AOEpeople / crawler

domain/queue/class.tx_crawler_domain_queue_entry.php

Indentation +14 added lines, -14 removed lines

@@ -24,21 +24,21 @@
 
 class tx_crawler_domain_queue_entry extends tx_crawler_domain_lib_abstract_dbobject {
 
-	/**
-	 * @var string table name
-	 */
-	protected static $tableName = 'tx_crawler_queue';
+    /**
+     * @var string table name
+     */
+    protected static $tableName = 'tx_crawler_queue';
 
-	/**
-	 * Returns the execution time of the record as int value
-	 *
-	 * @param void
-	 * @author Timo Schmidt <timo.schmidt@aoemedia.de>
-	 * @return int
-	 */
-	public function getExecutionTime(){
-		return $this->row['exec_time'];
-	}
+    /**
+     * Returns the execution time of the record as int value
+     *
+     * @param void
+     * @author Timo Schmidt <timo.schmidt@aoemedia.de>
+     * @return int
+     */
+    public function getExecutionTime(){
+        return $this->row['exec_time'];
+    }
 
 }
 

Spacing +1 added lines, -1 removed lines

@@ -36,7 +36,7 @@
 	 * @author Timo Schmidt <timo.schmidt@aoemedia.de>
 	 * @return int
 	 */
-	public function getExecutionTime(){
+	public function getExecutionTime() {
 		return $this->row['exec_time'];
 	}
 

domain/lib/class.tx_crawler_domain_lib_abstract_repository.php

Indentation +28 added lines, -28 removed lines

@@ -24,36 +24,36 @@
 
 abstract class tx_crawler_domain_lib_abstract_repository {
 
-	/**
-	 * @var string object class name
-	 */
-	protected $objectClassname;
+    /**
+     * @var string object class name
+     */
+    protected $objectClassname;
 
-	/**
-	 * @var string table name
-	 */
-	protected $tableName;
+    /**
+     * @var string table name
+     */
+    protected $tableName;
 
-	/**
-	 * Returns an instance of the TYPO3 database class.
-	 *
-	 * @return \TYPO3\CMS\Core\Database\DatabaseConnection
-	 */
-	protected function getDB() {
-		return $GLOBALS['TYPO3_DB'];
-	}
+    /**
+     * Returns an instance of the TYPO3 database class.
+     *
+     * @return \TYPO3\CMS\Core\Database\DatabaseConnection
+     */
+    protected function getDB() {
+        return $GLOBALS['TYPO3_DB'];
+    }
 
-	/**
-	 * Counts items by a given where clause
-	 *
-	 * @param  string $where    Where clause
-	 * @return integer
-	 */
-	protected function countByWhere($where) {
-		$db  = $this->getDB();
-		$rs  = $db->exec_SELECTquery('count(*) as anz', $this->tableName, $where);
-		$res = $db->sql_fetch_assoc($rs);
+    /**
+     * Counts items by a given where clause
+     *
+     * @param  string $where    Where clause
+     * @return integer
+     */
+    protected function countByWhere($where) {
+        $db  = $this->getDB();
+        $rs  = $db->exec_SELECTquery('count(*) as anz', $this->tableName, $where);
+        $res = $db->sql_fetch_assoc($rs);
 
-		return $res['anz'];
-	}
+        return $res['anz'];
+    }
 }

domain/process/class.tx_crawler_domain_process_repository.php

Indentation +81 added lines, -81 removed lines

@@ -24,97 +24,97 @@
 
 class tx_crawler_domain_process_repository extends tx_crawler_domain_lib_abstract_repository {
 
-	/**
-	 * @var string object class name
-	 */
-	protected $objectClassname = 'tx_crawler_domain_process';
+    /**
+     * @var string object class name
+     */
+    protected $objectClassname = 'tx_crawler_domain_process';
 
-	/**
-	 * @var string
-	 */
-	protected $tableName = 'tx_crawler_process';
+    /**
+     * @var string
+     */
+    protected $tableName = 'tx_crawler_process';
 
-	/**
-	 * This method is used to find all cli processes within a limit.
-	 * 
-	 * @param  string  $orderField
-	 * @param  string  $orderDirection
-	 * @param  integer $itemCount
-	 * @param  integer $offset
-	 * @param  string  $where
-	 * @return tx_crawler_domain_process_collection
-	 */
-	public function findAll($orderField = '', $orderDirection = 'DESC', $itemCount = NULL, $offset = NULL, $where = '') {
-		/** @var tx_crawler_domain_process_collection $collection */
-		$collection = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('tx_crawler_domain_process_collection');
+    /**
+     * This method is used to find all cli processes within a limit.
+     * 
+     * @param  string  $orderField
+     * @param  string  $orderDirection
+     * @param  integer $itemCount
+     * @param  integer $offset
+     * @param  string  $where
+     * @return tx_crawler_domain_process_collection
+     */
+    public function findAll($orderField = '', $orderDirection = 'DESC', $itemCount = NULL, $offset = NULL, $where = '') {
+        /** @var tx_crawler_domain_process_collection $collection */
+        $collection = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('tx_crawler_domain_process_collection');
 
