/**
* Add
*
* @param string $line
* @return bool
*/
public function add($line)
{
$uriParser = new UriParser($line);
$uri = $uriParser->encode();
if (!$uriParser->validate() || in_array($uri, $this->sitemaps)) {
return false;
}
$this->sitemaps[] = $uri;
return true;
}
/**
* Parse
*
* @param string $line
* @return string|false
*/
private function parse($line)
{
$uriParser = new UriParser($line);
$line = $uriParser->encode();
if ($uriParser->validateIP() || !$uriParser->validateHost() || parse_url($line, PHP_URL_SCHEME) !== null && !$uriParser->validateScheme()) {
return false;
}
$parts = $this->getParts($line);
return $parts['scheme'] . $parts['host'] . $parts['port'];
}
/**
* Is listed?
*
* @param string $uri
* @return bool
*/
public function isListed($uri)
{
$uriParser = new UriParser($uri);
$uri = $uriParser->encode();
$parts = ['scheme' => parse_url($uri, PHP_URL_SCHEME), 'host' => parse_url($uri, PHP_URL_HOST)];
$parts['port'] = is_int($port = parse_url($uri, PHP_URL_PORT)) ? $port : getservbyname($parts['scheme'], 'tcp');
$cases = [$parts['host'], $parts['host'] . ':' . $parts['port'], $parts['scheme'] . '://' . $parts['host'], $parts['scheme'] . '://' . $parts['host'] . ':' . $parts['port']];
foreach ($this->host as $host) {
if (in_array($host, $cases)) {
return true;
}
}
return false;
}
/**
* Add
*
* @param string $line
* @return bool
*/
public function add($line)
{
// split into parameter and path
$array = array_map('trim', mb_split('\\s+', $line, 2));
if (isset($array[1])) {
// strip any invalid characters from path prefix
$uriParser = new UriParser(preg_replace('/[^A-Za-z0-9\\.-\\/\\*\\_]/', '', $array[1]));
$path = rtrim($uriParser->encode(), '*');
}
$path = empty($path) ? '/' : $path;
$param = array_map('trim', explode('&', $array[0]));
foreach ($param as $key) {
$this->cleanParam[$key][] = $path;
}
return true;
}
/**
* Get path and query
*
* @param string $uri
* @return string
* @throws ClientException
*/
private function getPath($uri)
{
$uriParser = new UriParser($uri);
// Prepare uri
$uriParser->encode();
$uri = $uriParser->stripFragment();
if (mb_strpos($uri, '/') === 0) {
// URI is already an path
return $uri;
}
if (!$uriParser->validate()) {
throw new ClientException('Invalid URI');
}
$path = ($path = parse_url($uri, PHP_URL_PATH)) === null ? '/' : $path;
$query = ($query = parse_url($uri, PHP_URL_QUERY)) === null ? '' : '?' . $query;
return $path . $query;
}
