/**
* Strips session links, html entities and \r\n from the given text.
*
* @param string $text
* @return string
*/
public static function stripCrap($text)
{
// strip session links, security tokens and access tokens
$text = Regex::compile('(?<=\\?|&)([st]=[a-f0-9]{40}|at=\\d+-[a-f0-9]{40})')->replace($text, '');
// convert html entities (utf-8)
$text = Regex::compile('&#(3[2-9]|[4-9][0-9]|\\d{3,5});')->replace($text, new Callback(function ($matches) {
return StringUtil::getCharacter(intval($matches[1]));
}));
// unify new lines
$text = StringUtil::unifyNewlines($text);
// remove 4 byte utf-8 characters as MySQL < 5.5 does not support them
// see http://stackoverflow.com/a/16902461/782822
$text = preg_replace('/[\\xF0-\\xF7].../s', '', $text);
// remove control characters
$text = preg_replace('~[\\x00-\\x08\\x0B-\\x1F\\x7F]~', '', $text);
return $text;
}
