AOEpeople / crawler

domain/process/class.tx_crawler_domain_process_repository.php

Spacing +3 added lines, -3 removed lines

@@ -66,7 +66,7 @@ 
             $this->tableName,
             $where,
             '',
-            htmlspecialchars($orderField) . ' ' . htmlspecialchars($orderDirection),
+            htmlspecialchars($orderField).' '.htmlspecialchars($orderDirection),
             self::getLimitFromItemCountAndOffset($itemCount, $offset)
         );
 
@@ -111,7 +111,7 @@ 
      */
     public function countNotTimeouted($ttl)
     {
-        return $this->countByWhere('deleted = 0 AND ttl > ' . intval($ttl));
+        return $this->countByWhere('deleted = 0 AND ttl > '.intval($ttl));
     }
 
     /**
@@ -127,7 +127,7 @@ 
     {
         $itemCount = filter_var($itemCount, FILTER_VALIDATE_INT, ['options' => ['min_range' => 1, 'default' => 20]]);
         $offset = filter_var($offset, FILTER_VALIDATE_INT, ['options' => ['min_range' => 0, 'default' => 0]]);
-        $limit = $offset . ', ' . $itemCount;
+        $limit = $offset.', '.$itemCount;
 
         return $limit;
     }

domain/queue/class.tx_crawler_domain_queue_repository.php

Indentation +1 added lines, -1 removed lines

@@ -66,7 +66,7 @@
     {
         $db = $this->getDB();
         $where = 'process_id_completed='.$db->fullQuoteStr($process->getProcess_id(), $this->tableName).
-                 ' AND exec_time > 0 ';
+                    ' AND exec_time > 0 ';
         $limit = 1;
         $groupby = '';
 

Spacing +6 added lines, -6 removed lines

@@ -67,9 +67,9 @@ 
         $limit = 1;
         $groupby = '';
 
-        $res 	= $db->exec_SELECTgetRows('*', 'tx_crawler_queue', $where, $groupby, $orderby, $limit);
+        $res = $db->exec_SELECTgetRows('*', 'tx_crawler_queue', $where, $groupby, $orderby, $limit);
         if ($res) {
-            $first 	= $res[0];
+            $first = $res[0];
         } else {
             $first = [];
         }
@@ -143,7 +143,7 @@ 
     {
         $db 	= $this->getDB();
         $rs 	= $db->exec_SELECTquery('count(*) as anz', $this->tableName, $where);
-        $res 	= $db->sql_fetch_assoc($rs);
+        $res = $db->sql_fetch_assoc($rs);
 
         return $res['anz'];
     }
@@ -212,7 +212,7 @@ 
             $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)) {
@@ -258,7 +258,7 @@ 
      * @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(
@@ -292,7 +292,7 @@ 
         $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'
         );
 

cli/bootstrap.php

Spacing +10 added lines, -10 removed lines

@@ -12,12 +12,12 @@ 
 if (!isAbsPath($tempPathThisScript)) {
     $workingDirectory = $_SERVER['PWD'] ? $_SERVER['PWD'] : getcwd();
     if ($workingDirectory) {
-        $tempPathThisScript = $workingDirectory . '/' . preg_replace('/\.\//', '', $tempPathThisScript);
+        $tempPathThisScript = $workingDirectory.'/'.preg_replace('/\.\//', '', $tempPathThisScript);
         if (!@is_file($tempPathThisScript)) {
-            die('Relative path found, but an error occured during resolving of the absolute path: ' . $tempPathThisScript . PHP_EOL);
+            die('Relative path found, but an error occured during resolving of the absolute path: '.$tempPathThisScript.PHP_EOL);
         }
     } else {
-        die('Relative path found, but resolving absolute path is not supported on this platform.' . PHP_EOL);
+        die('Relative path found, but resolving absolute path is not supported on this platform.'.PHP_EOL);
     }
 }
 
