HTML Compressor with regex

I would like to compress a Magento HTML page using some regex, and this is what I have written:

function html_compress($string){

    global $idarray;
    $idarray=array();

    //Replace PRE and TEXTAREA tags
    $search=array(
                    '@(<)\s*?(pre\b[^>]*?)(>)([\s\S]*?)(<)\s*(/\s*?pre\s*?)(>)@',   //Find PRE Tag
                    '@(<)\s*?(textarea\b[^>]*?)(>)([\s\S]*?)(<)\s*?(/\s*?textarea\s*?)(>)@' //Find TEXTAREA
                );
    $string=preg_replace_callback($search,
                                    function($m){
                                        $id='<!['.uniqid().']!>';
                                        global $idarray;
                                        $idarray[]=array($id,$m[0]);
                                        return $id;
                                    },
                                    $string
    );

    //Remove blank useless space
    $search = array(
                    '@( |\t|\f)+@', // Shorten multiple whitespace sequences
                    '@(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+@',   //Remove blank lines
                    '@^(\s)+|( |\t|\0|\r\n)+$@' //Trim Lines
                    );
    $replace = array(' ',"\\1",'');
    $string = preg_replace($search, $replace, $string);

    //Replace IE COMMENTS, SCRIPT, STYLE and CDATA tags
    $search=array(
                    '@<!--\[if\s(?:[^<]+|<(?!!\[endif\]-->))*<!\[endif\]-->@',  //Find IE Comments
                    '@(<)\s*?(script\b[^>]*?)(>)([\s\S]*?)(<)\s*?(/\s*?script\s*?)(>)@',    //Find SCRIPT Tag
                    '@(<)\s*?(style\b[^>]*?)(>)([\s\S]*?)(<)\s*?(/\s*?style\s*?)(>)@',  //Find STYLE Tag
                    '@(//<!\[CDATA\[([\s\S]*?)//]]>)@', //Find commented CDATA
                    '@(<!\[CDATA\[([\s\S]*?)]]>)@'  //Find CDATA
                );
    $string=preg_replace_callback($search,
                                    function($m){
                                        $id='<!['.uniqid().']!>';
                                        global $idarray;
                                        $idarray[]=array($id,$m[0]);
                                        return $id;
                                    },
                                    $string
    );

    //Remove blank useless space
    $search = array(
                    '@(class|id|value|alt|href|src|style|title)=(\'\s*?\'|"\s*?")@',    //Remove empty attribute
                    '@<!--([\s\S]*?)-->@',  // Strip comments except IE
                    '@[\r\n|\n|\r]@', // Strip break line
                    '@[ |\t|\f]+@', // Shorten multiple whitespace sequences
                    '@(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+@', //Remove blank lines
                    '@^(\s)+|( |\t|\0|\r\n)+$@' //Trim Lines
                    );
    $replace = array(' ','',' ',' ',"\\1",'');
    $string = preg_replace($search, $replace, $string);

    //Replace unique id with original tag
    $c=count($idarray);
    for($i=0;$i<$c;$i++){
        $string = str_replace($idarray[$i][0], "\n".$idarray[$i][1]."\n", $string);
    }

    return $string;
}

It works, but I have got some concerns:

• Has it got sense to explit all the \s*? between the tags and retrieve with this (<)\s*?(style\b[^>]*?)(>)?
• Does this script eat resources and considerably delay the page load? Is there any possible optimization?
• Is the part that remove the white space redundant?
• Is all of this "cacheable"?