $color_fail, $color_pass;
echo "";
echo "| $function | ";
foreach ($args as $arg)
{
echo "$arg | ";
}
// Using the mbstring versions
echo "";
try {
$mbstring = call_user_func_array($function, $args);
}
catch (Throwable $t) {
$mbstring = get_class($t);
}
echo var_export($mbstring, true);
echo " | ";
// Using the MRBS emulations
echo "";
try {
$mrbs = call_user_func_array([__NAMESPACE__ . "\\Mbstring\\Mbstring", $function], $args);
}
catch (Throwable $t) {
$mrbs = get_class($t);
}
echo var_export($mrbs, true);
echo " | ";
// Compare the results
$color = ($mbstring === $mrbs) ? $color_pass : $color_fail;
echo '';
echo ($mbstring === $mrbs) ? 'Pass' : 'Fail';
echo " | ";
echo "
\n";
}
function test_strlen() : void
{
echo "\n";
echo thead_html(['string', 'encoding']);
echo "\n";
// Simple case
test('mb_strlen', ['abcd', 'UTF-8']);
// Multibyte
test('mb_strlen', ['會議室預約系統', 'UTF-8']);
test('mb_strlen', ['emojis 😀😨🙁', 'UTF-8']);
// Empty string
test('mb_strlen', ['', 'UTF-8']);
// 8bit testing
test('mb_strlen', ['', '8bit']);
test('mb_strlen', ['&', '8bit']);
test('mb_strlen', ['å', '8bit']);
test('mb_strlen', ['議', '8bit']);
test('mb_strlen', ['👽', '8bit']);
test('mb_strlen', ['z👽', '8bit']);
test('mb_strlen', ['åäö', '8bit']);
test('mb_strlen', ['👽統', '8bit']);
test('mb_strlen', ['👿🤩', '8bit']);
test('mb_strlen', ['系統åg', '8bit']);
echo "\n";
echo "
\n";
}
function test_all_codepoints(string $function) : void
{
global $color_fail, $intl_loaded, $max_codepoint;
echo "Testing all codepoints
\n";
$n_passed = 0;
$failures = [];
for ($i =0; $i<=$max_codepoint; $i++)
{
$str = mb_chr($i, 'UTF-8');
if (($str !== false) && mb_check_encoding($str, 'UTF-8'))
{
$mb = call_user_func($function, $str);
$mrbs = call_user_func([__NAMESPACE__ . "\\Mbstring\\Mbstring", $function], $str);
if ($mb === $mrbs)
{
$n_passed++;
}
else
{
$failures[] = [$str, $mb, $mrbs];
}
}
}
echo "$n_passed codepoints passed, " . count ($failures) . " failed.
\n";
if (!empty($failures))
{
echo "\n";
echo "\n";
echo '';
echo '| Codepoint | ';
echo 'mbstring | mrbs | Summary | ';
echo "
\n";
echo "\n";
echo "\n";
foreach ($failures as $failure)
{
echo '';
if ($intl_loaded)
{
echo '| ' . IntlChar::charName(mb_ord($failure[0])) . ' | ';
}
foreach ($failure as $char)
{
echo "$char | " . codepoint_notation(mb_ord($char)) . ' | ';
}
echo 'Fail | ' . "\n";
echo "
\n";
}
echo "\n";
echo "
\n";
}
}
function test_strtolower() : void
{
echo "\n";
echo thead_html(['string']);
echo "\n";
// Empty string
test('mb_strtolower', ['']);
// Simple string
test('mb_strtolower', ['ABcDeFgHI']);
// More complex
test('mb_strtolower', ['AÅÄÖ']);
// Turkish characters
test('mb_strtolower', ['CÇGĞIİSŞ']);
test('mb_strtolower', ['İ']);
// Other
test('mb_strtolower', ['Τάχιστη αλώπηξ βαφής']);
test('mb_strtolower', ['👽系😨z😎éÉ']);
echo "\n";
echo "
\n";
test_all_codepoints('mb_strtolower');
}
function test_strtoupper() : void
{
echo "\n";
echo thead_html(['string']);
echo "\n";
// Empty string
test('mb_strtoupper', ['']);
// Simple string
test('mb_strtoupper', ['ABcDeFgHI']);
// More complex
test('mb_strtoupper', ['aåäö']);
// Turkish characters
test('mb_strtoupper', ['cçgğiiı̇sş']);
// Other
test('mb_strtoupper', ['Τάχιστη αλώπηξ βαφής']);
test('mb_strtoupper', ['👽系😨z😎éÉ']);
// These fail with Transliterator
test('mb_strtoupper', ['ƛɤ']);
echo "\n";
echo "
\n";
test_all_codepoints('mb_strtoupper');
}
function test_substr() : void
{
echo "\n";
echo thead_html(['string', 'start', 'length']);
echo "\n";
// Empty string
test('mb_substr', ['', 0, null]);
test('mb_substr', ['', 1, null]);
test('mb_substr', ['', 0, 2]);
test('mb_substr', ['', 1, 2]);
// Multibyte
test('mb_substr', ['👽系😨z😎é', 0, null]);
test('mb_substr', ['👽系😨z😎é', 2, null]);
test('mb_substr', ['👽系😨z😎é', 2, 1]);
test('mb_substr', ['👽系😨z😎é', 2, -1]);
test('mb_substr', ['👽系😨z😎é', 2, -5]);
test('mb_substr', ['👽系😨z😎é', -1, null]);
test('mb_substr', ['👽系😨z😎é', -3, -1]);
test('mb_substr', ['👽系😨z😎é', -9, -1]);
test('mb_substr', ['👽系😨z😎é', -9, -12]);
echo "\n";
echo "
\n";
}
function test_pos() : void
