/**
* Recursively cleans arrays, objects, and strings. Removes ASCII control
* codes and converts to the requested charset while silently discarding
* incompatible characters.
*
* UTF8::clean($_GET); // Clean GET data
*
* @param mixed $var variable to clean
* @param string $charset character set, defaults to Phalcana\Phalcana::$charset
* @return mixed
* @uses UTF8::clean
* @uses UTF8::stripAsciiCtrl
* @uses UTF8::isAscii
*/
public static function clean($var, $charset = null)
{
if (!$charset) {
// Use the application character set
$charset = Phalcana::$charset;
}
if (is_array($var) || is_object($var)) {
foreach ($var as $key => $val) {
// Recursion!
$var[UTF8::clean($key)] = UTF8::clean($val);
}
} elseif (is_string($var) && $var !== '') {
// Remove control characters
$var = UTF8::stripAsciiCtrl($var);
if (!UTF8::isAscii($var)) {
// Temporarily save the mb_substitute_character() value into a variable
$mb_substitute_character = mb_substitute_character();
// Disable substituting illegal characters with the default '?' character
mb_substitute_character('none');
// convert encoding, this is expensive, used when $var is not ASCII
$var = mb_convert_encoding($var, $charset, $charset);
// Reset mb_substitute_character() value back to the original setting
mb_substitute_character($mb_substitute_character);
}
}
return $var;
}
/**
* Tests UTF8::clean
*
* @test
* @dataProvider provider_clean
*/
public function test_clean($input, $expected)
{
$this->assertSame($expected, UTF8::clean($input));
}