Read up on Win32API::File if you want to deal with unicode filenames.

Thanks, I have inserted some sample code

${^WIDE_SYSTEM_CALLS};

#use Encode;
#use utf8;

foreach $FILE ( `dir /B c:\\work\\intl-testout` )
#foreach $FILE ( `dir /B c:\\sigtools` )
{
    chomp($FILE);
    $FULLPATH = "c:/work/intl-testout/" . $FILE;
    
    #my $DECODE = decode("utf8",$FULLPATH);    
    
    #if ( -e $DECODE )
    #if ( -e $FULLPATH )
    if ( -e $FULLPATH )
    {
        print "DOS Dir listing File exists - $FULLPATH\n";
    }
    else
    {
        print "DOS Dir listing File DOES NOT exist - $FULLPATH\n";
    }
}
 
if ( ( opendir (HDIR,"c:/work/intl-testout" ) ) )
{
    while ( $FILE = readdir(HDIR) )
    {
        $FULLPATH = "c:/work/intl-testout/" . $FILE;
  
        if ( -e $FULLPATH )
        {
            print "Perl Dir listing File exists - $FULLPATH\n";
        }
        else
        {
            print "Perl Dir listing File DOES NOT exist - $FULLPATH\n"
+;
        }
    }
}

$FILE = q|c:/work/intl-testout/resumé.txt|;

print "$FILE\n";
if ( -e $FILE )
{
    print "File exists\n";
}
else
{
    print "Can not find File\n";
}
[download]

The above code lists a directory with some files with wide characters in the file name. I test if the file exists with the "-e" test, which should always return true. When I test the files with the first loop using the dos file listing, none of the files are found with -e. When I test the files with the Perl readdir, some files are found with -e, while others are not. The final test works as well. I suspect the characters returned from DOS listing are being translated to another character set in Perl, and then Windows can not resolve the file name when used in the "-e" test. I not very familiar with unicode, or Perl for that matter.

I couldn't find any information on file globbing in Win32API::File. Nor can I find anything to explain what is meant by "magic of NTFS." Can someone explain that? Thanks.

check Perl on Windows: file names with accented characters, UTF-8 and -e and in particular Unicode issues in Perl which looks into this exact thing