{
echo "\n";
echo thead_html(['haystack', 'needle', 'offset']);
echo "\n";
// mb_strpos()
// -----------
test('mb_strpos', ['0123456789a0123456789b0123456789c', 'c', 0]);
test('mb_strpos', ['0123456789a0123456789b0123456789c', 'd', 0]);
test('mb_strpos', ['0123456789a0123456789b0123456789c', 'c', -1]);
test('mb_strpos', ['0123456789a0123456789b0123456789c', 'c', -2]);
test('mb_strpos', ['TRUE', 'E_', 0]);
// Equivalence
test('mb_strpos', ['Jour précédent', 'e', 0]);
$old_locale = setlocale(LC_ALL, '0');
setlocale(LC_ALL, ['fr_FR', 'fr']);
test('mb_strpos', ['Jour précédent', 'e', 0]);
setlocale(LC_ALL, $old_locale);
// mb_stripos()
// -----------
test('mb_stripos', ['0123456789a0123456789b0123456789c', 'c', -1]);
test('mb_stripos', ['0123456789a0123456789b0123456789c', 'C', -1]);
// Multibyte
test('mb_stripos', ['會C議室預約系統', 'C', 2]);
test('mb_stripos', ['會議C室預約系統', 'C', 2]);
test('mb_stripos', ['會議C室預約系統', '預約', 2]);
// mb_strrpos()
// ------------
// Positive offsets, needle at start
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '0123456789a', 0]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '0123456789a', 1]);
// Positive offsets, needle partial match at the end
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '89cd', 10]);
// Positive offsets, needle longer than search area
test('mb_strrpos', ['abcde', 'cde', 2]);
test('mb_strrpos', ['abcde', 'cde', 3]);
// Negative offsets, needle longer than search area
test('mb_strrpos', ['abcdefg', 'cdefghi', -2]);
test('mb_strrpos', ['abcdefg', 'cdefghij', -2]);
// Negative offsets, needle in the middle of haystack
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '234', 0]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '234', -1]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '234', -2]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '234', -3]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '234', -9]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '234', -10]);
// Negative offsets, needle at the end of haystack
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789c', 0]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789c', -1]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789c', -2]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789c', -3]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789c', -4]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789c', -5]);
// Negative offsets, needle partial match at the end
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789cd', 0]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789cd', -1]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789cd', -2]);
test('mb_strrpos', ['0123456789a0123456789b0123456789c', '789cd', -3]);
// Multibyte
test('mb_strrpos', ['會🙂議C室🙂預約系統', '🙂', -1]);
// Empty haystack
test('mb_strrpos', ['', 'A', 0]);
// Case sensitivity
test('mb_strrpos', ['AaBb', 'a', 0]);
test('mb_strrpos', ['AaBb', 'A', 0]);
// Offset outside haystack
test('mb_strrpos', ['', 'A', 1]);
test('mb_strrpos', ['A', 'A', 2]);
test('mb_strrpos', ['A', 'A', -2]);
// mb_strripos()
// ------------
// Case sensitivity
test('mb_strripos', ['AaBb', 'a', 0]);
test('mb_strripos', ['AaBb', 'A', 0]);
echo "\n";
echo "
\n";
}
echo "mbstring emulation tests
\n";
$loaded_extensions = get_loaded_extensions();
echo "PHP version: " . PHP_VERSION;
echo "
\n";
echo "mbstring enabled: " . var_export(in_array('mbstring', $loaded_extensions), true);
echo "
\n";
echo "intl enabled: " . var_export(in_array('intl', $loaded_extensions), true);
echo "
\n";
echo "iconv enabled: " . var_export(in_array('iconv', $loaded_extensions), true);
echo "
\n";
echo "
\n";
if (!in_array('mbstring', $loaded_extensions))
{
die("This test needs the 'mbstring' PHP extension to be loaded.");
}
echo "mb_chr()
\n";
test_chr();
echo "mb_ord()
\n";
test_ord();
echo "mb_strlen()
\n";
test_strlen();
echo "mb_strtolower()
\n";
test_strtolower();
echo "mb_strtoupper()
\n";
test_strtoupper();
echo "mb_substr()
\n";
test_substr();
echo "mb_*pos()
\n";
test_pos();