PHP - RecursiveDirectoryIterator + RecursiveIteratorIterator + RegexIterator的混合字符集文件名（拉丁语，日语，韩语）错误

I'm reading my music directory to populate a JSON for jPlayer, as follow:

<?php
//tried utf-8, shift_jis, etc. No difference
header('Content-Type: application/json; charset=SHIFT_JIS');

//cant be blank so i put . to make current file dir as base
$Directory = new RecursiveDirectoryIterator('.');
$Iterator = new RecursiveIteratorIterator($Directory);
$Regex = new RegexIterator($Iterator, '/^.+\.mp3$/i', RecursiveRegexIterator::GET_MATCH);
//instead of glob(*/*.mp3) because isnt recursive

$filesJson = [];

foreach ($Regex as $key => $value) {
    $whatever = str_ireplace(['.mp3','.\\'], '', $key);
    $filesJson['mp3'][] = [
        'title' => htmlspecialchars($whatever),
        'mp3' => $key
    ];

}
echo json_encode($filesJson);
exit();
?>

The problem lies in files which filename isn't standard UTF-8 - as Latin, Japanese and Korean ones. Examples:

Japanese

Korean

Latin (pt-br)

Which converts into ?, or simply becomes null when parsing latin names ( Geração or 4º for e.g.)

So, how make the filenames/paths be parsed correctly with different kinds of languages? The header charset isn't helping.

Info:

XAMPP with Apache2 + PHP 5.4.2 at Win7 x86

Update #1:

Tried @infinity's answer but no changes. Still ? on JP, null on Latin.

<?php
header('Content-Type: application/json; charset=UTF-8');
mb_internal_encoding('UTF-8');

$Directory = new RecursiveDirectoryIterator('.');
$Iterator = new RecursiveIteratorIterator($Directory);
$Regex = new RegexIterator($Iterator, '/^.+\.mp3$/i', RecursiveRegexIterator::GET_MATCH);

$filesJson = [];

foreach ($Regex as $key => $value) {
    $whatever = mb_substr($key, 2, mb_strlen($key)-6, "utf-8"); // 2 to remove .\ and -6 to remove .mp3 (-4 + -2)
    $filesJson['mp3'][] = [
        'title' => $whatever, //tried with and without htmlspecialchars
        'mp3' => $key
    ];

}
echo json_encode($filesJson);
exit();
?>

If I use HTML-ENTITIES instead of utf-8 on mb_substr(), latin characters works but asian still ?.

<?php
header('Content-Type: application/json; charset=utf-8');
mb_internal_encoding('utf-8');

foreach ($Regex as $key => $value) {
    $whatever = mb_substr($key, 0, mb_strlen($str)-4, "utf-8");
    // ... rest of code
}

A short try on a recursive approach using dir():

myRecursiveScanDir($mypath);

function myRecursiveScanDir($path)
    $d = dir($path);
    while (false !== ($entry = $d->read())) {

       // Do something, ie just echo it
       echo $path."/".entry."<br/>";

       if(is_dir($path."/".entry))
           myRecursiveScanDir($path."/".entry);
    }
    $d->close();
)

getting file extension and/or basename could be a bit problematic too. You might have to debug and test how mb_substr,pathinfo and basename react to such filenames.

The operating system you're using may be important in this case:

Please reffer to this question: Why does Windows need to `utf8_decode` filenames for `file_get_contents` to work?

I think it may be relevant since the screenshots look very Microsoftish.

to match any letter/digits

\p{L}\p{N}