Take a look at the Unicode::String module on CPAN, specifically at the byteswap method. While reading the file, if your BOM is not in network order, byteswap each line then decode it to utf-8. Reverse the process as you write. Incidentally, ucs2 is another name for utf-16.

Here's a small demo script.

use warnings;
use strict;
use Unicode::String qw/byteswap2/;
use Encode qw/encode decode/;

load('Language.msg');

sub load {
    my $file = shift;
    open my $FH, '<:bytes', $file or die "$file: $!\n";
    my $char;
    sysread( $FH, $char, 2, 0 );
    my $swapbyteorder;
    if ( $char eq "\x{FF}\x{FE}" ) {
        $swapbyteorder = 1;
    }
    elsif ( $char eq "\x{FE}\x{FF}" ) {
        $swapbyteorder = 0;
    }
    else {
        die "No BOM found.\n";
    }
    close $FH;

    {
        local $/ = $swapbyteorder ? "\x{0600}" : "\x{0006}";
        open $FH, '<:encoding(utf16)', $file or die "$!\n";

        while ( my $line = <$FH> ) {
            chomp $line;
            byteswap2($line) if $swapbyteorder;
            $line = decode( 'utf16', $line );
            # do whatever with line, now in utf-8
        }
    }
}
[download]