add indexing task

inc/inc.ClassSchedulerTaskBase.php

@@ -37,11 +37,14 @@ class SeedDMS_SchedulerTaskBase {
 
 	var $logger;
 
-	public function __construct($dms=null, $user=null, $settings=null, $logger=null) { /* {{{ */
+	var $fulltextservice;
+
+	public function __construct($dms=null, $user=null, $settings=null, $logger=null, $fulltextservice=null) { /* {{{ */
 		$this->dms = $dms;
 		$this->user = $user;
 		$this->settings = $settings;
 		$this->logger = $logger;
+		$this->fulltextservice = $fulltextservice;
 	} /* }}} */
 
 	public function execute($task) { /* {{{ */

inc/inc.Tasks.php

@@ -7,7 +7,7 @@ require_once("inc/inc.ClassSchedulerTaskBase.php");
  *
  * @author  Uwe Steinmann <uwe@steinmann.cx>
  * @package SeedDMS
- * @subpackage  trash
+ * @subpackage  core
  */
 class SeedDMS_ExpiredDocumentsTask extends SeedDMS_SchedulerTaskBase { /* {{{ */
 
@@ -43,4 +43,143 @@ class SeedDMS_ExpiredDocumentsTask extends SeedDMS_SchedulerTaskBase { /* {{{ */
 	}
 } /* }}} */
 
+/**
+ * Class for processing a single folder
+ *
+ * SeedDMS_Task_Indexer_Process_Folder::process() is used as a callable when
+ * iterating over all folders recursively.
+ */
+class SeedDMS_Task_Indexer_Process_Folder { /* {{{ */
+	protected $forceupdate;
+
+	protected $fulltextservice;
+
+	public function __construct($fulltextservice, $forceupdate) { /* {{{ */
+		$this->fulltextservice = $fulltextservice;
+		$this->forceupdate = $forceupdate;
+	} /* }}} */
+
+	public function process($folder) { /* {{{ */
+		$documents = $folder->getDocuments();
+		if($documents) {
+			$lucenesearch = $this->fulltextservice->Search();
+			echo $folder->getFolderPathPlain()."\n";
+			foreach($documents as $document) {
+				echo $document->getId().":".$document->getName()." ";
+				/* If the document wasn't indexed before then just add it */
+				if(!($hit = $lucenesearch->getDocument($document->getId()))) {
+					try {
+						$idoc = $this->fulltextservice->IndexedDocument($document, true);
+						if(isset($GLOBALS['SEEDDMS_HOOKS']['indexDocument'])) {
+							foreach($GLOBALS['SEEDDMS_HOOKS']['indexDocument'] as $hookObj) {
+								if (method_exists($hookObj, 'preIndexDocument')) {
+									$hookObj->preIndexDocument(null, $document, $idoc);
+								}
+							}
+						}
+						$this->fulltextservice->Indexer()->addDocument($idoc);
+						echo " (Document added)".PHP_EOL;
+					} catch(Exception $e) {
+						echo " (Timeout)".PHP_EOL;
+					}
+				} else {
+					/* Check if the attribute created is set or has a value older
+					 * than the lastet content. Documents without such an attribute
+					 * where added when a new document was added to the dms. In such
+					 * a case the document content  wasn't indexed.
+					 */
+					try {
+						$created = (int) $hit->getDocument()->getFieldValue('created');
+					} catch (/* Zend_Search_Lucene_ */Exception $e) {
+						$created = 0;
+					}
+					$content = $document->getLatestContent();
+					if($created >= $content->getDate() && !$this->forceupdate) {
+						echo getMLText('index_document_unchanged').PHP_EOL;
+					} else {
+						$this->fulltextservice->Indexer()->delete($hit->id);
+						try {
+							$idoc = $this->fulltextservice->IndexedDocument($document, true);
+							if(isset($GLOBALS['SEEDDMS_HOOKS']['indexDocument'])) {
+								foreach($GLOBALS['SEEDDMS_HOOKS']['indexDocument'] as $hookObj) {
+									if (method_exists($hookObj, 'preIndexDocument')) {
+										$hookObj->preIndexDocument(null, $document, $idoc);
+									}
+								}
+							}
+							$this->fulltextservice->Indexer()->addDocument($idoc);
+							echo " (Document updated)".PHP_EOL;
+						} catch(Exception $e) {
+							echo " (Timeout)".PHP_EOL;
+						}
+					}
+				}
+			}
+		}
+	} /* }}} */
+} /* }}} */
+
+/**
+ * Class containing methods for running a scheduled task
+ *
+ * @author  Uwe Steinmann <uwe@steinmann.cx>
+ * @package SeedDMS
+ * @subpackage  core
+ */
+class SeedDMS_IndexingDocumentsTask extends SeedDMS_SchedulerTaskBase { /* {{{ */
+
+	/**
+	 * Run the task
+	 *
+	 * @param $task task to be executed
+	 * @param $dms dms
+	 * @return boolean true if task was executed succesfully, otherwise false
+	 */
+	public function execute($task) {
+		$dms = $this->dms;
+		$fulltextservice = $this->fulltextservice;
+		$taskparams = $task->getParameter();
+		$folder = $dms->getRootFolder();
+		$recreate = isset($taskparams['recreate']) ? $taskparams['recreate'] : false;
+
+		if($fulltextservice) {
+			if($recreate) {
+				$index = $fulltextservice->Indexer(true);
+				if(!$index) {
+					UI::exitError(getMLText("admin_tools"),getMLText("no_fulltextindex"));
+				}
+			} else {
+				$index = $fulltextservice->Indexer(false);
+				if(!$index) {
+					$index = $fulltextservice->Indexer(true);
+					if(!$index) {
+						UI::exitError(getMLText("admin_tools"),getMLText("no_fulltextindex"));
+					}
+				}
+			}
+		}
+
+		$folderprocess = new SeedDMS_Task_Indexer_Process_Folder($fulltextservice, $recreate);
+		$tree = new SeedDMS_FolderTree($folder, array($folderprocess, 'process'));
+		call_user_func(array($folderprocess, 'process'), $folder);
+
+		return true;
+	}
+
+	public function getDescription() {
+		return 'Indexing all new or updated documents';
+	}
+
+	public function getAdditionalParams() {
+		return array(
+			array(
+				'name'=>'recreate',
+				'type'=>'boolean',
+				'description'=> 'Force recreation of index',
+			)
+		);
+	}
+} /* }}} */
+
 $GLOBALS['SEEDDMS_SCHEDULER']['tasks']['core']['expireddocs'] = 'SeedDMS_ExpiredDocumentsTask';
+$GLOBALS['SEEDDMS_SCHEDULER']['tasks']['core']['indexingdocs'] = 'SeedDMS_IndexingDocumentsTask';

