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";
}
/**
* 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);
}
