/**
* Import data into wiki pages
*/
function importData(RDFIORequestData $requestData)
{
$rdfImporter = new RDFIORDFImporter();
if ($requestData->importSource === 'url') {
if ($requestData->externalRdfUrl === '') {
throw new RDFIOException('URL field is empty!');
} else {
if (!RDFIOUtils::isURI($requestData->externalRdfUrl)) {
throw new RDFIOException('Invalid URL provided!');
}
}
$rdfData = file_get_contents($requestData->externalRdfUrl);
} else {
if ($requestData->importSource === 'textfield') {
if ($requestData->importData === '') {
throw new RDFIOException('RDF field is empty!');
}
$rdfData = $requestData->importData;
} else {
throw new RDFIOException('Import source is not selected!');
}
}
switch ($requestData->dataFormat) {
case 'rdfxml':
$importInfo = $rdfImporter->importRdfXml($rdfData);
$triples = $importInfo['triples'];
break;
case 'turtle':
$importInfo = $rdfImporter->importTurtle($rdfData);
$triples = $importInfo['triples'];
break;
}
return $output = array('triples' => $triples);
}
protected function import($limit = 10, $offset = 0)
{
global $wgOut, $wgRequest;
//$rdfioUtils = new RDFIOUtils();
$externalSparqlUrl = $wgRequest->getText('extsparqlurl');
if ($externalSparqlUrl === '') {
throw new RDFIOException('Empty SPARQL Url provided!');
} else {
if (!RDFIOUtils::isURI($externalSparqlUrl)) {
throw new RDFIOException('Invalid SPARQL Url provided! (Must start with \'http://\' or \'https://\')');
}
}
$sparqlQuery = urlencode("SELECT DISTINCT * WHERE { ?s ?p ?o } OFFSET {$offset} LIMIT {$limit}");
$sparqlQueryUrl = $externalSparqlUrl . '/' . '?query=' . $sparqlQuery;
$sparqlResultXml = file_get_contents($sparqlQueryUrl);
$sparqlResultXmlObj = simplexml_load_string($sparqlResultXml);
$importTriples = array();
if (is_object($sparqlResultXmlObj)) {
foreach ($sparqlResultXmlObj->results->children() as $result) {
$triple = array();
// $wgOut->addHTML( print_r($result, true) );
foreach ($result as $binding) {
if ($binding['name'] == 's') {
$s = (string) $binding->uri[0];
if ($s == '') {
throw new Exception('Could not extract subject from empty string (' . print_r($binding->uri, true) . '), in SPARQLImport');
}
$triple['s'] = $s;
$triple['s_type'] = $this->resourceType($triple['s']);
} else {
if ($binding['name'] == 'p') {
$p = (string) $binding->uri[0];
if ($p == '') {
throw new Exception('Could not extract predicate from empty string (' . print_r($binding->uri, true) . '), in SPARQLImport');
}
$triple['p'] = $p;
$triple['p_type'] = $this->resourceType($triple['p']);
} else {
if ($binding['name'] == 'o') {
$o = (string) $binding->uri[0];
if ($o == '') {
throw new Exception('Could not extract object from empty string (' . print_r($binding->uri, true) . '), in SPARQLImport');
}
$triple['o'] = $o;
$triple['o_type'] = $this->resourceType($triple['o']);
$triple['o_datatype'] = '';
}
}
}
}
$importTriples[] = $triple;
}
$rdfImporter = new RDFIORDFImporter();
$rdfImporter->importTriples($importTriples);
$wgOut->addHTML($rdfImporter->showImportedTriples($importTriples));
} else {
RDFIOUtils::formatErrorHTML("Error", "There was a problem importing from the endpoint. Are you sure that the given URL is a valid SPARQL endpoint?");
}
return $output = array('externalSparqlUrl' => $externalSparqlUrl);
}
/**
* Use the namespaces from the RDF / SPARQL source, to shorten the URIs.
* @param string $uri
* @param array $nsPrefixes
* @return string
*/
function abbreviateParserNSPrefixes($uri, $nsPrefixes)
{
foreach ($nsPrefixes as $namespace => $prefix) {
$nslength = strlen($namespace);
$basepart = '';
$localpart = '';
$uriContainsNamepace = substr($uri, 0, $nslength) === $namespace;
if ($uriContainsNamepace) {
$localpart = substr($uri, $nslength);
$basepart = $prefix;
break;
}
}
/*
* Take care of some special cases:
*/
if ($basepart === '' && $localpart === '') {
$uriParts = $this->splitURI($uri);
$basepart = $uriParts[0];
$localpart = $uriParts[1];
}
if ($localpart === '') {
$abbreviatedUri = $basepart;
} elseif (RDFIOUtils::isURI($basepart)) {
// FIXME: Shouldn't the above check the local part instead??
// Change ARC:s default "random string", to indicate more clearly that
// it lacks title
$abbreviatedUri = str_replace('arc', 'untitled', $localpart);
} elseif (RDFIOUtils::isURI($basepart)) {
// If the abbreviation does not seem to have succeeded,
// fall back to use only the local part
$abbreviatedUri = $localpart;
} elseif (RDFIOUtils::endsWithColon($basepart)) {
// Don't add another colon
$abbreviatedUri = $basepart . $localpart;
} elseif ($basepart == false || $basepart == '') {
$abbreviatedUri = $localpart;
} else {
$abbreviatedUri = $basepart . ':' . $localpart;
}
return $abbreviatedUri;
}
function showImportedTriples($importedTriples)
{
$output = "";
$style_css = <<<EOD
\t \t table .rdfio- th {
\t \t font-weight: bold;
\t \t padding: 2px 4px;
\t \t }
\t \t table.rdfio-table td,
\t \t table.rdfio-table th {
\t \t font-size: 11px;
\t \t }
EOD;
$output .= "<style>{$style_css}</style>";
//$wgOut->addInlineStyle($style_css);
$output .= RDFIOUtils::formatSuccessMessageHTML("Success!", "Successfully imported the triples shown below!");
$output .= "<table class=\"wikitable sortable rdfio-table\"><tbody><tr><th>Subject</th><th>Predicate</th><th>Object</th></tr>";
foreach ($importedTriples as $triple) {
$s = $triple['s'];
$p = $triple['p'];
$o = $triple['o'];
if (RDFIOUtils::isURI($s)) {
$s = "<a href=\"{$s}\">{$s}</a>";
}
if (RDFIOUtils::isURI($p)) {
$p = "<a href=\"{$p}\">{$p}</a>";
}
if (RDFIOUtils::isURI($o)) {
$o = "<a href=\"{$o}\">{$o}</a>";
}
$output .= "<tr><td>{$s}</td><td>{$p}</td><td>{$o}</td></tr>";
}
$output .= "</tbody></table>";
return $output;
}
