KTS-673

"The search algorithm needs some work" Updated. Include version of document to the index. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/trunk@7239 c91229c3-7414-0410-bfa2-8a42b809f60b

KTS-673
"The search algorithm needs some work" Updated. Include version of document to the index. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/trunk@7239 c91229c3-7414-0410-bfa2-8a42b809f60b
conradverm
1 parent 03fc43f8
Showing 2 changed files with 157 additions and 14 deletions
search2/indexing/indexers/JavaXMLRPCLuceneIndexer.inc.php
search2/indexing/indexers/PHPLuceneIndexer.inc.php
 <?
-require_once('indexer.inc.php');
+
+require_once('indexing/lib/XmlRpcLucene.inc.php');
 class JavaXMLRPCLuceneIndexer extends Indexer
 {
-    protected function indexDocument($docid, $textfile)
+	/**
+	 * @var XmlRpcLucene
+	 */
+	private $lucene;
+
+	/**
+	 * The constructor for PHP Lucene
+	 *
+	 * @param boolean $create Optional. If true, the lucene index will be recreated.
+	 */
+	public function __construct()
+	{
+		parent::__construct();
+
+		$config =& KTConfig::getSingleton();
+		$javaServerUrl = $config->get('indexer/JavaLuceneURL', 'http://localhost:8875');
+		$this->lucene = new XmlRpcLucene($javaServerUrl);
+	}
+
+	/**
+	 * Creates an index to be used.
+	 *
+	 */
+	public static function createIndex()
+	{
+		// do nothing. The java lucene indexer will create the indexes if required
+	}
+
+	/**
+	 * Indexes a document based on a text file.
+	 *
+	 * @param int $docid
+	 * @param string $textfile
+	 * @return boolean
+	 */
+    protected function indexDocument($docid, $textfile, $title, $version)
     {
-		throw new Exception('TODO');
+    	try
+    	{
+	    	return $this->lucene->addDocument($docid, $textfile, '', $title, $version);
+    	}
+    	catch(Exception $e)
+    	{
+    		return false;
+    	}
     }
+
+    /**
+     * Indexes the content and discussions on a document.
+     *
+     * @param int $docid
+     * @param string $textfile
+     * @return boolean
+     */
+    protected function indexDocumentAndDiscussion($docid, $textfile, $title, $version)
+    {
+    	try
+    	{
+	    	$discussion = Indexer::getDiscussionText($docid);
+    		return $this->lucene->addDocument($docid, $textfile, $discussion, $title, $version);
+    	}
+    	catch(Exception $e)
+    	{
+    		return false;
+    	}
+    }
+
+    /**
+     * Indexes a discussion on a document..
+     *
+     * @param int $docid
+     * @return boolean
+     */
+    protected function indexDiscussion($docid)
+    {
+    	try
+    	{
+    		$discussion = Indexer::getDiscussionText($docid);
+    		return $this->lucene->updateDiscussion($docid, $discussion);
+    	}
+    	catch(Exception $e)
+    	{
+    		return false;
+    	}
+
+		return true;
+    }
+
+    /**
+     * Optimise the lucene index.
+     * This can be called periodically to optimise performance and size of the lucene index.
+     *
+     */
+    public function optimise()
+    {
+    	$this->lucene->optimize();
+    }
+
+    /**
+     * Removes a document from the index.
+     *
+     * @param int $docid
+     * @return array containing (content, discussion, title)
+     */
+    public function deleteDocument($docid)
+    {
+    	return $this->lucene->deleteDocument($docid);
+    }
+
+    /**
+     * Enter description here...
+     *
+     * @param string $query
+     * @return array
+     */
+    public function query($query)
+    {
+    	$results = array();
+    	$hits = $this->lucene->query($query);
+    	if (is_array($hits))
+    	{
+    		foreach ($hits as $hit)
+    		{
+
+
+    			$document_id 	= $hit->DocumentID;
+    			$content 		= $hit->Text;
+    			$discussion 	= $hit->Title; //TODO: fix to be discussion. lucen server is not returning discussion text as well..
+    			$title 			= $hit->Title;
+    			$score 			= $hit->Rank;
+
+    			// avoid adding duplicates. If it is in already, it has higher priority.