-		$orderField = trim($orderField);
-		$orderField = empty($orderField) ? 'process_id' : $orderField;
+        $orderField = trim($orderField);
+        $orderField = empty($orderField) ? 'process_id' : $orderField;
 
-		$orderDirection = strtoupper(trim($orderDirection));
-		$orderDirection = in_array($orderDirection, array('ASC', 'DESC')) ? $orderDirection : 'DESC';
+        $orderDirection = strtoupper(trim($orderDirection));
+        $orderDirection = in_array($orderDirection, array('ASC', 'DESC')) ? $orderDirection : 'DESC';
 
-		$rows = $this->getDB()->exec_SELECTgetRows(
-			'*',
-			$this->tableName,
-			$where,
-			'',
-			htmlspecialchars($orderField) . ' ' . htmlspecialchars($orderDirection),
-			self::getLimitFromItemCountAndOffset($itemCount, $offset)
-		);
+        $rows = $this->getDB()->exec_SELECTgetRows(
+            '*',
+            $this->tableName,
+            $where,
+            '',
+            htmlspecialchars($orderField) . ' ' . htmlspecialchars($orderDirection),
+            self::getLimitFromItemCountAndOffset($itemCount, $offset)
+        );
 
-		if (is_array($rows)) {
-			foreach($rows as $row) {
-				$collection->append(\TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance($this->objectClassname, $row));
-			}
-		}
+        if (is_array($rows)) {
+            foreach($rows as $row) {
+                $collection->append(\TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance($this->objectClassname, $row));
+            }
+        }
 
-		return $collection;
-	}
+        return $collection;
+    }
 
-	/**
-	 * This method is used to count all processes in the process table.
-	 *
-	 * @param  string $where    Where clause
-	 * @return integer
-	 * @author Timo Schmidt <timo.schmidt@aoe.com>
-	 */
-	public function countAll($where = '1 = 1') {
-		return $this->countByWhere($where);
-	}
+    /**
+     * This method is used to count all processes in the process table.
+     *
+     * @param  string $where    Where clause
+     * @return integer
+     * @author Timo Schmidt <timo.schmidt@aoe.com>
+     */
+    public function countAll($where = '1 = 1') {
+        return $this->countByWhere($where);
+    }
 
-	/**
-	 * Returns the number of active processes.
-	 *
-	 * @return integer
-	 */
-	public function countActive() {
-		return $this->countByWhere('active = 1 AND deleted = 0');
-	}
+    /**
+     * Returns the number of active processes.
+     *
+     * @return integer
+     */
+    public function countActive() {
+        return $this->countByWhere('active = 1 AND deleted = 0');
+    }
 
-	/**
-	 * Returns the number of processes that live longer than the given timestamp.
-	 *
-	 * @param  integer $ttl
-	 * @return integer
-	 */
-	public function countNotTimeouted($ttl) {
-		return $this->countByWhere('deleted = 0 AND ttl > ' . intval($ttl));
-	}
+    /**
+     * Returns the number of processes that live longer than the given timestamp.
+     *
+     * @param  integer $ttl
+     * @return integer
+     */
+    public function countNotTimeouted($ttl) {
+        return $this->countByWhere('deleted = 0 AND ttl > ' . intval($ttl));
+    }
 
-	/**
-	 * Get limit clause
-	 *
-	 * @param  integer $itemCount   Item count
-	 * @param  integer $offset      Offset
-	 * @return string               Limit clause
-	 * @author Fabrizio Branca <fabrizio.branca@aoem.com>
-	 */
-	public static function getLimitFromItemCountAndOffset($itemCount, $offset) {
-		$itemCount = filter_var($itemCount, FILTER_VALIDATE_INT, array('options' => array('min_range' => 1, 'default' => 20)));
-		$offset = filter_var($offset, FILTER_VALIDATE_INT, array('options' => array('min_range' => 0, 'default' => 0)));
-		$limit = $offset . ', ' . $itemCount;
+    /**
+     * Get limit clause
+     *
+     * @param  integer $itemCount   Item count
+     * @param  integer $offset      Offset
+     * @return string               Limit clause
+     * @author Fabrizio Branca <fabrizio.branca@aoem.com>
+     */
+    public static function getLimitFromItemCountAndOffset($itemCount, $offset) {
+        $itemCount = filter_var($itemCount, FILTER_VALIDATE_INT, array('options' => array('min_range' => 1, 'default' => 20)));
+        $offset = filter_var($offset, FILTER_VALIDATE_INT, array('options' => array('min_range' => 0, 'default' => 0)));
+        $limit = $offset . ', ' . $itemCount;
 
