/**
* 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, $uri);
}
public function getResourceFromUri($uri)
{
EasyRdf_Format::unregister('json');
// prevent parsing errors for sources which return invalid JSON
// using apc cache for the resource if available
if (function_exists('apc_store') && function_exists('apc_fetch')) {
$key = 'fetch: ' . EasyRdf_Utils::removeFragmentFromUri($uri);
$resource = apc_fetch($key);
if ($resource === null || $resource === false) {
// was not found in cache, or previous request failed
$resource = $this->fetchResourceFromUri($uri);
apc_store($key, $resource, $this->URI_FETCH_TTL);
}
} else {
// APC not available, parse on every request
$resource = $this->fetchResourceFromUri($uri);
}
return $resource;
}
public function testRemoveFragmentExtraHash()
{
$this->assertSame('http://example.com/', EasyRdf_Utils::removeFragmentFromUri('http://example.com/#foo#bar'));
}
