public function readData($source_definition, $rest_parameters = array())
{
$uri = $source_definition['uri'];
// Keep track of the prefix URI's
$this->prefixes = array();
// Check for caching
if (Cache::has($uri)) {
$data = Cache::get($uri);
} else {
// Fetch the data
$data = @file_get_contents($uri);
if (!empty($data)) {
Cache::put($uri, $data, $source_definition['cache']);
} else {
$uri = $source_definition['uri'];
\App::abort(500, "Cannot retrieve data from the XML file located on {$uri}.");
}
}
$data_result = new Data();
$data_result->data = $data;
$data_result->semantic = $this->prefixes;
$data_result->preferred_formats = $this->getPreferredFormats();
if (!empty($source_definition['geo_formatted']) && $source_definition['geo_formatted']) {
$data_result->geo_formatted = true;
$data_result->preferred_formats = array('geojson', 'map', 'php');
}
return $data_result;
}
public function readData($source_definition, $rest_parameters = array())
{
$uri = $source_definition['uri'];
// Check for caching
if (Cache::has($uri)) {
$data = Cache::get($uri);
} else {
// Fetch the data
$data = [];
if (!filter_var($uri, FILTER_VALIDATE_URL) === false) {
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $uri);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
$data = curl_exec($ch);
curl_close($ch);
} else {
$data = @file_get_contents($uri);
}
if ($data) {
Cache::put($uri, $data, $source_definition['cache']);
} else {
$uri = $source_definition['uri'];
\App::abort(500, "Cannot retrieve data from the JSON file located on {$uri}.");
}
}
$php_object = json_decode($data);
$data_result = new Data();
$data_result->data = $php_object;
$data_result->preferred_formats = $this->getPreferredFormats();
return $data_result;
}
/**
* Add triples to the graph and return it based on limit, offset and the SPARQL query
*
* @param string $base_uri
* @param EasyRdf_Graph $graph
* @param int $limit
* @param int $offset
* @param integer $depth The depth the queries should have, handlers should not override this if given
*
* @return EasyRdf_Graph
*/
public function addTriples($base_uri, $graph, $limit, $offset, $depth = null)
{
$total_triples = $graph->countTriples();
// Iterate the sparql endpoints
foreach ($this->sparql_repo->getAll() as $sparql_source) {
$endpoint = $sparql_source['endpoint'];
$pw = $sparql_source['endpoint_password'];
$user = $sparql_source['endpoint_user'];
$endpoint = rtrim($endpoint, '/');
if (is_null($depth)) {
$depth = $sparql_source['depth'];
}
$count_query = $this->query_builder->createCountQuery($base_uri, \Request::root(), $sparql_source['named_graph'], $depth);
\Log::info("sparql : " . $count_query . " base uri : " . $base_uri);
// Check if the response has been cached yet
$cache_string = $this->buildCacheString($sparql_source['id'], $count_query);
if (Cache::has($cache_string)) {
$result = Cache::get($cache_string);
} else {
$count_query = urlencode($count_query);
$count_query = str_replace("+", "%20", $count_query);
$query_uri = $endpoint . '?query=' . $count_query . '&format=' . urlencode("application/sparql-results+json");
// Make a request with the count query to the SPARQL endpoint
$result = $this->executeUri($query_uri, array(), $user, $pw);
Cache::put($cache_string, $result, 5);
}
$response = json_decode($result);
if (!empty($response)) {
$count = $response->results->bindings[0]->count->value;
// If the amount of matching triples is higher than the offset
// add them and update the offset, if not higher, then only update the offset
if ($count > $offset) {
// Read the triples from the sparql endpoint
$query_limit = $limit - $total_triples;
$query = $this->query_builder->createFetchQuery($base_uri, \Request::root(), $sparql_source['named_graph'], $query_limit, $offset, $depth);
$query = urlencode($query);
$query = str_replace("+", "%20", $query);
$query_uri = $endpoint . '?query=' . $query . '&format=' . urlencode("application/rdf+xml");
// Check for caching
$cache_string = $this->buildCacheString($sparql_source['id'], $query_uri);
if (Cache::has($cache_string)) {
$result = Cache::get($cache_string);
} else {
$result = $this->executeUri($query_uri, array(), $user, $pw);
}
if (!empty($result) && $result[0] == '<') {
// Parse the triple response and retrieve the triples from them
$result_graph = new \EasyRdf_Graph();
