This should also work, taking into account, for example, page anchors (#) and at least some of those "weird characters" you mention but don't seem worried about:
function remove_query_part($url, $term)
{
$query_str = parse_url($url, PHP_URL_QUERY);
if ($frag = parse_url($url, PHP_URL_FRAGMENT)) {
$frag = '#' . $frag;
}
parse_str($query_str, $query_arr);
unset($query_arr[$term]);
$new = '?' . http_build_query($query_arr) . $frag;
return str_replace(strstr($url, '?'), $new, $url);
}
Demo:
$string[] = 'http://domain.com/php/doc.php?arg1=0&arg2=1&arg3=0';
$string[] = 'http://domain.com/php/doc.php?arg1=0&arg2=1';
$string[] = 'http://domain.com/php/doc.php?arg1=0&arg2=1&arg3=0#frag';
$string[] = 'http://domain.com/php/doc.php?arg1=0&arg2=1&arg3=0&arg4=4';
$string[] = 'http://domain.com/php/doc.php';
$string[] = 'http://domain.com/php/doc.php#frag';
$string[] = 'http://example.com?arg1=question?mark&arg2=equal=sign&arg3=hello';
foreach ($string as $str) {
echo remove_query_part($str, 'arg3') . "\n";
}
Output:
http://domain.com/php/doc.php?arg1=0&arg2=1
http://domain.com/php/doc.php?arg1=0&arg2=1
http://domain.com/php/doc.php?arg1=0&arg2=1#frag
http://domain.com/php/doc.php?arg1=0&arg2=1&arg4=4
http://domain.com/php/doc.php
http://domain.com/php/doc.php#frag
http://example.com?arg1=question%3Fmark&arg2=equal%3Dsign
Tested only as shown.