/**
* 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);
}
public function execute()
{
$indata_file = $this->getOption('indata', '');
$chunksize = intval($this->getOption('chunksize', 0));
$chunksleep = floatval($this->getOption('chunksleep', 0.0));
$offset = intval($this->getOption('offset', 0));
$verbose = $this->getOption('verbose', false);
echo "Starting import from file: {$indata_file}\n";
if ($offset > 0) {
echo "Starting with offset {$offset} ...\n";
}
$rdfImporter = new RDFIORDFImporter();
$indata_fh = fopen($indata_file, 'r');
$lineinchunk = 1;
$chunkindex = 1;
$lineindex = 0;
$totalimported = 0;
$importdata = '';
while ($line = fgets($indata_fh)) {
if ($lineindex >= $offset) {
if ($chunksize > 0 && $lineinchunk == 1) {
if ($verbose) {
echo "Starting chunk {$chunkindex} ...\n";
}
}
$importdata .= $line;
if ($verbose) {
echo "Appended line {$lineinchunk} in chunk {$chunkindex}, to indata ...\n";
}
$totalimported++;
if ($chunksize != 0 && $lineinchunk == $chunksize) {
$rdfImporter->importTurtle($importdata);
$totalwithoffset = $totalimported + $offset;
echo "Imported {$chunksize} triples in chunk {$chunkindex} ({$totalimported} triples imported in total, and {$totalwithoffset} including offset)!\n";
// Reset variables
$lineinchunk = 0;
$importdata = '';
// Bump chunk index
$chunkindex++;
if ($verbose) {
echo 'Now sleeping for ' . strval($chunksleep) . ' seconds before continuing with next chunk ...';
}
sleep($chunksleep);
}
$lineinchunk++;
}
$lineindex++;
}
// Import any remaining stuff, or all the stuff, if chunksize = 0
$rdfImporter->importTurtle($importdata);
fclose($indata_fh);
echo "Finished importing everything ({$totalimported} triples in total)!\n";
}
/**
* After a query is parsed, import the parsed data to the wiki
*/
function importTriplesInQuery()
{
if ($this->checkAllowInsert()) {
$triples = $this->requestdata->query_parsed['query']['construct_triples'];
$rdfImporter = new RDFIORDFImporter();
$rdfImporter->importTriples($triples);
$this->successMsg("Successfully imported the triples!");
}
}
protected function import($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);
}
