#!/usr/bin/perl

use strict;
use warnings;
use Encode qw();

BEGIN { eval 'use Unicode::UTF8 qw()'; }      # Must exist for Match to work
BEGIN { eval 'use Unicode::Normalize qw()'; }

# Stolen from src/lib/Sympa/Tools/Text.pm
sub canonic_text {
    my $text = shift;

    return undef unless defined $text;

    # Normalize text. See also discussion on
    # https://listes.renater.fr/sympa/arc/sympa-developpers/2018-03/thrd1.html
    #
    # N.B.: Corresponding modules are optional by now, and should be
    # mandatory in the future.
    my $utext;
    if (Encode::is_utf8($text)) {
        $utext = $text;
    } elsif ($Unicode::UTF8::VERSION) {
        no warnings 'utf8';
        $utext = Unicode::UTF8::decode_utf8($text);
    } else {
        $utext = Encode::decode_utf8($text);
    }
    if ($Unicode::Normalize::VERSION) {
        $utext = Unicode::Normalize::normalize('NFC', $utext);
    }

    # Remove DOS linefeeds (^M) that cause problems with Outlook 98, AOL,
    # and EIMS:
    $utext =~ s/\r\n|\r/\n/g;

    if (Encode::is_utf8($text)) {
        return $utext;
    } else {
        return Encode::encode_utf8($utext);
    }
}

# Derived from t/Tools_Text.t 

my $a = canonic_text(
    "\xED\xA0\x80\n\xF4\x8F\xBF\xBE\n\xF4\x90\x80\x80\n\xF8\x88\x80\x80\x80\n"
    );

my $b = Encode::encode_utf8(
    "\x{FFFD}\x{FFFD}\x{FFFD}\n\x{FFFD}\n\x{FFFD}\x{FFFD}\x{FFFD}\x{FFFD}\n\x{FFFD}\x{FFFD}\x{FFFD}\x{FFFD}\x{FFFD}\n"
    );

if ($a eq $b) {
    print "Match\n";
}