/**
* Cleans the html of any none-html information.
*
* @param string $str
* @return string
*/
protected function clean($str)
{
if ($this->options->get('cleanupInput') != true) {
// skip entire cleanup step
return $str;
}
// remove white space before closing tags
$str = mb_eregi_replace("'\\s+>", "'>", $str);
$str = mb_eregi_replace('"\\s+>', '">', $str);
// clean out the \n\r
$replace = ' ';
if ($this->options->get('preserveLineBreaks')) {
$replace = ' ';
}
$str = str_replace(["\r\n", "\r", "\n"], $replace, $str);
// strip the doctype
$str = mb_eregi_replace("<!doctype(.*?)>", '', $str);
// strip out comments
$str = mb_eregi_replace("<!--(.*?)-->", '', $str);
// strip out cdata
$str = mb_eregi_replace("<!\\[CDATA\\[(.*?)\\]\\]>", '', $str);
// strip out <script> tags
if ($this->options->get('removeScripts') == true) {
$str = mb_eregi_replace("<\\s*script[^>]*[^/]>(.*?)<\\s*/\\s*script\\s*>", '', $str);
$str = mb_eregi_replace("<\\s*script\\s*>(.*?)<\\s*/\\s*script\\s*>", '', $str);
}
// strip out <style> tags
if ($this->options->get('removeStyles') == true) {
$str = mb_eregi_replace("<\\s*style[^>]*[^/]>(.*?)<\\s*/\\s*style\\s*>", '', $str);
$str = mb_eregi_replace("<\\s*style\\s*>(.*?)<\\s*/\\s*style\\s*>", '', $str);
}
// strip out server side scripts
$str = mb_eregi_replace("(<\\?)(.*?)(\\?>)", '', $str);
// strip smarty scripts
$str = mb_eregi_replace("(\\{\\w)(.*?)(\\})", '', $str);
return $str;
}
public function testAddingOver()
{
$options = new Options();
$options->setOptions(['test' => false])->setOptions(['test' => true, 'whitespaceTextNode' => false]);
$this->assertFalse($options->get('whitespaceTextNode'));
}
public function testGettingNoOption()
{
$options = new Options();
$this->assertEquals(null, $options->get('doesnotexist'));
}