+    			if (!array_key_exists($document_id, $results) || $score > $results[$document_id]->Score)
+    			{
+    				$results[$document_id] = new QueryResultItem($document_id,  $score, $title,  $content, $discussion);
+    			}
+    		}
+    	}
+    	else
+    	{
+			 $_SESSION['KTErrorMessage'][] = _kt('The XMLRPC Server did not respond correctly. Please notify the system administrator to investigate.');
+    	}
+        return $results;
+    }
+
 }
 ?>
 \ No newline at end of file
@@ -30,7 +30,7 @@ class PHPLuceneIndexer extends Indexer
 	{
 		$config =& KTConfig::getSingleton();
 		$indexPath = $config->get('indexer/luceneDirectory');
-		$lucene = new Zend_Search_Lucene($indexPath, true);
+		new Zend_Search_Lucene($indexPath, true);
 	}
@@ -41,13 +41,14 @@ class PHPLuceneIndexer extends Indexer
 	 * @param string $content
 	 * @param string $discussion
 	 */
-	private function addDocument($docid, $content, $discussion, $title='')
+	private function addDocument($docid, $content, $discussion, $title, $version)
 	{
 		$doc = new Zend_Search_Lucene_Document();
 		$doc->addField(Zend_Search_Lucene_Field::Text('DocumentID', PHPLuceneIndexer::longToString($docid)));
 		$doc->addField(Zend_Search_Lucene_Field::Text('Content', $content, 'UTF-8'));
 		$doc->addField(Zend_Search_Lucene_Field::Text('Discussion', $discussion, 'UTF-8'));
 		$doc->addField(Zend_Search_Lucene_Field::Text('Title', $title, 'UTF-8'));
+		$doc->addField(Zend_Search_Lucene_Field::Text('Version', $version, 'UTF-8'));
 		$this->lucene->addDocument($doc);
 	}
@@ -58,7 +59,7 @@ class PHPLuceneIndexer extends Indexer
 	 * @param string $textfile
 	 * @return boolean
 	 */
-    protected function indexDocument($docid, $textfile, $title='')
+    protected function indexDocument($docid, $textfile, $title, $version)
     {
     	global $default;
@@ -68,9 +69,9 @@ class PHPLuceneIndexer extends Indexer
     		return false;
     	}
-    	list($content, $discussion) = $this->deleteDocument($docid);
+    	list($content, $discussion, $title2, $version2) = $this->deleteDocument($docid);
-    	$this->addDocument($docid, file_get_contents($textfile), $discussion, $title);
+    	$this->addDocument($docid, file_get_contents($textfile), $discussion, $title, $version);
 		return true;
     }
@@ -82,7 +83,7 @@ class PHPLuceneIndexer extends Indexer
      * @param string $textfile
      * @return boolean
      */
-    protected function indexDocumentAndDiscussion($docid, $textfile, $title='')
+    protected function indexDocumentAndDiscussion($docid, $textfile, $title, $version)
     {
 		global $default;
@@ -94,7 +95,7 @@ class PHPLuceneIndexer extends Indexer
     	$this->deleteDocument($docid);
-    	$this->addDocument($docid, file_get_contents($textfile), Indexer::getDiscussionText($docid), $title);
+    	$this->addDocument($docid, file_get_contents($textfile), Indexer::getDiscussionText($docid), $title, $version);
     	return true;
     }
@@ -107,9 +108,9 @@ class PHPLuceneIndexer extends Indexer
      */
     protected function indexDiscussion($docid)
     {
-		list($content, $discussion, $title) = $this->deleteDocument($docid);
+		list($content, $discussion, $title, $version) = $this->deleteDocument($docid);
-		$this->addDocument($docid, $content, Indexer::getDiscussionText($docid), $title);
+		$this->addDocument($docid, $content, Indexer::getDiscussionText($docid), $title, $version);
 		return true;
     }
@@ -142,11 +143,11 @@ class PHPLuceneIndexer extends Indexer
     		$content = $hit->Content;
     		$discussion = $hit->Discussion;
     		$title = $hit->Title;
-    		$title='';
+    		$version = $hit->Version;
     		$this->lucene->delete($hit);
     	}
-    	return array($content, $discussion, $title);
+    	return array($content, $discussion, $title, $version);
     }
     /**