-		return $limit;
-	}
+        return $limit;
+    }
 }

Spacing +4 added lines, -4 removed lines

@@ -59,12 +59,12 @@ 
 			$this->tableName,
 			$where,
 			'',
-			htmlspecialchars($orderField) . ' ' . htmlspecialchars($orderDirection),
+			htmlspecialchars($orderField).' '.htmlspecialchars($orderDirection),
 			self::getLimitFromItemCountAndOffset($itemCount, $offset)
 		);
 
 		if (is_array($rows)) {
-			foreach($rows as $row) {
+			foreach ($rows as $row) {
 				$collection->append(\TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance($this->objectClassname, $row));
 			}
 		}
@@ -99,7 +99,7 @@ 
 	 * @return integer
 	 */
 	public function countNotTimeouted($ttl) {
-		return $this->countByWhere('deleted = 0 AND ttl > ' . intval($ttl));
+		return $this->countByWhere('deleted = 0 AND ttl > '.intval($ttl));
 	}
 
 	/**
@@ -113,7 +113,7 @@ 
 	public static function getLimitFromItemCountAndOffset($itemCount, $offset) {
 		$itemCount = filter_var($itemCount, FILTER_VALIDATE_INT, array('options' => array('min_range' => 1, 'default' => 20)));
 		$offset = filter_var($offset, FILTER_VALIDATE_INT, array('options' => array('min_range' => 0, 'default' => 0)));
-		$limit = $offset . ', ' . $itemCount;
+		$limit = $offset.', '.$itemCount;
 
 		return $limit;
 	}

domain/process/class.tx_crawler_domain_process_manager.php

Indentation +158 added lines, -158 removed lines

@@ -27,176 +27,176 @@
  *
  */
 class tx_crawler_domain_process_manager  {
-	/**
-	 * @var $timeToLive integer
-	 */
-	private $timeToLive;
-	/**
-	 * @var integer
-	 */
-	private $countInARun;
+    /**
+     * @var $timeToLive integer
+     */
+    private $timeToLive;
+    /**
+     * @var integer
+     */
+    private $countInARun;
 
-	/**
-	 * @var integer
-	 */
-	private $processLimit;
+    /**
+     * @var integer
+     */
+    private $processLimit;
 
-	/**
-	 * @var $crawlerObj tx_crawler_lib
-	 */
-	private $crawlerObj;
+    /**
+     * @var $crawlerObj tx_crawler_lib
+     */
+    private $crawlerObj;
 
-	/**
-	 * @var $queueRepository tx_crawler_domain_queue_repository
-	 */
-	private $queueRepository;
+    /**
+     * @var $queueRepository tx_crawler_domain_queue_repository
+     */
+    private $queueRepository;
 
-	/**
-	 * @var tx_crawler_domain_process_repository
-	 */
-	private $processRepository;
+    /**
+     * @var tx_crawler_domain_process_repository
+     */
+    private $processRepository;
 
-	/**
-	 * @var $verbose boolean
-	 */
-	private $verbose;
+    /**
+     * @var $verbose boolean
+     */
+    private $verbose;
 
-	/**
-	 * the constructor
-	 */
-	public function __construct() {
-		$this->processRepository	= new tx_crawler_domain_process_repository();
-		$this->queueRepository	= new tx_crawler_domain_queue_repository();
-		$this->crawlerObj = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('tx_crawler_lib');
-		$this->timeToLive = intval($this->crawlerObj->extensionSettings['processMaxRunTime']);
-		$this->countInARun = intval($this->crawlerObj->extensionSettings['countInARun']);
-		$this->processLimit = intval($this->crawlerObj->extensionSettings['processLimit']);
-		$this->verbose = intval($this->crawlerObj->extensionSettings['processVerbose']);
-	}
+    /**
+     * the constructor
+     */
+    public function __construct() {
+        $this->processRepository	= new tx_crawler_domain_process_repository();
+        $this->queueRepository	= new tx_crawler_domain_queue_repository();
+        $this->crawlerObj = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('tx_crawler_lib');
+        $this->timeToLive = intval($this->crawlerObj->extensionSettings['processMaxRunTime']);
+        $this->countInARun = intval($this->crawlerObj->extensionSettings['countInARun']);
+        $this->processLimit = intval($this->crawlerObj->extensionSettings['processLimit']);
+        $this->verbose = intval($this->crawlerObj->extensionSettings['processVerbose']);
+    }
 
-	/**
-	 * starts multiple processes
-	 *
-	 * @param integer $timeout
-	 */
-	public function multiProcess( $timeout ) {
+    /**
+     * starts multiple processes
+     *
+     * @param integer $timeout
+     */
+    public function multiProcess( $timeout ) {
 
