/**
* Load RDF data into the graph from a URI.
*
* If no URI is given, then the URI of the graph will be used.
*
* The document type is optional but should be specified if it
* can't be guessed or got from the HTTP headers.
*
* @param string $uri The URI of the data to load
* @param string $format Optional format of the data (eg. rdfxml)
* @return integer The number of triples added to the graph
*/
public function load($uri = null, $format = null)
{
$this->checkResourceParam($uri, true);
if (!$uri) {
throw new EasyRdf_Exception("No URI given to load() and the graph does not have a URI.");
}
// Setup the HTTP client
$client = EasyRdf_Http::getDefaultHttpClient();
$client->resetParameters(true);
$client->setConfig(array('maxredirects' => 0));
$client->setMethod('GET');
$client->setHeaders('Accept', EasyRdf_Format::getHttpAcceptHeader());
$requestUrl = $uri;
$response = null;
$redirectCounter = 0;
do {
// Have we already loaded it into the graph?
$requestUrl = EasyRdf_Utils::removeFragmentFromUri($requestUrl);
if (in_array($requestUrl, $this->loaded)) {
return 0;
}
// Make the HTTP request
$client->setHeaders('host', null);
$client->setUri($requestUrl);
$response = $client->request();
// Add the URL to the list of URLs loaded
$this->loaded[] = $requestUrl;
if ($response->isRedirect() and $location = $response->getHeader('location')) {
// Avoid problems with buggy servers that add whitespace
$location = trim($location);
// Some servers return relative URLs in the location header
// resolve it in relation to previous request
$baseUri = new EasyRdf_ParsedUri($requestUrl);
$requestUrl = $baseUri->resolve($location)->toString();
$requestUrl = EasyRdf_Utils::removeFragmentFromUri($requestUrl);
// If it is a 303 then drop the parameters
if ($response->getStatus() == 303) {
$client->resetParameters();
}
++$redirectCounter;
} elseif ($response->isSuccessful()) {
// If we didn't get any location, stop redirecting
break;
} else {
throw new EasyRdf_Http_Exception("HTTP request for {$requestUrl} failed: " . $response->getMessage(), $response->getStatus(), null, $response->getBody());
}
} while ($redirectCounter < $this->maxRedirects);
if (!$format or $format == 'guess') {
list($format, $params) = EasyRdf_Utils::parseMimeType($response->getHeader('Content-Type'));
}
// Parse the data
return $this->parse($response->getBody(), $format, $uri);
}
public function testResolve()
{
$base = new EasyRdf_ParsedUri('http://example.com');
$this->assertStringEquals("http://example.com/filename", $base->resolve("filename"));
}
/**
* Send the HTTP request and return an HTTP response object
*
* @return EasyRdf_Http_Response
* @throws EasyRdf_Exception
*/
public function request($method = null)
{
if (!$this->uri) {
throw new EasyRdf_Exception("Set URI before calling EasyRdf_Http_Client->request()");
}
if ($method) {
$this->setMethod($method);
}
$this->redirectCounter = 0;
$response = null;
// Send the first request. If redirected, continue.
do {
// Clone the URI and add the additional GET parameters to it
$uri = parse_url($this->uri);
if ($uri['scheme'] === 'http') {
$host = $uri['host'];
} elseif ($uri['scheme'] === 'https') {
$host = 'ssl://' . $uri['host'];
} else {
throw new EasyRdf_Exception("Unsupported URI scheme: " . $uri['scheme']);
}
if (isset($uri['port'])) {
$port = $uri['port'];
} else {
if ($uri['scheme'] === 'https') {
$port = 443;
} else {
$port = 80;
}
}
if (!empty($this->paramsGet)) {
if (!empty($uri['query'])) {
$uri['query'] .= '&';
} else {
$uri['query'] = '';
}
$uri['query'] .= http_build_query($this->paramsGet, null, '&');
}
$headers = $this->prepareHeaders($uri['host'], $port);
// Open socket to remote server
$socket = @fsockopen($host, $port, $errno, $errstr, $this->config['timeout']);
if (!$socket) {
throw new EasyRdf_Exception("Unable to connect to {$host}:{$port} ({$errstr})");
}
// Write the request
$path = $uri['path'];
if (empty($path)) {
$path = '/';
}
if (isset($uri['query'])) {
$path .= '?' . $uri['query'];
}
fwrite($socket, "{$this->method} {$path} HTTP/1.1\r\n");
foreach ($headers as $k => $v) {
if (is_string($k)) {
$v = ucfirst($k) . ": {$v}";
}
fwrite($socket, "{$v}\r\n");
}
fwrite($socket, "\r\n");
// Send the request body, if there is one set
if (isset($this->rawPostData)) {
fwrite($socket, $this->rawPostData);
}
// Read in the response
$content = '';
while (!feof($socket)) {
$content .= fgets($socket);
}
// FIXME: support HTTP/1.1 100 Continue
// Close the socket
@fclose($socket);
// Parse the response string
$response = EasyRdf_Http_Response::fromString($content);
// If we got redirected, look for the Location header
if ($response->isRedirect() && ($location = $response->getHeader('location'))) {
// Avoid problems with buggy servers that add whitespace at the
// end of some headers (See ZF-11283)
$location = trim($location);
// Some servers return relative URLs in the location header
// resolve it in relation to previous request
$baseUri = new EasyRdf_ParsedUri($this->uri);
$location = $baseUri->resolve($location)->toString();
// If it is a 303 then drop the parameters and send a GET request
if ($response->getStatus() == 303) {
$this->resetParameters();
$this->setMethod('GET');
}
// If we got a well formed absolute URI
if (parse_url($location)) {
$this->setHeaders('host', null);
$this->setUri($location);
} else {
throw new EasyRdf_Exception("Failed to parse Location header returned by " . $this->uri);
}
++$this->redirectCounter;
} else {
// If we didn't get any location, stop redirecting
break;
}
} while ($this->redirectCounter < $this->config['maxredirects']);
return $response;
}