@@ -35,7 +35,7 @@ 
             list($key, $value) = explode(':', $additionalHeader, 2);
             $key = str_replace('-', '_', strtoupper(trim($key)));
             if ($key != 'HOST') {
-                $_SERVER['HTTP_' . $key] = $value;
+                $_SERVER['HTTP_'.$key] = $value;
             }
         }
     }
@@ -57,18 +57,18 @@ 
 $typo3SitePath = $_SERVER['argv'][1];
 
     // faking the environment
-$_SERVER['DOCUMENT_ROOT'] = preg_replace('#' . preg_quote($typo3SitePath, '#') . '$#', '', $typo3Root);
+$_SERVER['DOCUMENT_ROOT'] = preg_replace('#'.preg_quote($typo3SitePath, '#').'$#', '', $typo3Root);
 $_SERVER['HTTP_USER_AGENT'] = 'CLI Mode';
 $_SERVER['HTTP_HOST'] = $_SERVER['SERVER_NAME'] = $urlParts['host'];
-$_SERVER['SCRIPT_NAME'] = $_SERVER['PHP_SELF'] = $typo3SitePath . 'index.php';
-$_SERVER['SCRIPT_FILENAME'] = $_SERVER['PATH_TRANSLATED'] = $typo3Root . 'index.php';
+$_SERVER['SCRIPT_NAME'] = $_SERVER['PHP_SELF'] = $typo3SitePath.'index.php';
+$_SERVER['SCRIPT_FILENAME'] = $_SERVER['PATH_TRANSLATED'] = $typo3Root.'index.php';
 $_SERVER['QUERY_STRING'] = (isset($urlParts['query']) ? $urlParts['query'] : '');
-$_SERVER['REQUEST_URI'] = $urlParts['path'] . (isset($urlParts['query']) ? '?' . $urlParts['query'] : '');
+$_SERVER['REQUEST_URI'] = $urlParts['path'].(isset($urlParts['query']) ? '?'.$urlParts['query'] : '');
 $_SERVER['REQUEST_METHOD'] = 'GET';
 
     // Define a port if used in the URL:
 if (isset($urlParts['port'])) {
-    $_SERVER['HTTP_HOST'] .= ':' . $urlParts['port'];
+    $_SERVER['HTTP_HOST'] .= ':'.$urlParts['port'];
     $_SERVER['SERVER_PORT'] = $urlParts['port'];
 }
 
@@ -78,7 +78,7 @@ 
 }
 
 chdir($typo3Root);
-include($typo3Root . '/index.php');
+include($typo3Root.'/index.php');
 
 
 /**

cli/class.tx_crawler_cli_im.php

Indentation +5 added lines, -5 removed lines

@@ -49,11 +49,11 @@
 
         // Adding options to help archive:
         /**
-        * We removed the "proc" option as it seemd not to be working any more. But as the complete handling of the crawler has changed regarding the configuration
-        * this is completely ok. Since configuration records were introduced to configure "what should be done" additionally to page ts the way to setup jobs
-        * has drifted from selecting filtering processing instructions to selecting/filtering configuration keys (you can configure the processing instructions there).
-        * This is also reflected in the backend modules and allows you a much clearer and powerful way to work with the crawler extension.
-        */
+         * We removed the "proc" option as it seemd not to be working any more. But as the complete handling of the crawler has changed regarding the configuration
+         * this is completely ok. Since configuration records were introduced to configure "what should be done" additionally to page ts the way to setup jobs
+         * has drifted from selecting filtering processing instructions to selecting/filtering configuration keys (you can configure the processing instructions there).
+         * This is also reflected in the backend modules and allows you a much clearer and powerful way to work with the crawler extension.
+         */
         // $this->cli_options[] = array('-proc listOfProcInstr', 'Comma list of processing instructions. These are the "actions" carried out when crawling and you must specify at least one. Depends on third-party extensions. Examples are "tx_cachemgm_recache" from "cachemgm" extension (will recache pages), "tx_staticpub_publish" from "staticpub" (publishing pages to static files) or "tx_indexedsearch_reindex" from "indexed_search" (indexes pages).');
         // TODO: cleanup here!
         $this->cli_options[] = ['-d depth', 'Tree depth, 0-99', "How many levels under the 'page_id' to include."];