-		if ($this->processLimit <= 1) {
-			throw new RuntimeException('To run crawler in multi process mode you have to configure the processLimit > 1.' . PHP_EOL);
-		}
+        if ($this->processLimit <= 1) {
+            throw new RuntimeException('To run crawler in multi process mode you have to configure the processLimit > 1.' . PHP_EOL);
+        }
 
-		$pendingItemsStart = $this->queueRepository->countAllPendingItems();
-		$itemReportLimit = 20;
-		$reportItemCount = 	$pendingItemsStart -  $itemReportLimit;
-		if ($this->verbose) {
-			$this->reportItemStatus();
-		}
-		$this->startRequiredProcesses();
-		$nextTimeOut = time() + $this->timeToLive;
-		for ($i=0; $i<$timeout; $i++) {
-			$currentPendingItems = $this->queueRepository->countAllPendingItems();
-			if ($this->startRequiredProcesses($this->verbose)) {
-				$nextTimeOut = time() + $this->timeToLive;
-			}
-			if ($currentPendingItems == 0) {
-				if ($this->verbose) {
-					echo 'Finished...'.chr(10);
-				}
-				break;
-			}
-			if ($currentPendingItems < $reportItemCount) {
-				if ($this->verbose) {
-					$this->reportItemStatus();
-				}
-				$reportItemCount = $currentPendingItems -  $itemReportLimit;
-			}
-			sleep(1);
-			if ($nextTimeOut < time()) {
-				$timedOutProcesses = $this->processRepository->findAll('','DESC',NULL,0,'ttl >'.$nextTimeOut);
-				$nextTimeOut = time() + $this->timeToLive;
-				if ($this->verbose) {
-					echo 'Cleanup'.implode(',',$timedOutProcesses->getProcessIds()).chr(10);
-				}
-				$this->crawlerObj->CLI_releaseProcesses($timedOutProcesses->getProcessIds(),true);
-			}
-		}
-		if ($currentPendingItems > 0 && $this->verbose) {
-			echo 'Stop with timeout'.chr(10);
-		}
-	}
+        $pendingItemsStart = $this->queueRepository->countAllPendingItems();
+        $itemReportLimit = 20;
+        $reportItemCount = 	$pendingItemsStart -  $itemReportLimit;
+        if ($this->verbose) {
+            $this->reportItemStatus();
+        }
+        $this->startRequiredProcesses();
+        $nextTimeOut = time() + $this->timeToLive;
+        for ($i=0; $i<$timeout; $i++) {
+            $currentPendingItems = $this->queueRepository->countAllPendingItems();
+            if ($this->startRequiredProcesses($this->verbose)) {
+                $nextTimeOut = time() + $this->timeToLive;
+            }
+            if ($currentPendingItems == 0) {
+                if ($this->verbose) {
+                    echo 'Finished...'.chr(10);
+                }
+                break;
+            }
+            if ($currentPendingItems < $reportItemCount) {
+                if ($this->verbose) {
+                    $this->reportItemStatus();
+                }
+                $reportItemCount = $currentPendingItems -  $itemReportLimit;
+            }
+            sleep(1);
+            if ($nextTimeOut < time()) {
+                $timedOutProcesses = $this->processRepository->findAll('','DESC',NULL,0,'ttl >'.$nextTimeOut);
+                $nextTimeOut = time() + $this->timeToLive;
+                if ($this->verbose) {
+                    echo 'Cleanup'.implode(',',$timedOutProcesses->getProcessIds()).chr(10);
+                }
+                $this->crawlerObj->CLI_releaseProcesses($timedOutProcesses->getProcessIds(),true);
+            }
+        }
+        if ($currentPendingItems > 0 && $this->verbose) {
+            echo 'Stop with timeout'.chr(10);
+        }
+    }
 
-	/**
-	 * Reports curent Status of queue
-	 */
-	protected function reportItemStatus() {
-		echo 'Pending:'.$this->queueRepository->countAllPendingItems().' / Assigned:'.$this->queueRepository->countAllAssignedPendingItems().chr(10);
-	}
+    /**
+     * Reports curent Status of queue
+     */
+    protected function reportItemStatus() {
+        echo 'Pending:'.$this->queueRepository->countAllPendingItems().' / Assigned:'.$this->queueRepository->countAllAssignedPendingItems().chr(10);
+    }
 
