Description:
------------
In php 5.3.x when using fgetcsv to read a unicode file including a UTF-8 Byte 
Order Mark (BOM) prefix 0xEF,0xBB,0xBF the first row of the file is not read 
correctly.  If the BOM is removed fgetcsv reads the file correctly. 

I have tried this with and without setlocale and the result is always wrong.  I 
have run the same program on PHP 5.2.4 and it works.

Test File is the simplest possible csv with the BOM prefix "a" followed by a 
newline contains (7 characters in total)

0xEF,0xBB,0xBF,0x22,0x61,0x22,0x0A

When processed by fgetcsv the doublequotes should get removed and the value a 
should be in the array returned.  



Test script:
---------------
<?php

echo mb_detect_encoding(file_get_contents($argv[1]))."\n";

setlocale(LC_CTYPE, 'en_GB.utf8');

$handle = fopen($argv[1], "r");
$data = fgetcsv($handle, 1000, ",");
print_r($data);
?>


Expected result:
----------------
UTF-8
Array
(
    [0] => a
)


Actual result:
--------------
UTF-8
Array
(
    [0] => "a"
)