AOEpeople / crawler

Classes/Hooks/StaticFileCacheCreateUriHook.php

Spacing +3 added lines, -3 removed lines

@@ -66,10 +66,10 @@
         if ($this->isCrawlerExtensionRunning($frontend) && preg_match('#^/index.php\?&?id=(\d+)(&.*)?$#', $uri, $matches)) {
             $speakingUri = $frontend->cObj->typoLink_URL(array('parameter' => $matches[1], 'additionalParams' => $matches[2]));
             $speakingUriParts = parse_url($speakingUri);
-            if(false === $speakingUriParts){
-                throw new \Exception('Could not parse URI: ' . $speakingUri, 1289915976);
+            if (false === $speakingUriParts) {
+                throw new \Exception('Could not parse URI: '.$speakingUri, 1289915976);
             }
-            $speakingUrlPath = '/' . ltrim($speakingUriParts['path'], '/');
+            $speakingUrlPath = '/'.ltrim($speakingUriParts['path'], '/');
             // Don't change anything if speaking URL is part of old URI:
             // (it might be the case the using the speaking URL failed)
             if (strpos($uri, $speakingUrlPath) !== 0 || $speakingUrlPath === '/') {

Classes/Task/CrawlerQueueTask.php

Spacing +2 added lines, -2 removed lines

@@ -92,7 +92,7 @@ 
             $this->startPage = 0;
         }
 
-        $_SERVER['argv'] = array($_SERVER['argv'][0], $this->startPage,'-ss', '-d', $this->depth, '-o', self::MODE, '-conf', implode(',', $this->configuration));
+        $_SERVER['argv'] = array($_SERVER['argv'][0], $this->startPage, '-ss', '-d', $this->depth, '-o', self::MODE, '-conf', implode(',', $this->configuration));
     }
 
 
@@ -109,6 +109,6 @@ 
             $this->startPage = 0;
         }
 
-        return implode(',', $this->configuration) . ' (depth: ' . $this->depth . ', startPage:' . $this->startPage . ')';
+        return implode(',', $this->configuration).' (depth: '.$this->depth.', startPage:'.$this->startPage.')';
     }
 }
\ No newline at end of file

domain/queue/class.tx_crawler_domain_queue_repository.php

Spacing +14 added lines, -14 removed lines

@@ -37,7 +37,7 @@ 
 	 * @return tx_crawler_domain_queue_entry $entry
 	 */
 	public function findYoungestEntryForProcess(tx_crawler_domain_process $process) {
-		return $this->getFirstOrLastObjectByProcess($process,'exec_time ASC');
+		return $this->getFirstOrLastObjectByProcess($process, 'exec_time ASC');
 	}
 
 	/**
@@ -47,7 +47,7 @@ 
 	 * @return tx_crawler_domain_queue_entry
 	 */
 	public function findOldestEntryForProcess(tx_crawler_domain_process $process) {
-		return $this->getFirstOrLastObjectByProcess($process,'exec_time DESC');
+		return $this->getFirstOrLastObjectByProcess($process, 'exec_time DESC');
 	}
 
 
@@ -60,15 +60,15 @@ 
 	 */
 	protected function getFirstOrLastObjectByProcess($process, $orderby) {
 		$db = $this->getDB();
-		$where = 'process_id_completed='.$db->fullQuoteStr($process->getProcess_id(),$this->tableName).
+		$where = 'process_id_completed='.$db->fullQuoteStr($process->getProcess_id(), $this->tableName).
 				 ' AND exec_time > 0 ';
 		$limit = 1;
 		$groupby = '';
 
-		$res 	= $db->exec_SELECTgetRows('*','tx_crawler_queue',$where,$groupby,$orderby,$limit);
-		if($res) {
-			$first 	= $res[0];
-		}else{
+		$res = $db->exec_SELECTgetRows('*', 'tx_crawler_queue', $where, $groupby, $orderby, $limit);
+		if ($res) {
+			$first = $res[0];
+		} else {
 			$first = array();
 		}
 		$resultObject = new tx_crawler_domain_queue_entry($first);
@@ -83,7 +83,7 @@ 
 	 */
 	public function countExtecutedItemsByProcess($process) {
 
-		return $this->countItemsByWhereClause('exec_time > 0 AND process_id_completed = '.$this->getDB()->fullQuoteStr($process->getProcess_id(),$this->tableName));
+		return $this->countItemsByWhereClause('exec_time > 0 AND process_id_completed = '.$this->getDB()->fullQuoteStr($process->getProcess_id(), $this->tableName));
 	}
 