-	/**
-	 * according to the given count of pending items and the countInARun Setting this method
-	 * starts more crawling processes
-	 * @return boolean if processes are started
-	 */
-	private function startRequiredProcesses() {
-		$ret = FALSE;
-		$currentProcesses= $this->processRepository->countActive();
-		$availableProcessesCount = $this->processLimit-$currentProcesses;
-		$requiredProcessesCount = ceil($this->queueRepository->countAllUnassignedPendingItems() / $this->countInARun);
-		$startProcessCount = min(array($availableProcessesCount,$requiredProcessesCount));
-		if ($startProcessCount <= 0) {
-			return $ret;
-		}
-		if ($startProcessCount && $this->verbose) {
-			echo 'Start '.$startProcessCount.' new processes (Running:'.$currentProcesses.')';
-		}
-		for($i=0;$i<$startProcessCount;$i++) {
-			usleep(100);
-			if ($this->startProcess()) {
-				if ($this->verbose) {
-					echo '.';
-					$ret = TRUE;
-				}
-			}
-		}
-		if ($this->verbose) {
-			echo chr(10);
-		}
-		return $ret;
-	}
+    /**
+     * according to the given count of pending items and the countInARun Setting this method
+     * starts more crawling processes
+     * @return boolean if processes are started
+     */
+    private function startRequiredProcesses() {
+        $ret = FALSE;
+        $currentProcesses= $this->processRepository->countActive();
+        $availableProcessesCount = $this->processLimit-$currentProcesses;
+        $requiredProcessesCount = ceil($this->queueRepository->countAllUnassignedPendingItems() / $this->countInARun);
+        $startProcessCount = min(array($availableProcessesCount,$requiredProcessesCount));
+        if ($startProcessCount <= 0) {
+            return $ret;
+        }
+        if ($startProcessCount && $this->verbose) {
+            echo 'Start '.$startProcessCount.' new processes (Running:'.$currentProcesses.')';
+        }
+        for($i=0;$i<$startProcessCount;$i++) {
+            usleep(100);
+            if ($this->startProcess()) {
+                if ($this->verbose) {
+                    echo '.';
+                    $ret = TRUE;
+                }
+            }
+        }
+        if ($this->verbose) {
+            echo chr(10);
+        }
+        return $ret;
+    }
 
-	/**
-	 * starts new process
-	 * @throws Exception if no crawlerprocess was started
-	 */
-	public function startProcess() {
-		$ttl = (time() + $this->timeToLive -1);
-		$current = $this->processRepository->countNotTimeouted($ttl);
-		$completePath = '(' .escapeshellcmd($this->getCrawlerCliPath()) . ' &) > /dev/null';
-		if (system($completePath) === FALSE) {
-			throw new Exception('could not start process!');
-		}
-		else {
-			for ($i=0;$i<10;$i++) {
-				if ($this->processRepository->countNotTimeouted($ttl) > $current) {
-					return true;
-				}
-				sleep(1);
-			}
-			throw new Exception('Something went wrong: process did not appear within 10 seconds.');
-		}
-	}
+    /**
+     * starts new process
+     * @throws Exception if no crawlerprocess was started
+     */
+    public function startProcess() {
+        $ttl = (time() + $this->timeToLive -1);
+        $current = $this->processRepository->countNotTimeouted($ttl);
+        $completePath = '(' .escapeshellcmd($this->getCrawlerCliPath()) . ' &) > /dev/null';
+        if (system($completePath) === FALSE) {
+            throw new Exception('could not start process!');
+        }
+        else {
+            for ($i=0;$i<10;$i++) {
+                if ($this->processRepository->countNotTimeouted($ttl) > $current) {
+                    return true;
+                }
+                sleep(1);
+            }
+            throw new Exception('Something went wrong: process did not appear within 10 seconds.');
+        }
+    }
 
-	/**
-	 * Returns the path to start the crawler from the command line
-	 *
-	 * @return string
-	 */
-	public function getCrawlerCliPath(){
-		$phpPath 		= $this->crawlerObj->extensionSettings['phpPath'] . ' ';
-		$pathToTypo3 	= rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_DOCUMENT_ROOT'), '/');
-		$pathToTypo3 	.= rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_SITE_PATH'), '/');
-		$cliPart	 	= '/typo3/cli_dispatch.phpsh crawler';
-		return $phpPath.$pathToTypo3.$cliPart;
-	}
+    /**
+     * Returns the path to start the crawler from the command line
+     *
+     * @return string
+     */
+    public function getCrawlerCliPath(){
+        $phpPath 		= $this->crawlerObj->extensionSettings['phpPath'] . ' ';
+        $pathToTypo3 	= rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_DOCUMENT_ROOT'), '/');
+        $pathToTypo3 	.= rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_SITE_PATH'), '/');
+        $cliPart	 	= '/typo3/cli_dispatch.phpsh crawler';
+        return $phpPath.$pathToTypo3.$cliPart;
+    }
 
