/**
* @param $name string, unit name used in ontology and mapping configuration
* @param $labels list of strings, unit labels used in template property values
*/
public function __construct($name, $labels)
{
parent::__construct($name, OntologyNamespaces::getUri($name, OntologyNamespaces::DBPEDIA_ONTOLOGY_NAMESPACE));
if (!is_array($labels)) {
throw new \InvalidArgumentException('labels must be an array');
}
$this->labels = $labels;
}
public function parse(Node $node)
{
if ($node instanceof PropertyNode) {
$children = $node->getChildren('LinkNode');
foreach ($children as $child) {
if (!$child->isExternalLink()) {
$link = $child->getDestination();
if ($link) {
return OntologyNamespaces::getUri($link->encoded(), OntologyNamespaces::DBPEDIA_INSTANCE_NAMESPACE);
}
}
}
}
return null;
}
private function writeGeoQuad($node, $subjectUri, $lat, $long)
{
if ($lat != '' && $long != '') {
try {
if (isset($this->ontologyProperty)) {
// TODO: replace label with local part w/o class
$ontologyPropertyLabel = $this->ontologyProperty->getLabel();
$originalSubjectUri = $subjectUri;
$subjectUri = $this->pageContext->generateUri(OntologyNamespaces::getUri($node->getRoot()->getTitle()->encoded(), OntologyNamespaces::DBPEDIA_INSTANCE_NAMESPACE) . '__' . $ontologyPropertyLabel, null);
$quad = new RdfQuad($subjectUri, $this->ontology->getProperty("rdf:type"), $this->ontology->getClass("gml:_Feature")->getUri(), $node->getSourceUri());
$this->destination->addQuad($quad);
$quad = new RdfQuad($originalSubjectUri, $this->ontologyProperty, $subjectUri, $node->getSourceUri());
$this->destination->addQuad($quad);
}
$geoLat = $this->ontology->getProperty("geo:lat");
$geoLong = $this->ontology->getProperty("geo:long");
$georssPoint = $this->ontology->getProperty("georss:point");
$quad = new RdfQuad($subjectUri, $geoLat, $lat, $node->getSourceUri());
$this->destination->addQuad($quad);
$quad = new RdfQuad($subjectUri, $geoLong, $long, $node->getSourceUri());
$this->destination->addQuad($quad);
$quad = new RdfQuad($subjectUri, $georssPoint, $lat . " " . $long, $node->getSourceUri());
$this->destination->addQuad($quad);
return true;
} catch (\InvalidArgumentException $e) {
$this->logger->warn($e->getMessage() . ' (Page: ' . $node->getRoot()->getTitle() . ')');
}
}
}
/**
* Extracts all pages in the source directory according to the mappings
*/
private function extract()
{
$this->logger->info('Extracting...');
$extractor = $this->extractor;
$pageCount = 0;
$startTime = microtime(true);
$logger = $this->logger;
$callback = function ($page) use($extractor, &$pageCount, $startTime, $logger) {
$pageUri = OntologyNamespaces::getUri($page->getTitle()->encoded(), OntologyNamespaces::DBPEDIA_INSTANCE_NAMESPACE);
$pageContext = new mapping\PageContext($pageUri);
$extractor->extract($page, $pageUri, $pageContext);
$pageCount++;
if ($pageCount % 10 == 0) {
$timePerPage = round((microtime(true) - $startTime) * 1000.0 / $pageCount, 0);
$logger->info('Extracted pages: ' . $pageCount . ' (' . $timePerPage . ' ms per page)');
}
};
$this->pageSource->processPages($callback);
$this->logger->info('Extraction finished (' . $pageCount . ' pages)');
}
/**
* @param $name string, dimension name used in ontology and mapping configuration
*/
public function __construct($name)
{
parent::__construct($name, OntologyNamespaces::getUri($name, OntologyNamespaces::DBPEDIA_ONTOLOGY_NAMESPACE));
}
private function processData()
{
if ($this->parsedTitle->nsCode() !== WikiTitle::NS_MAIN) {
return false;
}
if (!$this->parsePage()) {
return true;
}
// we're done, give up
try {
$configHolder = $this->configHandler->loadConfigFile();
} catch (\Exception $e) {
return $this->response(500, 'Internal server error', 'Failed to load config: ' . $e);
}
// forward source to ultrapedia
if (!$this->sendWiki()) {
return true;
}
// error response already sent
$secs = microtime(true);
$extractor = $configHolder->extractor;
// destinations must be SingletonQuadDestinations, so key doesn't matter
// destination must be a StringQuadDestination
// TODO: check types
$destination = $configHolder->destinations->getDestination('');
$pageUri = OntologyNamespaces::appendUri(OntologyNamespaces::DBPEDIA_INSTANCE_NAMESPACE, $this->parsedTitle->encoded());
$pageContext = new mapping\PageContext($pageUri);
$extractor->extract($this->parsedPage, $pageUri, $pageContext);
$delete = $this->delete($this->parsedPage->getSourceUriPrefix());
$insert = $this->insert((string) $destination);
$sparul = $this->sparul(array($delete, $insert));
$this->log->info('Successfully extracted data (' . (microtime(true) - $secs) * 1000 . ' millis)');
// send triples to triple store
if (!$this->sendSparul($sparul)) {
return true;
}
// error response already sent
return $this->response(200, 'Successfully extracted data', $this->reply ? "Successfully extracted data.\n\n" . $sparul : null);
}
/**
* The uri of this property.
*
* @return string The uri of this property e.g. http://xmlns.com/foaf/0.1/name
*/
public function getUri()
{
return OntologyNamespaces::getUri($this->name, OntologyNamespaces::DBPEDIA_PROPERTY_NAMESPACE);
}