$parser = new \EasyRdf_Parser_RdfXml();
$parser->parse($result_graph, $result, 'rdfxml', null);
$graph = $this->mergeGraph($graph, $result_graph);
$total_triples += $count - $offset;
} else {
$sparql_id = $sparql_source['id'];
\Log::error("Something went wrong while fetching the triples from the sparql source with id {$sparql_id}. The error was " . $result . ". The query was : " . $query_uri);
}
} else {
// Update the offset
$offset -= $count;
}
if ($offset < 0) {
$offset = 0;
}
}
}
return $graph;
}
/**
* Retrieve a Data object identified by $uri
*
* @param string $uri The identifier that identifies a resource
*
* @return \Response
*/
public function get($uri)
{
// Check permissions
Auth::requirePermissions('dataset.view');
// Split for an (optional) extension
list($uri, $extension) = $this->processURI($uri);
// Check for caching
// Based on: URI / Rest parameters / Query parameters / Paging headers
$cache_string = $uri;
list($limit, $offset) = Pager::calculateLimitAndOffset();
$cache_string .= '/limit=' . $limit . 'offset=' . $offset;
$cache_string .= http_build_query(\Input::except('limit', 'offset', 'page', 'page_size'));
$cache_string = sha1($cache_string);
if (Cache::has($cache_string)) {
return ContentNegotiator::getResponse(Cache::get($cache_string), $extension);
} else {
// Get definition
$definition = $this->definition->getByIdentifier($uri);
if ($definition) {
// Get source definition
$source_definition = $this->definition->getDefinitionSource($definition['source_id'], $definition['source_type']);
if ($source_definition) {
$source_type = $source_definition['type'];
// Create the right datacontroller
$controller_class = 'Tdt\\Core\\DataControllers\\' . $source_type . 'Controller';
$data_controller = \App::make($controller_class);
// Get REST parameters
$rest_parameters = str_replace($definition['collection_uri'] . '/' . $definition['resource_name'], '', $uri);
$rest_parameters = ltrim($rest_parameters, '/');
$rest_parameters = explode('/', $rest_parameters);
if (empty($rest_parameters[0]) && !is_numeric($rest_parameters[0])) {
$rest_parameters = array();
}
// Retrieve dataobject from datacontroller
$data = $data_controller->readData($source_definition, $rest_parameters);
$data->rest_parameters = $rest_parameters;
// REST filtering
if ($source_type != 'INSTALLED' && count($data->rest_parameters) > 0) {
$data->data = self::applyRestFilter($data->data, $data->rest_parameters);
}
// Add definition to the object
$data->definition = $definition;
// Add source definition to the object
$data->source_definition = $source_definition;
// Add the available, supported formats to the object
$format_helper = new FormatHelper();
$data->formats = $format_helper->getAvailableFormats($data);
// Store in cache
Cache::put($cache_string, $data, $source_definition['cache']);
// Return the formatted response with content negotiation
return ContentNegotiator::getResponse($data, $extension);
} else {
\App::abort(404, "Source for the definition could not be found.");
}
} else {
// Coulnd't find a definition, but it might be a collection
$resources = $this->definition->getByCollection($uri);
if (count($resources) > 0) {
$data = new Data();
$data->data = new \stdClass();
$data->data->datasets = array();
$data->data->collections = array();
foreach ($resources as $res) {
// Check if it's a subcollection or a dataset
$collection_uri = rtrim($res['collection_uri'], '/');
if ($collection_uri == $uri) {
array_push($data->data->datasets, \URL::to($collection_uri . '/' . $res['resource_name']));
} else {
// Push the subcollection if it's not already in the array
if (!in_array(\URL::to($collection_uri), $data->data->collections)) {
array_push($data->data->collections, \URL::to($collection_uri));
}
}
}
// Fake a definition
$data->definition = new \Definition();
$uri_array = explode('/', $uri);
$last_chunk = array_pop($uri_array);
$data->definition->collection_uri = join('/', $uri_array);
$data->definition->resource_name = $last_chunk;
// Return the formatted response with content negotiation
return ContentNegotiator::getResponse($data, $extension);
} else {
\App::abort(404, "The dataset or collection you were looking for could not be found (URI: {$uri}).");
}
}
}
}
/**
* Retrieve a Data object identified by $uri
*
* @param string $uri The identifier that identifies a resource
*
