If you by writing "non letters and numbers" exclude more than [A-Za-z0-9]
(ie. considering letters like åäö
to be letters to) and want to be able to accurately handle UTF-8 strings \p{L}
and \p{N}
will be of aid.
\p{N}
will match any "Number"\p{L}
will match any "Letter Character", which includes
Documentation PHP: Unicode Character Properties
$data = "Thäre!wouldn't%bé#äny";
$new_data = str_replace ("'", "", $data);
$new_data = preg_replace ('/[^\p{L}\p{N}]/u', '_', $new_data);
var_dump (
$new_data
);
output
string(23) "Thäre_wouldnt_bé_äny"