app/DoctrineMigrations/Version20140129151724.php
changeset 109 fc56f9e28cdb
parent 108 48af4fad8a44
child 110 7602e5ca9c30
--- a/app/DoctrineMigrations/Version20140129151724.php	Thu Jan 30 17:52:14 2014 +0100
+++ b/app/DoctrineMigrations/Version20140129151724.php	Fri Jan 31 16:08:50 2014 +0100
@@ -20,17 +20,26 @@
         // this up() migration is autogenerated, please modify it to your needs
         $this->abortIf($this->connection->getDatabasePlatform()->getName() != "mysql");
         
+        
         // First we get all tags.
         $em = $GLOBALS["kernel"]->getContainer()->get( 'doctrine.orm.entity_manager' );
-        $query = $em->createQuery('SELECT t FROM WikiTagBundle:Tag t ORDER BY t.label ASC');//->setMaxResults(40)->setFirstResult(5000);
+        // Avoid php annoying memory leaks
+        $em->getConnection()->getConfiguration()->setSQLLogger(null);
+        
+        // First step : we populate the dbpedia uris thanks to the dbpedia-owl:wikiPageID
+        echo "\nFIRST STEP";
+        $query = $em->createQuery('SELECT t FROM WikiTagBundle:Tag t ORDER BY t.label ASC');//->setMaxResults(240)->setFirstResult(500);
         $tags = $query->getResult();
-        
         $i = 1;
+        $nb_set = 0;
         foreach($tags as $tag){
         	$l = $tag->getLabel();
-        	$uri = WikiTagUtils::getDbpediaUri($tag->getLabel(), [], false);
+        	$uri = WikiTagUtils::getDbpediaUri($tag->getWikipediaPageId(), [], false, "pageid");
         	$tag->setDbpediaUri($uri);
         	$em->persist($tag);
+         if($uri!=NULL && $uri!=""){
+             $nb_set++;
+         }
         	if( $i % 50 == 0 ){
         		$em->flush();
         		echo "\n    FLUSH";
@@ -39,6 +48,57 @@
         	echo "\n$i : $l \t\t: $uri";
         }
         $em->flush();
+        echo "\nFIRST STEP : $nb_set uris found";
+        
+        
+        // Second step : we populate the dbpedia uris not found thanks to the foaf:isPrimaryTopicOf
+        echo "\nSECOND STEP";
+        $query = $em->createQuery('SELECT t FROM WikiTagBundle:Tag t WHERE (t.dbpediaUri=\'\' OR t.dbpediaUri IS NULL) ORDER BY t.label ASC');//->setMaxResults(240);
+        $tags = $query->getResult();
+        $i = 1;
+        $nb_set = 0;
+        foreach($tags as $tag){
+        	$l = $tag->getLabel();
+        	$uri = WikiTagUtils::getDbpediaUri($tag->getWikipediaUrl(), [], false, "wikiurl");
+        	$tag->setDbpediaUri($uri);
+        	$em->persist($tag);
+            if($uri!=NULL && $uri!=""){
+                $nb_set++;
+            }
+        	if( $i % 50 == 0 ){
+        		$em->flush();
+        		echo "\n    FLUSH";
+        	}
+        	$i++;
+        	echo "\n$i : $l \t\t: $uri";
+        }
+        $em->flush();
+        echo "\nSECOND STEP : $nb_set uris found";
+        
+        
+        // Third step : we populate the dbpedia uris not found thanks to the rdfs:label
+        echo "\nTHIRD STEP";
+        $query = $em->createQuery('SELECT t FROM WikiTagBundle:Tag t WHERE (t.dbpediaUri=\'\' OR t.dbpediaUri IS NULL) ORDER BY t.label ASC');//->setMaxResults(240);
+        $tags = $query->getResult();
+        $i = 1;
+        $nb_set = 0;
+        foreach($tags as $tag){
+        	$l = $tag->getLabel();
+        	$uri = WikiTagUtils::getDbpediaUri($tag->getLabel(), [], false);
+        	$tag->setDbpediaUri($uri);
+        	$em->persist($tag);
+            if($uri!=NULL && $uri!=""){
+                $nb_set++;
+            }
+        	if( $i % 50 == 0 ){
+        		$em->flush();
+        		echo "\n    FLUSH";
+        	}
+        	$i++;
+        	echo "\n$i : $l \t\t: $uri";
+        }
+        $em->flush();
+        echo "\nTHIRD STEP : $nb_set uris found";
     }
 
     public function down(Schema $schema)