thanks for your code, sorry I didn't explain the problem clear enough.

The input could be encoded in iso-8859-1 \x{f6}\x{f6}, or, maybe in utf-8, \x{c3}\x{b6}, I have to find out what is the charset first.

Encode::Detect::Detector is the one I am using to find out what is the charset of the string, utf-8 or iso-8859-1,

the logic is like:
$charset = = Encode::Detect::Detector::detect($input);
if($charset eq 'UTF-8'){
  # do NFC ...
}elsif($charset eq 'iso-8859-1'){
  # do NFD ...
}
[download]

Text::Unaccent unac_string($charset, $str) in my case. Text::Unaccent is working well if Detector can find it the correct code, it failed if Detector failed, of course, no charset.

Encode::Detect::Detector normally working well, but failed if input = \x{f6}\x{f6}.