Translating utf8 byte strings

decode_utf8.pl

# Steal bits, but don't take at face value
# Originally stolen from https://gist.github.com/nicwolff/1484073
sub _decode {
  my ($self, $in) = @_;
  my $log = $self->app->log;
  my $out = '';
  use bytes;  # Who said C was dead?
  my ($i, $length) = (0, length $in);
  while ($i < $length) {
    my $b = substr $in, $i, 1;
    if (ord $b < 0x80) {
      # 7-bit ascii
      $out .= $b;
    }
    elsif (ord($b) >> 5 == 6 and $i + 1 < $length - 1) {
      # Possible 2-byte char 
      my $next = substr $in, $i + 1, 1;
      if (ord($next) >> 6 == 2) { 
        $out .= $b . $next;
        ++$i;
      }
      else {
        $out .= decode 'cp1252', $b;
      }
    } 
    elsif (ord($b) >> 4 == 14 and $i + 2 < $length - 1) {
      # Possible 3-byte char
      my $next      = substr $in, $i + 1, 1;
      my $nextagain = substr $in, $i + 2, 1;
      if (ord($next) >> 6 == 2 and ord($nextagain) >> 6 == 2) {
        $out .= $b . $next . $nextagain;
        $i += 2;
      }
      else {
        $out .= decode 'cp1252', $b;
      }
    } 
    else {
      $out .= decode 'cp1252', $b;
    }
    ++$i;
  } 
  return $out;
}