/**
* Enhanced strip_tags() :: will revert also special entities like nbsp; and more
*
* @param ARRAY $yhtmlcode :: HTML Code to be stripped of tags
* @param YES/NO $y_mode :: yes to convert <br> to new lines \n, otherwise (if no) will convert <br> to spaces
*
* @return STRING :: The processed HTML Code
*/
public static function striptags($yhtmlcode, $ynewline = 'yes')
{
//--
$yhtmlcode = (string) $yhtmlcode;
$ynewline = (string) $ynewline;
//-- fix xhtml tag ends and add spaces between tags
$yhtmlcode = (string) str_replace(array(' />', '/>', '>'), array('>', '>', '> '), (string) $yhtmlcode);
//-- remove special tags
$html_regex_h = array('#<head[^>]*?>.*?</head[^>]*?>#si', '#<style[^>]*?>.*?</style[^>]*?>#si', '#<script[^>]*?>.*?</script[^>]*?>#si', '#<noscript[^>]*?>.*?</noscript[^>]*?>#si', '#<frameset[^>]*?>.*?</frameset[^>]*?>#si', '#<frame[^>]*?>.*?</frame[^>]*?>#si', '#<iframe[^>]*?>.*?</iframe[^>]*?>#si', '#<canvas[^>]*?>.*?</canvas[^>]*?>#si', '#<audio[^>]*?>.*?</audio[^>]*?>#si', '#<video[^>]*?>.*?</video[^>]*?>#si', '#<applet[^>]*?>.*?</applet[^>]*?>#si', '#<param[^>]*?>.*?</param[^>]*?>#si', '#<object[^>]*?>.*?</object[^>]*?>#si', '#<form[^>]*?>.*?</form[^>]*?>#si', '#<link[^>]*?>#si', '#<img[^>]*?>#si');
$html_regex_r = array(' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ');
$yhtmlcode = (string) preg_replace((array) $html_regex_h, (array) $html_regex_r, (string) $yhtmlcode);
$yhtmlcode = str_replace(["\r\n", "\r", "\t", "\f"], ["\n", "\n", ' ', ' '], $yhtmlcode);
//-- replace new line tags
if ((string) $ynewline == 'yes') {
$yhtmlcode = (string) str_ireplace(['<br>', '</br>'], ["\n", ''], (string) $yhtmlcode);
} else {
$yhtmlcode = (string) str_ireplace(['<br>', '</br>'], [' ', ''], (string) $yhtmlcode);
}
//end if else
//-- strip the tags
$yhtmlcode = (string) strip_tags((string) $yhtmlcode);
//-- restore some usual html entities
$regex_h = array(' ', '&', '"', '<', '>', '©', '€', '·');
$regex_r = array(' ', '&', '"', '<', '>', '(c)', 'EURO', '.');
$yhtmlcode = (string) str_ireplace((array) $regex_h, (array) $regex_r, (string) $yhtmlcode);
$yhtmlcode = (string) strip_tags((string) $yhtmlcode);
// fix: after all fixes when reversing entities, new tags can appear that were encoded, so needs run again for safety ...
//-- restore html unicode entities
$html_accents = (array) SmartUnicode::accented_html_entities();
$yhtmlcode = (string) str_replace((array) array_values($html_accents), (array) array_keys($html_accents), (string) $yhtmlcode);
//-- clean any other remaining html entities
$yhtmlcode = (string) preg_replace('/&\\#?([0-9a-z]+);/i', ' ', (string) $yhtmlcode);
//-- cleanup multiple spaces with just one space
$yhtmlcode = (string) preg_replace('/[ \\t]+/', ' ', (string) $yhtmlcode);
// replace any horizontal whitespace character ' since PHP 5.2.4 can be /[\h]+/
$yhtmlcode = (string) preg_replace('/^\\s*[\\n]{2,}/m', '', (string) $yhtmlcode);
// fix: replace multiple consecutive lines that may also contain before optional leading spaces
$yhtmlcode = (string) preg_replace('/[^\\S\\r\\n]+$/m', '', (string) $yhtmlcode);
// remove trailing spaces on each line
//--
return (string) trim((string) $yhtmlcode);
//--
}
