/**
* 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'];
}
/**
* Check
*
* @param string $directive
* @param string $uri
* @return bool
* @throws ClientException
*/
private function check($directive, $uri)
{
$uriParser = new UriParser($uri);
$uri = $uriParser->convertToFull($this->base);
if ($this->base !== $uriParser->base()) {
throw new ClientException('URI belongs to a different robots.txt');
}
if (($result = $this->checkOverride($uri)) !== false) {
return $directive === $result;
}
// Path check
return $this->checkPath($directive, $uri);
}
/**
* TxtClient constructor.
*
* @param string $baseUri
* @param string $content
* @param string|null $effectiveUri
*/
public function __construct($baseUri, $content, $effectiveUri = null)
{
mb_internal_encoding(self::ENCODING);
$baseParser = new UriParser($baseUri);
$baseUri = $baseParser->base();
$effectiveBase = $baseUri;
if ($effectiveUri !== null) {
$effectiveParser = new UriParser($effectiveUri);
$effectiveBase = $effectiveParser->base();
}
$this->handler = new RootDirectiveHandler($baseUri, $effectiveBase);
$this->parseTxt($content);
}
/**
* 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;
}
/**
* Invalidate cache
*
* @param $baseUri
* @return bool
* @throws ClientException
*/
public function invalidate($baseUri)
{
$parser = new UriParser($baseUri);
return $this->handler->invalidate($parser->base());
}
/**
* cURL request
*
* @param array $options
* @return bool
*/
private function request($options = [])
{
$curl = curl_init();
// Set default cURL options
curl_setopt_array($curl, [CURLOPT_AUTOREFERER => true, CURLOPT_CAINFO => CaBundle::getSystemCaRootBundlePath(), CURLOPT_CONNECTTIMEOUT => 30, CURLOPT_ENCODING => 'identity', CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_NONE, CURLOPT_IPRESOLVE => CURL_IPRESOLVE_WHATEVER, CURLOPT_SSL_VERIFYHOST => 2, CURLOPT_SSL_VERIFYPEER => true, CURLOPT_TIMEOUT => 120, CURLOPT_USERAGENT => self::CURL_USER_AGENT]);
// Apply custom cURL options
curl_setopt_array($curl, $options);
$this->headerParser = new Parser\HeaderParser($curl);
// Make sure these cURL options stays untouched
curl_setopt_array($curl, [CURLOPT_FAILONERROR => false, CURLOPT_FOLLOWLOCATION => true, CURLOPT_FTPSSLAUTH => CURLFTPAUTH_DEFAULT, CURLOPT_HEADER => false, CURLOPT_HEADERFUNCTION => [$this->headerParser, 'curlCallback'], CURLOPT_HTTPAUTH => CURLAUTH_BASIC, CURLOPT_MAXREDIRS => self::MAX_REDIRECTS, CURLOPT_NOBODY => false, CURLOPT_PROTOCOLS => CURLPROTO_FTP | CURLPROTO_FTPS | CURLPROTO_HTTP | CURLPROTO_HTTPS | CURLPROTO_SFTP, CURLOPT_REDIR_PROTOCOLS => CURLPROTO_FTP | CURLPROTO_FTPS | CURLPROTO_HTTP | CURLPROTO_HTTPS | CURLPROTO_SFTP, CURLOPT_RETURNTRANSFER => true, CURLOPT_URL => $this->base . self::PATH, CURLOPT_USERPWD => 'anonymous:anonymous@']);
// Execute cURL request
if (($this->rawContents = curl_exec($curl)) === false) {
// Request failed
return false;
}
$this->time = time();
$this->rawStatusCode = curl_getinfo($curl, CURLINFO_HTTP_CODE);
// also works with FTP status codes
$uriParser = new UriParser(curl_getinfo($curl, CURLINFO_EFFECTIVE_URL));
$this->effective = $uriParser->base();
curl_close($curl);
$this->rawEncoding = $this->headerParser->getCharset();
$this->rawMaxAge = $this->headerParser->getMaxAge();
return true;
}
/**
* Get the RAW data
*
* @param string $baseUri
* @return array
*/
public function debug($baseUri)
{
$parser = new UriParser($baseUri);
return $this->switch->delayManager()->debug($parser->base());
}
/**
* 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;
}
