String anonymisation

anonymize.php

<?php

/**
 * This tries to anonymise all letters in a string.
 *
 * @param  string $s
 * @return string
 */
function anonymize_letters($s) {
    $tokens = str_split($s, 1);
    $ns = '';
    $l = strlen($s);
    for ($i = 0; $i < $l; $i++) {
        $cp = ord($s[$i]);
        if ($cp < 128) {
            $offset = null;
            $range = null;

            if ($cp > 64 && $cp < 91) { // roman capital letters
                $offset = 65;
                $range = 25;
            } elseif ($cp > 96 && $cp < 123) { // roman lowercase letters
                $offset = 97;
                $range = 25;
            } elseif ($cp > 47 && $cp < 58) { // numbers
                $offset = 48;
                $range = 9;
            }

            if ($offset && $range) {
                $ns .= chr($offset + \mt_rand(0, $range));
            } else {
                $ns .= $s[$i];
            }
        } else { // a non-ASCII, hopefully UTF-8 string
            $t = '';
            // byte length according to UTF-8 standard
            $bl = $cp < 224 ? 2 : ($cp < 240 ? 3 : 4);

            if ($cp < 224) {
                $t = $s[$i] . $s[$i + 1];
                $i++;
            } elseif ($cp < 240) {
                $t = $s[$i] . $s[$i + 1] . $s[$i + 2];
                $i += 2;
            } else {
                $t = $s[$i] . $s[$i + 1] . $s[$i + 2] . $s[$i + 3];
                $i += 3;
            }

            if (preg_match('/\p{L}/', $t)) {
                $cp = mb_ord($t);
                $n = $t;
                $j = 0;
                do {
                    $n = mb_chr($cp + (\mt_rand(-10, 10))); // find a letter nearby
                    $j++;
                } while ($j < 10 && (!preg_match('/\p{L}/', $n) || ctype_upper($n) !== ctype_upper($t)));
                $ns .= $n;
            } else {
                $ns .= $t;
            }
        }
    }
    return $ns;
}