Windows provides two interface to its functions. One supports single byte characters (Called "ANSI" by MS), and one supports two-byte characters (Called "Wide" or "UNICODE" by MS, they are encoded as "UCS-2le"). For example, to create/open a file, one would call CreateFileA if we passed a single-byte char string, or CreateFileW if we passed a UCS-2le char string.

I'm not sure how about all the details, but one thing in sure. If Perl is using CreateFileA (and I think it does), you have a problem.

>perl -e"use Encode; print encode('UCS-2le', 'ABC')" | od -c
0000000    A  \0   B  \0   C  \0
0000006
[download]

Since CreateFileA accepts a NUL terminated file name, it will think encode('UCS-2le', 'ABC') is just 'A'. CreateFileW must be used to open that file. For the same reason, FindFirstFileW and FindNextFileW must be used list the contents of the directory.

I don't know if there's anything "out there" that does what you need. Win32::API will definitely get you there, but you'll have to do some leg work.

Oh, dear! I was hoping there would be an easier and dirtier solution to my problem than Win32::API but... a monk's got to do what a monk's got to do. ;-)

I'm ready to dive. I'll let you know if this gets me anywhere.

#!/usr/bin/perl -w
use strict;
use Encode qw(from_to);
use Encode::Byte;

my $file = "C:\\Αντίγραφο
+\\VOLINFO.TXT";
from_to($file, "utf8", "cp1253");

open FILE, $file or die "can't open $file: $!";
while (<FILE>) {
    print;
}
close FILE;
[download]

Hi, I don't know if this will work on Win32, but it's worth a try -> Re: problems with extended ascii characters in filenames. It works on linux.

---

I'm not really a human, but I play one on earth. Cogito ergo sum a bum

PS: Of course, I'm not using html entities but the corresponding unicode characters.

Perhaps your filesystem doesn't encode filenames using UTF-8. I'd try opening up the directory with opendir() and looking at what readdir() produces for the files in question. Most likely you'll have to produce the same thing to open the file.

-sam

Thanks for the reply, Sam!

Well, perhaps I should have mentioned that I resorted to UTF-8 after having tried Windows-1253 and ISO-8859-7. I even created a Wx::FileDialog object to make sure I'm getting the right path/filename but to no avail.

Of course, there is no problem when plain English letters only appear in the path.

Did you try UTF-16? I have some vague recollection that it's a popular encoding for Windows filesystems. Wikipedia seems to agree: UTF-16.

-sam