function run()
{
$dataset_description = '';
$ldir = parent::GetParameterValue('indir');
$odir = parent::GetParameterValue('outdir');
//download
if ($this->GetParameterValue('download') == true) {
$list = $this->getFtpFileList('ftp.ncbi.nih.gov');
$total = count($list);
$counter = 1;
foreach ($list as $f) {
echo "downloading file {$counter} out of {$total} :" . parent::getParameterValue('download_url') . $f . "... " . PHP_EOL;
file_put_contents($ldir . $f, file_get_contents(parent::GetParameterValue('download_url') . $f));
$counter++;
}
}
//if download
//iterate over the files
$paths = $this->getFilePaths($ldir, 'gz');
$lfile = null;
foreach ($paths as $aPath) {
$lfile = $aPath;
$ofile = $odir . basename($aPath, ".gz") . "." . parent::getParameterValue('output_format');
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
parent::setWriteFile($ofile, $gz);
parent::setReadFile($ldir . $lfile, true);
$source_file = (new DataResource($this))->setURI(parent::getParameterValue('download_url') . basename($aPath))->setTitle('NCBI UniSTS filename: ' . basename($aPath))->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($ldir . $lfile)))->setFormat('xml/unists-format')->setFormat('application/zip')->setPublisher('https://www.ncbi.nlm.nih.gov')->setHomepage('https://www.ncbi.nlm.nih.gov/unists')->setRights('use')->setRights('attribution')->setLicense('https://www.nlm.nih.gov/copyright.html')->setDataset(parent::getDatasetURI());
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = date("Y-m-d\\TG:i:s\\Z");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/unists/unists.php")->setCreateDate($date)->setHomepage("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$prefix}.html")->setPublisher("http://bio2rdf.org")->setRights("use-share-modify")->setRights("restricted-by-source-license")->setLicense("http://creativecommons/licenses/by/3.0/")->setDataset(parent::getDatasetURI());
$dataset_description .= $output_file->toRDF() . $source_file->toRDF();
echo "processing {$aPath} ...";
$this->process();
echo "done!" . PHP_EOL;
$this->setWriteFile($odir . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
}
//foreach
}