inc/inc.Utils.php

@@ -650,13 +650,13 @@ function seed_pass_verify($password, $hash) { /* {{{ */
 } /* }}} */
 
 function resolveTask($task) {
-	global $dms, $user, $settings, $logger;
+	global $dms, $user, $settings, $logger, $fulltextservice;
 
 	if(is_object($task))
 		return $task;
 	if(is_string($task)) {
 		if(class_exists($task)) {
-			$task = new $task($dms, $user, $settings, $logger);
+			$task = new $task($dms, $user, $settings, $logger, $fulltextservice);
 		}
 	}
 	return $task;

op/op.SchedulerTaskMgr.php

@@ -53,7 +53,7 @@ if ($action == "addtask") { /* {{{ */
 	$description = $_POST["description"];
 	$frequency = $_POST["frequency"];
 	$disabled = isset($_POST["disabled"]) ? $_POST["disabled"] : 0;
-	$params = $_POST["params"];
+	$params = isset($_POST["params"]) ? $_POST["params"] : null;
 
 	$newtask = $scheduler->addTask($extension, $task, $name, $description, $frequency, $disabled, $params);
 	if ($newtask) {
@@ -90,7 +90,7 @@ else if ($action == "edittask") { /* {{{ */
 	$description = $_POST["description"];
 	$frequency = $_POST["frequency"];
 	$disabled = isset($_POST["disabled"]) ? $_POST["disabled"] : 0;
-	$params = $_POST["params"];
+	$params = isset($_POST["params"]) ? $_POST["params"] : null;
 
 	if ($editedtask->getName() != $name)
 		$editedtask->setName($name);

views/bootstrap/class.Indexer.php

@@ -27,7 +27,7 @@ require_once("class.Bootstrap.php");
 class SeedDMS_View_Indexer_Process_Folder { /* {{{ */
 	protected $forceupdate;
 
-	protected $index;
+	protected $fulltextservice;
 
 	public function __construct($fulltextservice, $forceupdate) { /* {{{ */
 		$this->fulltextservice = $fulltextservice;