Decodes UTF-8, interpreting ill-formed UTF-8 sequences as CP1252, posted in reply to <http://blog.endpoint.com/2011/12/sanitizing-supposed-utf-8-data.html>

fix_latin.pl

#!/usr/bin/perl
use strict;
use warnings;

use Encode        qw[find_encoding];
use Unicode::UTF8 qw[decode_utf8];

{
    my $encoding = find_encoding('Windows-1252')
      or die q/Couldn't find Windows-1252 encoding/;

    my $fallback = sub {
        my ($octets, $is_usv) = @_;
        return $is_usv ? "\x{FFFD}" : $encoding->decode($octets);
    };

    sub fix_latin {
        @_ == 1 || die q/Usage: fix_latin($octets)/;
        no warnings 'utf8';
        return decode_utf8($_[0], $fallback);
    }
}

my $octets = "\x91 Foo \xE2\x98\xBA \x92";

printf "<%s>\n", 
  join ' ', map { sprintf 'U+%.4X', ord $_ } split //, fix_latin($octets);