* @return \Response
*/
public function get($uri)
{
// Check permissions
Auth::requirePermissions('dataset.view');
// Split for an (optional) extension
list($uri, $extension) = $this->processURI($uri);
// Check for caching
// Based on: URI / Rest parameters / Query parameters / Paging headers
$cache_string = $uri;
list($limit, $offset) = Pager::calculateLimitAndOffset();
$cache_string .= '/limit=' . $limit . 'offset=' . $offset;
$omit = ['limit', 'offset', 'page', 'page_size'];
$query_string_params = \Input::get();
foreach ($query_string_params as $key => $val) {
if (in_array($key, $omit)) {
unset($query_string_params[$key]);
}
}
$cache_string .= http_build_query($query_string_params);
$cache_string = sha1($cache_string);
if (Cache::has($cache_string)) {
return ContentNegotiator::getResponse(Cache::get($cache_string), $extension);
} else {
// Get definition
$definition = $this->definition->getByIdentifier($uri);
if ($definition) {
// Get source definition
$source_definition = $this->definition->getDefinitionSource($definition['source_id'], $definition['source_type']);
if ($source_definition) {
$source_type = $source_definition['type'];
// Create the right datacontroller
$controller_class = 'Tdt\\Core\\DataControllers\\' . $source_type . 'Controller';
$data_controller = \App::make($controller_class);
// Get REST parameters
$uri_segments = explode('/', $uri);
$rest_parameters = array_diff($uri_segments, array($definition['collection_uri'], $definition['resource_name']));
$rest_parameters = array_values($rest_parameters);
$throttle_response = $this->applyThrottle($definition);
if (!empty($throttle_response)) {
return $throttle_response;
}
// Retrieve dataobject from datacontroller
$data = $data_controller->readData($source_definition, $rest_parameters);
// If the source type is XML, just return the XML contents, don't transform
if (strtolower($source_type) == 'xml' && $extension == 'xml') {
return $this->createXMLResponse($data->data);
}
$data->rest_parameters = $rest_parameters;
// REST filtering
if ($source_type != 'INSTALLED' && count($data->rest_parameters) > 0) {
$data->data = self::applyRestFilter($data->data, $data->rest_parameters);
}
// Semantic paging with the hydra voc
if ($data->is_semantic && !empty($data->paging)) {
\EasyRdf_Namespace::set('hydra', 'http://www.w3.org/ns/hydra/core#');
$graph = $data->data;
$url = \URL::to($definition['collection_uri'] . '/' . $definition['resource_name']);
$request_url = \Request::url();
$graph->addResource($request_url, 'void:subset', $url);
foreach ($data->paging as $key => $val) {
$paged_url = $request_url . '?offset=' . $val[0] . '&limit=' . $val[1] . Pager::buildQuerystring();
switch ($key) {
case 'next':
$graph->addResource($request_url, 'hydra:nextPage', $paged_url);
break;
case 'previous':
$graph->addResource($request_url, 'hydra:previousPage', $paged_url);
break;
case 'last':
$graph->addResource($request_url, 'hydra:lastPage', $paged_url);
break;
case 'first':
$graph->addResource($request_url, 'hydra:firstPage', $paged_url);
break;
}
}
$graph->addResource($url, 'a', 'dcat:Dataset');
$title = null;
if (!empty($definition['title'])) {
$title = $definition['title'];
} else {
$title = $definition['collection_uri'] . '/' . $definition['resource_name'];
}
$graph->addLiteral($url, 'dc:title', $title);
$graph->addLiteral($url, 'dc:description', $source_definition['description']);
$graph->addResource($url, 'dcat:distribution', $url . '.json');
$data->data = $graph;
}
// Add definition to the object
$data->definition = $definition;
// Add source definition to the object
$data->source_definition = $source_definition;
// Add the available, supported formats to the object
$format_helper = new FormatHelper();
$data->formats = $format_helper->getAvailableFormats($data);
// Store in cache
Cache::put($cache_string, $data, $source_definition['cache']);
// Return the formatted response with content negotiation
return ContentNegotiator::getResponse($data, $extension);
} else {
\App::abort(404, "Source for the definition could not be found.");
}
} else {
// Coulnd't find a definition, but it might be a collection
$resources = $this->definition->getByCollection($uri);
if (count($resources) > 0) {
$data = new Data();
$data->data = new \stdClass();
$data->data->datasets = array();
$data->data->collections = array();
