Merge pull request #421 from micheldumontier/release3

Major updates to scripts
bio2rdf · Oct 7, 2015 · a3ef2b4 · a3ef2b4
2 parents 6539d4d + 86c22e6
commit a3ef2b4
Show file tree

Hide file tree

Showing 19 changed files with 1,217 additions and 996 deletions.
diff --git a/README.md b/README.md
diff --git a/bioportal/bioportal.php b/bioportal/bioportal.php
@@ -89,8 +89,9 @@ function Run()
 		foreach($ontologies AS $i => $o) {
 			$label = (string) $o->name;
 			$abbv = (string) $o->acronym;
-
-			if(array_search($abbv,$exclude_list) !== FALSE) continue;
+			if(array_search($abbv,$exclude_list) !== FALSE) {
+				continue;
+			}
 			if($include_list[0] != 'all') {
 				// ignore if we don't find it in the include list OR we do find it in the exclude list
 				if( (array_search($abbv,$include_list) === FALSE)
@@ -106,9 +107,7 @@ function Run()
 			if(!isset($ls['hasOntologyLanguage'])) {echo 'insufficient metadata'.PHP_EOL;continue;}
 
 			$format = strtolower($ls['hasOntologyLanguage']);
-/***********/
 			if($format != 'owl' and $format != 'obo') continue;
-
 			echo "Processing ($i/$total) $abbv ... ";
 
 			$version = $ls['version'];
@@ -117,7 +116,6 @@ function Run()
 
 			$rfile = $ls['ontology']['links']['download'];
 
-
 			$lfile = $abbv.".".$format.".gz";
 			if(parent::getParameterValue('download') == 'true') {
 				echo "downloading ... ";
@@ -138,6 +136,7 @@ function Run()
 				if(isset($m[1])) {
 					$filename = $m[1];
 					if(strstr($filename,".zip"))  continue;
+
 				} else {echo "error: no filename".PHP_EOL;continue;}
 
 				$body = substr($ret, $header_size);
@@ -148,6 +147,7 @@ function Run()
 
 				$lz = "compress.zlib://".$idir.$lfile;
 				file_put_contents($lz,$body);
+
 				echo "done".PHP_EOL;
 			}
 
@@ -160,6 +160,11 @@ function Run()
 				// process
 				echo "converting ... ";
 				set_time_limit(0);
+				// let's double check the format
+				$fp = gzopen($idir.$lfile,"r");
+				$l = gzgets($fp);
+				if(strstr($l,"xml")) $format= "owl";
+				gzclose($fp);
 				if($format == 'obo') {
 					$this->OBO2RDF($abbv);
 				} else if($format == 'owl') {
@@ -366,6 +371,7 @@ function OBO2RDF($abbv)
 		if($abbv == "doid") $abbv = "do";
 		$minimal = (parent::getParameterValue('detail') == 'min')?true:false;
 		$minimalp = (parent::getParameterValue('detail') == 'min+')?true:false;
+		$version = parent::getParameterValue("bio2rdf_release");
 
 		$tid = '';
 		$first = true;
@@ -374,7 +380,7 @@ function OBO2RDF($abbv)
 		$min = $buf = '';
 		$ouri = "http://bio2rdf.org/lsr:".$abbv;
 
-		$dataset_uri = $abbv."_resource:bio2rdf.dataset.$abbv.R3";
+		$dataset_uri = $abbv."_resource:bio2rdf.dataset.$abbv.R".$version;
 		parent::setGraphURI($dataset_uri);
 		$buf = parent::triplify($ouri,"rdf:type","owl:Ontology");
 		$graph_uri = '<'.parent::getRegistry()->getFQURI(parent::getGraphURI()).'>';

diff --git a/clinicaltrials/clinicaltrials.php b/clinicaltrials/clinicaltrials.php
@@ -233,12 +233,11 @@ function parse_dir(){
 	**/
 	function process_file($infile) {
 		$indir = parent::getParameterValue('indir');
-		$xml = new CXML($indir,basename($infile));
+		$xml = new CXML($infile);
 		$this->setCheckPoint('file');
 		while($xml->Parse("clinical_study") == TRUE) {
 			$this->setCheckPoint('record');
 			$this->root = $root = $xml->GetXMLRoot();
-
 			$this->nct_id = $nct_id = $this->getString("//id_info/nct_id");
 			$this->study_id = $study_id = parent::getNamespace()."$nct_id";
 
@@ -759,7 +758,7 @@ function process_file($infile) {
 						parent::triplifyString($location_uri,parent::getVoc()."status", $this->getString('//status',$location)).
 						parent::triplify($study_id,parent::getVoc()."location",$location_uri).
 						parent::triplify($location_uri, parent::getVoc()."address", $this->makeAddress($address)).
-						parent::triplify($location_uri, parent::getVoc()."contact", $this->makeContact($contact))
+						($contact != null?parent::triplify($location_uri, parent::getVoc()."contact", $this->makeContact($contact)):"")
 					);
 					if($backups) {
 						foreach($backups AS $backup) {
@@ -831,11 +830,13 @@ function process_file($infile) {
 			try{
 				$links = $root->xpath('//link');
 				foreach($links AS $i => $link) {
-					$lid = parent::getRes().md5($this->getString('./url',$link));
+					$url = $this->getString('./url',$link);
+					$url = preg_replace("/>.*$/","",$url);
+					$lid = parent::getRes().md5($url);
 					parent::addRDF(
 						parent::describeIndividual($lid, $this->getString('./description',$link), parent::getVoc()."Link").
 						parent::describeClass(parent::getVoc()."Link","Link").
-						parent::triplify($lid,parent::getVoc()."url",preg_replace("/>$/","",$this->getString('./url',$link))).
+						parent::triplify($lid,parent::getVoc()."url",$url).
 						parent::triplify($study_id,parent::getVoc()."link",$lid)
 					);
 				}
@@ -1166,7 +1167,7 @@ function process_file($infile) {
 						if(!$et) continue;
 						$ev_uri = parent::getVoc().str_replace(" ","-",$ev_label);
 
-						$categories = array_shift($et->xpath('./category_list'));
+						$categories = @array_shift($et->xpath('./category_list'));
 						foreach($categories AS $category) {
 							$major_title = $this->getString('./title', $category);
 							$major_title_uri = parent::getRes().md5($major_title);
@@ -1272,8 +1273,8 @@ public function getDatetimeFromDate($date)
 
 	public function makeContact($contact)
 	{
-		if($contact == null) return null;
-		$contact_uri = parent::getRes().md5($contact->asXML());		
+		if($contact == null) return '';
+		$contact_uri = parent::getRes().md5($contact->asXML());
 		$contact_type_uri = parent::getVoc()."Contact";
 		$contact_label = trim($this->getString('//first_name',$contact)." ".$this->getString('//last_name', $contact));
 		parent::addRDF(