Description:
------------
fgetcsv() outputs different results with and without Unicode on Windows and Linux.

Reproduce code:
---------------
--TEST--
Test fgetcsv()
--FILE--
<?php

$filename = dirname(__FILE__) . '/fgetcsv.tmp';


// create the file and add the content with has csv fields
$file_handle = fopen($filename, "w+t" );

$delimiter = '-';
$enclosure = '-';
$csv_field = '-water---fruit---air-';


@fwrite($file_handle, $csv_field );

// rewind the file pointer to bof
rewind($file_handle);

// use the right delimiter and enclosure with max length
var_dump( fgetcsv($file_handle, 1024, $delimiter, $enclosure) );

// close the file
fclose($file_handle);
//delete file
unlink($filename);

echo "Done\n";
?>
--EXPECT--

Expected result:
----------------
array(2) {
  [0]=>
  string(11) "water-fruit"
  [1]=>
  string(3) "air"
}
Done

Actual result:
--------------
array(1) {
  [0]=>
  unicode(12) "wat^@^@^@fruitt"
}
Done