foreach ($resources as $res) {
// Check if it's a subcollection or a dataset
$collection_uri = rtrim($res['collection_uri'], '/');
if ($collection_uri == $uri) {
array_push($data->data->datasets, \URL::to($collection_uri . '/' . $res['resource_name']));
} else {
// Push the subcollection if it's not already in the array
if (!in_array(\URL::to($collection_uri), $data->data->collections)) {
array_push($data->data->collections, \URL::to($collection_uri));
}
}
}
// Fake a definition
$data->definition = new \Definition();
$uri_array = explode('/', $uri);
$last_chunk = array_pop($uri_array);
$data->definition->collection_uri = join('/', $uri_array);
$data->definition->resource_name = $last_chunk;
// Return the formatted response with content negotiation
return ContentNegotiator::getResponse($data, $extension);
} else {
\App::abort(404, "The dataset or collection you were looking for could not be found (URI: {$uri}).");
}
}
}
}
/**
* Resolve a graph pattern query (/all route)
*
* @param string $format The format of the request
*
* @return \Response
*/
public function solveQuery($format = null)
{
$data;
if (!empty($format)) {
$format = ltrim($format, '.');
}
// Ignore the rest of the uri after /all and work with the request parameters as they were given
$cache_string = sha1($this->getRawRequestURI(\Request::root()));
// Check if the response to the query has been cached already
if (Cache::has($cache_string)) {
$data = Cache::get($cache_string);
} else {
// Get the graph pattern query string parameters from the request
list($s, $p, $o) = $this->getTemplateParameters();
// Pass them to our sparql query builder
SparqlQueryBuilder::setParameters(array($s, $p, $o));
$base_uri = null;
// If no parameter has been filled in, the URI we have to match triples with is the root of our application
if ($s == '?s' && $p == '?p' && $o == '?o') {
$base_uri = \Request::root();
}
// Fetch matching triples
$result = $this->triples->getTriples($base_uri, \Request::get('limit', 100), \Request::get('offset', 0));
// If the graph contains no triples, then the graph pattern couldn't resolve to anything, 404 it is
if ($result->countTriples() == 0) {
\App::abort(404, "The resource couldn't be found, nor dereferenced.");
}
$definition = array('resource_name' => "all", 'collection_uri' => "");
$source_definition = array('description' => 'Semantic data collected out the configured semantic data sources.', 'type' => 'Semantic');
$data = new Data();
$data->definition = $definition;
$data->source_definition = $source_definition;
$data->data = $result;
$data->is_semantic = true;
// Add the available, supported formats to the object
$format_helper = new FormatHelper();
$data->formats = $format_helper->getAvailableFormats($data);
// Store in cache for a default of 5 minutes
Cache::put($cache_string, $data, 5);
}
// Add the hydra namespace, it's not present in the easy rdf namespaces by default
\EasyRdf_Namespace::set('hydra', 'http://www.w3.org/ns/hydra/core#');
// Return the formatted response with content negotiation
$response = ContentNegotiator::getResponse($data, $format);
// Pass a Vary header so that browsers know they have to take the accept header
// into consideration when they apply caching client side
$response->header('Vary', 'Accept');
// Allow CORS
$response->header('Access-Control-Allow-Origin', '*');
return $response;
}
private function getPlainJson($uri)
{
$data = [];
if (Cache::has($uri)) {
return Cache::get($uri);
}
if (!filter_var($uri, FILTER_VALIDATE_URL) === false) {
$parts = parse_url($uri);
if ($parts['scheme'] != 'file') {
$data = $this->getRemoteData($uri);
} else {
$data = @file_get_contents($uri);
}
} else {
$data = @file_get_contents($uri);
}
if ($data) {
Cache::put($uri, $data, $this->cache);
} else {
\App::abort(500, "Cannot retrieve data from the JSON file located on {$uri}.");
}
return $data;
}
/**
* Add triples to the graph and return it based on limit, offset and the SPARQL query
*
* @param string $base_uri
* @param EasyRdf_Graph $graph
* @param int $limit
* @param int $offset
*
* @return EasyRdf_Graph
*/
public function addTriples($base_uri, $graph, $limit, $offset)
{
$total_triples = $graph->countTriples();
// Iterate the LDF end points, note that ldf servers don't necessarily have page size's as a parameter
// But rather have a fixed page size