 }

Spacing +22 added lines, -22 removed lines

@@ -26,7 +26,7 @@ 
  * Manages cralwer processes and can be used to start a new process or multiple processes
  *
  */
-class tx_crawler_domain_process_manager  {
+class tx_crawler_domain_process_manager {
 	/**
 	 * @var $timeToLive integer
 	 */
@@ -65,8 +65,8 @@ 
 	 * the constructor
 	 */
 	public function __construct() {
-		$this->processRepository	= new tx_crawler_domain_process_repository();
-		$this->queueRepository	= new tx_crawler_domain_queue_repository();
+		$this->processRepository = new tx_crawler_domain_process_repository();
+		$this->queueRepository = new tx_crawler_domain_queue_repository();
 		$this->crawlerObj = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('tx_crawler_lib');
 		$this->timeToLive = intval($this->crawlerObj->extensionSettings['processMaxRunTime']);
 		$this->countInARun = intval($this->crawlerObj->extensionSettings['countInARun']);
@@ -79,21 +79,21 @@ 
 	 *
 	 * @param integer $timeout
 	 */
-	public function multiProcess( $timeout ) {
+	public function multiProcess($timeout) {
 
 		if ($this->processLimit <= 1) {
-			throw new RuntimeException('To run crawler in multi process mode you have to configure the processLimit > 1.' . PHP_EOL);
+			throw new RuntimeException('To run crawler in multi process mode you have to configure the processLimit > 1.'.PHP_EOL);
 		}
 
 		$pendingItemsStart = $this->queueRepository->countAllPendingItems();
 		$itemReportLimit = 20;
-		$reportItemCount = 	$pendingItemsStart -  $itemReportLimit;
+		$reportItemCount = $pendingItemsStart - $itemReportLimit;
 		if ($this->verbose) {
 			$this->reportItemStatus();
 		}
 		$this->startRequiredProcesses();
 		$nextTimeOut = time() + $this->timeToLive;
-		for ($i=0; $i<$timeout; $i++) {
+		for ($i = 0; $i < $timeout; $i++) {
 			$currentPendingItems = $this->queueRepository->countAllPendingItems();
 			if ($this->startRequiredProcesses($this->verbose)) {
 				$nextTimeOut = time() + $this->timeToLive;
@@ -108,16 +108,16 @@ 
 				if ($this->verbose) {
 					$this->reportItemStatus();
 				}
-				$reportItemCount = $currentPendingItems -  $itemReportLimit;
+				$reportItemCount = $currentPendingItems - $itemReportLimit;
 			}
 			sleep(1);
 			if ($nextTimeOut < time()) {
-				$timedOutProcesses = $this->processRepository->findAll('','DESC',NULL,0,'ttl >'.$nextTimeOut);
+				$timedOutProcesses = $this->processRepository->findAll('', 'DESC', NULL, 0, 'ttl >'.$nextTimeOut);
 				$nextTimeOut = time() + $this->timeToLive;
 				if ($this->verbose) {
-					echo 'Cleanup'.implode(',',$timedOutProcesses->getProcessIds()).chr(10);
+					echo 'Cleanup'.implode(',', $timedOutProcesses->getProcessIds()).chr(10);
 				}
-				$this->crawlerObj->CLI_releaseProcesses($timedOutProcesses->getProcessIds(),true);
+				$this->crawlerObj->CLI_releaseProcesses($timedOutProcesses->getProcessIds(), true);
 			}
 		}
 		if ($currentPendingItems > 0 && $this->verbose) {
@@ -139,17 +139,17 @@ 
 	 */
 	private function startRequiredProcesses() {
 		$ret = FALSE;
-		$currentProcesses= $this->processRepository->countActive();
-		$availableProcessesCount = $this->processLimit-$currentProcesses;
+		$currentProcesses = $this->processRepository->countActive();
+		$availableProcessesCount = $this->processLimit - $currentProcesses;
 		$requiredProcessesCount = ceil($this->queueRepository->countAllUnassignedPendingItems() / $this->countInARun);
-		$startProcessCount = min(array($availableProcessesCount,$requiredProcessesCount));
+		$startProcessCount = min(array($availableProcessesCount, $requiredProcessesCount));
 		if ($startProcessCount <= 0) {
 			return $ret;
 		}
 		if ($startProcessCount && $this->verbose) {
 			echo 'Start '.$startProcessCount.' new processes (Running:'.$currentProcesses.')';
 		}
-		for($i=0;$i<$startProcessCount;$i++) {
+		for ($i = 0; $i < $startProcessCount; $i++) {
 			usleep(100);
 			if ($this->startProcess()) {
 				if ($this->verbose) {
@@ -169,14 +169,14 @@ 
 	 * @throws Exception if no crawlerprocess was started
 	 */
 	public function startProcess() {
-		$ttl = (time() + $this->timeToLive -1);
+		$ttl = (time() + $this->timeToLive - 1);
 		$current = $this->processRepository->countNotTimeouted($ttl);
-		$completePath = '(' .escapeshellcmd($this->getCrawlerCliPath()) . ' &) > /dev/null';
+		$completePath = '('.escapeshellcmd($this->getCrawlerCliPath()).' &) > /dev/null';
 		if (system($completePath) === FALSE) {
 			throw new Exception('could not start process!');
 		}
 		else {
-			for ($i=0;$i<10;$i++) {
+			for ($i = 0; $i < 10; $i++) {
 				if ($this->processRepository->countNotTimeouted($ttl) > $current) {
 					return true;
 				}
@@ -191,10 +191,10 @@ 
 	 *
 	 * @return string
 	 */
-	public function getCrawlerCliPath(){
-		$phpPath 		= $this->crawlerObj->extensionSettings['phpPath'] . ' ';
-		$pathToTypo3 	= rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_DOCUMENT_ROOT'), '/');
-		$pathToTypo3 	.= rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_SITE_PATH'), '/');
+	public function getCrawlerCliPath() {
+		$phpPath 		= $this->crawlerObj->extensionSettings['phpPath'].' ';
+		$pathToTypo3 = rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_DOCUMENT_ROOT'), '/');
+		$pathToTypo3 .= rtrim(\TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv('TYPO3_SITE_PATH'), '/');
 		$cliPart	 	= '/typo3/cli_dispatch.phpsh crawler';
 		return $phpPath.$pathToTypo3.$cliPart;
 	}

