I used the regex from @GabrielAnderson and devised this function which behaves according to RFC 2616 (when no quality value is given to a language, it defaults to 1).
When several languages share the same quality value, the most specific are given priority over the less specific ones. (this behaviour is not part of the RFC which provides no recommendation for this specific case)
function Get_Client_Prefered_Language ($getSortedList = false, $acceptedLanguages = false)
{
if (empty($acceptedLanguages))
$acceptedLanguages = $_SERVER["HTTP_ACCEPT_LANGUAGE"];
// regex inspired from @GabrielAnderson on http://stackoverflow.com/questions/6038236/http-accept-language
preg_match_all('/([a-z]{1,8}(-[a-z]{1,8})*)\s*(;\s*q\s*=\s*(1|0\.[0-9]+))?/i', $acceptedLanguages, $lang_parse);
$langs = $lang_parse[1];
$ranks = $lang_parse[4];
// (create an associative array 'language' => 'preference')
$lang2pref = array();
for($i=0; $i<count($langs); $i++)
$lang2pref[$langs[$i]] = (float) (!empty($ranks[$i]) ? $ranks[$i] : 1);
// (comparison function for uksort)
$cmpLangs = function ($a, $b) use ($lang2pref) {
if ($lang2pref[$a] > $lang2pref[$b])
return -1;
elseif ($lang2pref[$a] < $lang2pref[$b])
return 1;
elseif (strlen($a) > strlen($b))
return -1;
elseif (strlen($a) < strlen($b))
return 1;
else
return 0;
};
// sort the languages by prefered language and by the most specific region
uksort($lang2pref, $cmpLangs);
if ($getSortedList)
return $lang2pref;
// return the first value's key
reset($lang2pref);
return key($lang2pref);
}
Example:
print_r(Get_Client_Prefered_Language(true, 'en,en-US,en-AU;q=0.8,fr;q=0.6,en-GB;q=0.4'));
Outputs:
Array
(
[en-US] => 1
[en] => 1
[en-AU] => 0.8
[fr] => 0.6
[en-GB] => 0.4
)
As you can notice, 'en-US' appears in first position despite the fact that 'en' was first in the given string.
So you could use this function and just replace your first line of code by:
$http_lang = substr(Get_Client_Prefered_Language(),0,2);
