SlevomatCodingStandard.PHP.UselessParentheses.UselessParentheses - false positive

[bkdotcom]

These parens aren't useless... am I doing something wrong?

class Utf8
{
    // Get Unicode code point of utf9-encoded character (without need for mbstring extension)
    public static function ord($char)
    {
        $ord = \ord($char[0]);
        if ($ord < 0x80) {
            return $ord;
        } elseif ($ord < 0xe0) {
            return ($ord - 0xc0 << 6)        //  UselessParentheses (invalid)
                + \ord($char[1]) - 0x80;
        } elseif ($ord < 0xf0) {
            return ($ord - 0xe0 << 12)      // UselessParentheses (invalid)
                + (\ord($char[1]) - 0x80 << 6)   // UselessParentheses (invalid)
                + \ord($char[2]) - 0x80;
        } elseif ($ord < 0xf8) {
            return ($ord - 0xf0 << 18)      // UselessParentheses (invalid)
                + (\ord($char[1]) - 0x80 << 12)  // UselessParentheses (invalid)
                + (\ord($char[2]) - 0x80 << 6)   // UselessParentheses (invalid)
                + \ord($char[3]) - 0x80;
        }
        return false;
    }
}

unit tests:

assertSame(97, Utf8::ord('a'));   // 1-bype
assertSame(169, Utf8::ord('©'));   // 2-byte
assertSame(65049, Utf8::ord('︙'));   // 3-byte
assertSame(128169, Utf8::ord('💩'));  // 4-type

to be clear: without parens, PHP evaluates $ord - 0xc0 << 6 + \ord($char[1]) - 0x80 as ($ord - 0xc0) << (6 + \ord($char[1]) - 0x80)

so... ($ord - 0xc0 << 6) + \ord($char[1]) - 0x80 the parens are very much necessary also tried: ($ord - 0xc0 << 6) + (\ord($char[1]) - 0x80) and $ord - 0xc0 << 6 + (\ord($char[1]) - 0x80)

[bkdotcom]

Possibly similar to #1672, #1678