Braces +1 added lines, -2 removed lines

@@ -174,8 +174,7 @@
 		$completePath = '(' .escapeshellcmd($this->getCrawlerCliPath()) . ' &) > /dev/null';
 		if (system($completePath) === FALSE) {
 			throw new Exception('could not start process!');
-		}
-		else {
+		} else {
 			for ($i=0;$i<10;$i++) {
 				if ($this->processRepository->countNotTimeouted($ttl) > $current) {
 					return true;

domain/process/class.tx_crawler_domain_process.php

Indentation +134 added lines, -134 removed lines

@@ -24,140 +24,140 @@
 
 class tx_crawler_domain_process extends tx_crawler_domain_lib_abstract_dbobject {
 
-	/**
-	 * @var string table name
-	 */
-	protected static $tableName = 'tx_crawler_process';
-
-	/**
-	 * Returns the activity state for this process
-	 *
-	 * @param void
-	 * @return boolean
-	 */
-	public function getActive() {
-		return $this->row['active'];
-	}
-
-	/**
-	 * Returns the identifier for the process
-	 *
-	 * @return string
-	 */
-	public function getProcess_id() {
-		return $this->row['process_id'];
-	}
-
-	/**
-	 * Returns the timestamp of the exectime for the first relevant queue item.
-	 * This can be used to determine the runtime
-	 *
-	 * @return int
-	 */
-	public function getTimeForFirstItem() {
-		$queueRepository = new tx_crawler_domain_queue_repository();
-		$entry = $queueRepository->findYoungestEntryForProcess($this);
-
-		return $entry->getExecutionTime();
-	}
-
-	/**
-	 * Returns the timestamp of the exectime for the last relevant queue item.
-	 * This can be used to determine the runtime
-	 *
-	 * @return int
-	 */
-	public function getTimeForLastItem() {
-		$queueRepository = new tx_crawler_domain_queue_repository();
-		$entry = $queueRepository->findOldestEntryForProcess($this);
-
-		return $entry->getExecutionTime();
-	}
-
-	/**
-	 * Returns the difference between first and last processed item
-	 *
-	 * @return int
-	 */
-	public function getRuntime() {
-		return $this->getTimeForLastItem() - $this->getTimeForFirstItem();
-	}
-
-	/**
-	 * Returns the ttl of the process
-	 *
-	 * @return int
-	 */
-	public function getTTL() {
-		return $this->row['ttl'];
-	}
-
-	/**
-	 * Counts the number of items which need to be processed
-	 *
-	 * @author Timo Schmidt <timo.schmidt@aoemedia.de>
-	 * @param void
-	 * @return int
-	 */
-	public function countItemsProcessed() {
-		$queueRepository = new tx_crawler_domain_queue_repository();
-		return $queueRepository->countExtecutedItemsByProcess($this);
-	}
-
-	/**
-	 * Counts the number of items which still need to be processed
-	 *
-	 * @author Timo Schmidt <timo.schmidt@aoemedia.de>
-	 * @param void
-	 * @return int
-	 */
-	public function countItemsToProcess() {
-		$queueRepository = new tx_crawler_domain_queue_repository();
-		return $queueRepository->countNonExecutedItemsByProcess($this);
-	}
-
-	/**
-	 * Returns the Progress of a crawling process as a percentage value
-	 *
-	 * @param void
-	 * @return float
-	 */
-	public function getProgress() {
-		$all = $this->countItemsAssigned();
-		if ($all > 0) {
-			$res = round((100 / $all) * $this->countItemsProcessed());
-		} else {
-			$res = 0;
-		}
-		return $res;
-	}
-
-	/**
-	 * Returns the number of assigned Entrys
-	 *
-	 * @return int
-	 */
-	public function countItemsAssigned() {
-		return $this->row['assigned_items_count'];
-	}
-
-	/**
-	 * Return the processes current state
-	 *
-	 * @param void
-	 * @return string 'running'|'cancelled'|'completed'
-	 */
-	public function getState() {
-		// TODO: use class constants for these states
-		if ($this->getActive() && $this->getProgress() < 100) {
-			$stage = 'running';
-		} elseif(!$this->getActive() && $this->getProgress() < 100) {
-			$stage = 'cancelled';
-		} else {
-			$stage = 'completed';
-		}
-		return $stage;
-	}
+    /**
+     * @var string table name
+     */
+    protected static $tableName = 'tx_crawler_process';
+
+    /**