 	/**
@@ -93,7 +93,7 @@ 
 	 * @return int
 	 */
 	public function countNonExecutedItemsByProcess($process) {
-		return $this->countItemsByWhereClause('exec_time = 0 AND process_id = '.$this->getDB()->fullQuoteStr($process->getProcess_id(),$this->tableName));
+		return $this->countItemsByWhereClause('exec_time = 0 AND process_id = '.$this->getDB()->fullQuoteStr($process->getProcess_id(), $this->tableName));
 	}
 
 	/**
@@ -135,8 +135,8 @@ 
 	 */
 	protected function countItemsByWhereClause($where) {
 		$db 	= $this->getDB();
-		$rs 	= $db->exec_SELECTquery('count(*) as anz',$this->tableName,$where);
-		$res 	= $db->sql_fetch_assoc($rs);
+		$rs 	= $db->exec_SELECTquery('count(*) as anz', $this->tableName, $where);
+		$res = $db->sql_fetch_assoc($rs);
 
 		return $res['anz'];
 	}
@@ -202,7 +202,7 @@ 
 			$res = $db->exec_SELECTquery(
 				'configuration, count(*) as c',
 				$this->tableName,
-				'set_id in ('. implode(',',$setIds).') AND scheduled < '.time(),
+				'set_id in ('.implode(',', $setIds).') AND scheduled < '.time(),
 				'configuration'
 			);
 			while ($row = $db->sql_fetch_assoc($res)) {
@@ -247,7 +247,7 @@ 
 	 * @author Fabrizio Branca <fabrizio.branca@aoemedia.de>
 	 * @since 2010-11-16
 	 */
-	public function getLastProcessedEntries($selectFields='*', $limit='100') {
+	public function getLastProcessedEntries($selectFields = '*', $limit = '100') {
 		$db = $this->getDB();
 		$res = $db->exec_SELECTquery(
 			$selectFields,
@@ -279,7 +279,7 @@ 
 		$res = $db->exec_SELECTquery(
 			'process_id_completed, min(exec_time) as start, max(exec_time) as end, count(*) as urlcount',
 			$this->tableName,
-			'exec_time != 0 and exec_time >= '.intval($start). ' and exec_time <= ' . intval($end),
+			'exec_time != 0 and exec_time >= '.intval($start).' and exec_time <= '.intval($end),
 			'process_id_completed'
 		);
 

domain/queue/class.tx_crawler_domain_queue_entry.php

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/process/class.tx_crawler_domain_process_repository.php

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.php

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';

Classes/Task/CrawlMultiProcessTask.php

Spacing +1 added lines, -1 removed lines

@@ -47,7 +47,7 @@
     public function execute()
     {
         $processManager = new \tx_crawler_domain_process_manager();
-        $timeout = is_int($this->timeOut) ? (int)$this->timeOut : 1800;
+        $timeout = is_int($this->timeOut) ? (int) $this->timeOut : 1800;
 
         try {
             $processManager->multiProcess($timeout);

cli/crawler_multiprocess.php

Spacing +2 added lines, -2 removed lines

@@ -4,10 +4,10 @@
 }
 
 $processManager = new tx_crawler_domain_process_manager();
-$timeout = isset($_SERVER['argv'][1] ) ? intval($_SERVER['argv'][1]) : 1800;
+$timeout = isset($_SERVER['argv'][1]) ? intval($_SERVER['argv'][1]) : 1800;
 
 try {
 	$processManager->multiProcess($timeout);
 } catch (Exception $e) {
-	echo PHP_EOL . $e->getMessage();
+	echo PHP_EOL.$e->getMessage();
 }

domain/process/class.tx_crawler_domain_process_manager.php

Spacing +19 added lines, -19 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']);
@@ -81,7 +81,7 @@ 
 	 */
 	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 = $this->queueRepository->countAllPendingItems();
@@ -99,7 +99,7 @@ 
 			}
 			if ($currentPendingItems == 0) {
 				if ($this->verbose) {
-					echo 'Finished...' . chr(10);
+					echo 'Finished...'.chr(10);
 				}
 				break;
 			}
@@ -111,16 +111,16 @@ 
 			}
 			sleep(1);
 			if ($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);
 			}
 		}
 		if ($currentPendingItems > 0 && $this->verbose) {
-			echo 'Stop with timeout' . chr(10);
+			echo 'Stop with timeout'.chr(10);
 		}
 	}
 
@@ -138,17 +138,17 @@ 
 	 */
 	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) {
@@ -168,14 +168,14 @@ 
 	 * @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;
 				}
@@ -190,10 +190,10 @@ 
 	 *
 	 * @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;
 	}