You could use Encode and its friends to change the encoding of your data before writing.

Or you could use the functionality of open that allows you to specify the encoding of your data, as given in the example code open(FH, "<:utf8", "file") (which is for reading - writig will work analogously). I havent used that yet, but it sounds like the right solution.

I don't know about codepages though, I'd suggest you read the Encode documentation on it.

Cheers, Sören

use utf8;

# $string being your string with wicked chars
utf8::decode($string);
[download]

hope this helps...

use strict;
use diagnostics;
use Encode 'from_to';

my @filenames    = map {
    from_to(
        $_,
        'cp1253' => 'utf8',
    );
    $_;
} glob '*';

{
    open my $filelisting, '>', 'filelisting' or die "could not open <f
+ilelisting> for writing: $!";
        binmode $filelisting, ':utf8';
        print $filelisting join "\n", @filenames;
    close $filelisting;
};
[download]

Problems with character encodings for non-ASCII text are difficult to describe clearly. If you know what specific encoding(s) you are dealing with (or would like to use), you need to be clear about that -- name them. If you can't figure out the explicit name(s), then it is certainly helpful to say what language you are dealing with (thanks for doing that).

When characters are simply not coherent, it helps to have an explicit numeric rendering of the byte sequence (e.g. using hex digits for each byte value). The pack function is handy for this, though it can be a little tough to figure out sometimes... Another way is:

my $hexstring = join " ", map { sprintf("0x%04x",ord()) } split //, $f
+oobar_string;
[download]

That will print every "character" of a string as a four-digit hex number (with leading zeros, e.g. 0x0041 for "A"). If the original string is being treated by perl as single-byte "characters", the hex numbers will all start with "0x00"; if the string is flagged as containing utf8 characters, the hex numbers will be the 16-bit values representing unicode "codepoints". Using a technique like this to know what is in your data (whether file contents or file names) is usually very helpful.