+     * Returns the activity state for this process
+     *
+     * @param void
+     * @return boolean
+     */
+    public function getActive() {
+        return $this->row['active'];
+    }
+
+    /**
+     * Returns the identifier for the process
+     *
+     * @return string
+     */
+    public function getProcess_id() {
+        return $this->row['process_id'];
+    }
+
+    /**
+     * Returns the timestamp of the exectime for the first relevant queue item.
+     * This can be used to determine the runtime
+     *
+     * @return int
+     */
+    public function getTimeForFirstItem() {
+        $queueRepository = new tx_crawler_domain_queue_repository();
+        $entry = $queueRepository->findYoungestEntryForProcess($this);
+
+        return $entry->getExecutionTime();
+    }
+
+    /**
+     * Returns the timestamp of the exectime for the last relevant queue item.
+     * This can be used to determine the runtime
+     *
+     * @return int
+     */
+    public function getTimeForLastItem() {
+        $queueRepository = new tx_crawler_domain_queue_repository();
+        $entry = $queueRepository->findOldestEntryForProcess($this);
+
+        return $entry->getExecutionTime();
+    }
+
+    /**
+     * Returns the difference between first and last processed item
+     *
+     * @return int
+     */
+    public function getRuntime() {
+        return $this->getTimeForLastItem() - $this->getTimeForFirstItem();
+    }
+
+    /**
+     * Returns the ttl of the process
+     *
+     * @return int
+     */
+    public function getTTL() {
+        return $this->row['ttl'];
+    }
+
+    /**
+     * Counts the number of items which need to be processed
+     *
+     * @author Timo Schmidt <timo.schmidt@aoemedia.de>
+     * @param void
+     * @return int
+     */
+    public function countItemsProcessed() {
+        $queueRepository = new tx_crawler_domain_queue_repository();
+        return $queueRepository->countExtecutedItemsByProcess($this);
+    }
+
+    /**
+     * Counts the number of items which still need to be processed
+     *
+     * @author Timo Schmidt <timo.schmidt@aoemedia.de>
+     * @param void
+     * @return int
+     */
+    public function countItemsToProcess() {
+        $queueRepository = new tx_crawler_domain_queue_repository();
+        return $queueRepository->countNonExecutedItemsByProcess($this);
+    }
+
+    /**
+     * Returns the Progress of a crawling process as a percentage value
+     *
+     * @param void
+     * @return float
+     */
+    public function getProgress() {
+        $all = $this->countItemsAssigned();
+        if ($all > 0) {
+            $res = round((100 / $all) * $this->countItemsProcessed());
+        } else {
+            $res = 0;
+        }
+        return $res;
+    }
+
+    /**
+     * Returns the number of assigned Entrys
+     *
+     * @return int
+     */
+    public function countItemsAssigned() {
+        return $this->row['assigned_items_count'];
+    }
+
+    /**
+     * Return the processes current state
+     *
+     * @param void
+     * @return string 'running'|'cancelled'|'completed'
+     */
+    public function getState() {
+        // TODO: use class constants for these states
+        if ($this->getActive() && $this->getProgress() < 100) {
+            $stage = 'running';
+        } elseif(!$this->getActive() && $this->getProgress() < 100) {
+            $stage = 'cancelled';
+        } else {
+            $stage = 'completed';
+        }
+        return $stage;
+    }
 }
 
 ?>

Spacing +1 added lines, -1 removed lines

@@ -151,7 +151,7 @@
 		// TODO: use class constants for these states
 		if ($this->getActive() && $this->getProgress() < 100) {
 			$stage = 'running';
-		} elseif(!$this->getActive() && $this->getProgress() < 100) {
+		} elseif (!$this->getActive() && $this->getProgress() < 100) {
 			$stage = 'cancelled';
 		} else {
 			$stage = 'completed';