foreach ($this->ldf_repo->getAll() as $ldf_conf) {
// Build the query string (raw)
$query_string = $_SERVER['QUERY_STRING'];
$q_string_raw = '';
$query_parts = explode('&', $query_string);
// Don't let paging parameters in the re-constructed query string
$invalid_q_string = array('page', 'page_size', 'limit', 'offset');
foreach ($query_parts as $part) {
if (!empty($part)) {
$couple = explode('=', $part);
if (!in_array($couple[0], $invalid_q_string)) {
$q_string_raw .= $couple[0] . '=' . $couple[1] . '&';
}
}
}
$q_string_raw = rtrim($q_string_raw, '&');
$start_fragment = $ldf_conf['startfragment'];
$entire_fragment = $start_fragment . '?' . $q_string_raw;
$entire_fragment = rtrim($entire_fragment, '?');
// Make the LDF query (basic GET to the endpoint, should provide us with a hydra:totalItems or void:triples entry)
$accept = array("Accept: text/turtle,*/*;q=0.0");
$response = '';
if (Cache::has($entire_fragment)) {
$response = Cache::get($entire_fragment);
} else {
$response = $this->executeUri($entire_fragment, $accept);
}
if ($response) {
// Try decoding it into turtle, if not something is wrong with the response body
try {
$tmp_graph = new \EasyRdf_Graph();
$parser = new \EasyRdf_Parser_Turtle();
\EasyRdf_Namespace::set('hydra', 'http://www.w3.org/ns/hydra/core#');
$parser->parse($tmp_graph, $response, 'turtle', null);
// Fetch the count (hydra:totalItems or void:triples)
$count = $tmp_graph->getLiteral($entire_fragment, 'hydra:totalItems');
$page_size = $tmp_graph->getLiteral($entire_fragment, 'hydra:itemsPerPage');
if (is_null($count)) {
$count = $tmp_graph->getLiteral($entire_fragment, 'void:triples');
}
if (is_null($count) || is_null($page_size)) {
// Skip, the count has not been found on this endpoint
$count = -1;
\Log::warning("An LDF endpoint's count could not be retrieved from the uri: {$entire_fragment}");
} else {
$count = $count->getValue();
$page_size = $page_size->getValue();
Cache::put($entire_fragment, $response, 5);
}
// If the amount of matching triples is higher than the offset
// add them and update the offset, if not higher, then only update the offset
if ($count > $offset) {
// Read the triples from the LDF
$query_limit = $limit - $total_triples;
// There's no way of giving along the page size (not that we can presume)
// So we have to make a numer of requests
$amount_of_requests = ceil($query_limit / $page_size);
// Calculate the page offset from the offset parameter
$page_offset = ceil($offset / $page_size);
for ($i = $page_offset; $i < $amount_of_requests + $page_offset; $i++) {
$paged_fragment = $entire_fragment;
if (!empty($q_string_raw)) {
$paged_fragment .= '&page=' . $i;
} else {
$paged_fragment .= '?page=' . $i;
}
// Ask for turtle
$accept = array('Accept: text/turtle');
$response = '';
if (Cache::has($paged_fragment)) {
$response = Cache::get($paged_fragment);
} else {
$response = $this->executeUri($paged_fragment, $accept);
}
if ($response) {
// Try decoding it into turtle, if not something is wrong with the response body
try {
$tmp_graph = new \EasyRdf_Graph();
$parser = new \EasyRdf_Parser_Turtle();
$parser->parse($tmp_graph, $response, 'turtle', $start_fragment);
// Fetch the count (hydra:totalItems or void:triples)
$total_items = $tmp_graph->getLiteral($paged_fragment, 'hydra:totalItems');
if (is_null($total_items)) {
$total_items = $tmp_graph->getLiteral($paged_fragment, 'void:triples');
}
if (!is_null($total_items)) {
Cache::put($paged_fragment, $tmp_graph, 5);
// This needs to be a function of a different helper class for LDF endpoints
$tmp_graph = $this->rebaseGraph($start_fragment, $tmp_graph);
$graph = $this->mergeGraph($graph, $tmp_graph);
$total_triples += $page_size;
}
} catch (\EasyRdf_Parser_Exception $ex) {
\Log::error("Failed to parse turtle content from the LDF endpoint: {$start_fragment}");
}
} else {
\Log::error("Something went wrong while fetching the triples from a LDF source. The error was " . $response . ". The query was : " . $paged_fragment);
}
}
} else {
// Update the offset
$offset -= $count;
}
if ($offset < 0) {
$offset = 0;
}
} catch (\EasyRdf_Parser_Exception $ex) {
\Log::error("Failed to parse turtle content from the LDF endpoint: {$endpoint}");
}
} else {
\Log::warning("Couldn't fetch a proper response for the fragment: {$entire_fragment}.");
}
}
return $graph;
}
