/**
* Prepares the given HTTP-query-string for the HTTP-request.
*
* HTTP-query-strings always should be utf8-encoded and urlencoded afterwards.
* So "/path/file?test=tatütata" will be converted to "/path/file?test=tat%C3%BCtata":
*
* @param stirng The quetry-string (like "/path/file?test=tatütata")
* @return string
*/
protected function prepareHTTPRequestQuery($query)
{
// If string already is a valid URL -> do nothing
if (PHPCrawlerUtils::isValidUrlString($query)) {
return $query;
}
// Decode query-string (for URLs that are partly urlencoded and partly not)
$query = rawurldecode($query);
// if query is already utf-8 encoded -> simply urlencode it,
// otherwise encode it to utf8 first.
if (PHPCrawlerUtils::isUTF8String($query) == true) {
$query = rawurlencode($query);
} else {
$query = rawurlencode(utf8_encode($query));
}
// Replace url-specific signs back
$query = str_replace("%2F", "/", $query);
$query = str_replace("%3F", "?", $query);
$query = str_replace("%3D", "=", $query);
$query = str_replace("%26", "&", $query);
return $query;
}