Spacing +2 added lines, -2 removed lines

@@ -59,7 +59,7 @@ 
         $this->cli_options[] = ['-d depth', 'Tree depth, 0-99', "How many levels under the 'page_id' to include."];
         $this->cli_options[] = ['-o mode', 'Output mode: "url", "exec", "queue"', "Specifies output modes\nurl : Will list URLs which wget could use as input.\nqueue: Will put entries in queue table.\nexec: Will execute all entries right away!"];
         $this->cli_options[] = ['-n number', 'Number of items per minute.', 'Specifies how many items are put in the queue per minute. Only valid for output mode "queue"'];
-        $this->cli_options[] = ['-conf configurationkeys','List of Configuration Keys','A commaseperated list of crawler configurations'];
+        $this->cli_options[] = ['-conf configurationkeys', 'List of Configuration Keys', 'A commaseperated list of crawler configurations'];
         #		$this->cli_options[] = array('-v level', 'Verbosity level 0-3', "The value of level can be:\n  0 = all output\n  1 = info and greater (default)\n  2 = warnings and greater\n  3 = errors");
 
         // Setting help texts:
@@ -67,7 +67,7 @@ 
         $this->cli_help['synopsis'] = 'page_id ###OPTIONS###';
         $this->cli_help['description'] = "Works as a CLI interface to some functionality from the Web > Info > Site Crawler module; It can put entries in the queue from command line options, return the list of URLs and even execute all entries right away without having to queue them up - this can be useful for immediate re-cache, re-indexing or static publishing from command line.";
         $this->cli_help['examples'] = "/.../cli_dispatch.phpsh crawler_im 7 -d=2 -conf=<configurationKey> -o=exec\nWill re-cache pages from page 7 and two levels down, executed immediately.\n";
-        $this->cli_help['examples'].= "/.../cli_dispatch.phpsh crawler_im 7 -d=0 -conf=<configurationKey> -n=4 -o=queue\nWill put entries for re-caching pages from page 7 into queue, 4 every minute.\n";
+        $this->cli_help['examples'] .= "/.../cli_dispatch.phpsh crawler_im 7 -d=0 -conf=<configurationKey> -n=4 -o=queue\nWill put entries for re-caching pages from page 7 into queue, 4 every minute.\n";
         $this->cli_help['author'] = 'Kasper Skaarhoj, Daniel Poetzinger, Fabrizio Branca, Tolleiv Nietsch, Timo Schmidt - AOE media 2009';
     }
 }

cli/crawler_multiprocess.php

Spacing +1 added lines, -1 removed lines

@@ -9,5 +9,5 @@
 try {
     $processManager->multiProcess($timeout);
 } catch (Exception $e) {
-    echo PHP_EOL . $e->getMessage();
+    echo PHP_EOL.$e->getMessage();
 }

template/process/list.php

Spacing +3 added lines, -3 removed lines

@@ -53,16 +53,16 @@
 	</thead>
 	<tbody>
 		<?php foreach ($this->getProcessCollection() as $process): /* @var $process tx_crawler_domain_process */ ?>
-			<tr class="<?php echo (++$count % 2 == 0) ? 'odd': 'even' ?>">
+			<tr class="<?php echo (++$count % 2 == 0) ? 'odd' : 'even' ?>">
 				<td><?php echo $this->getIconForState(htmlspecialchars($process->getState())); ?></td>
 				<td><?php echo htmlspecialchars($process->getProcess_id()); ?></td>
 				<td><?php echo htmlspecialchars($this->asDate($process->getTimeForFirstItem())); ?></td>
 				<td><?php echo htmlspecialchars($this->asDate($process->getTimeForLastItem())); ?></td>
-				<td><?php echo htmlspecialchars(floor($process->getRuntime()/ 60)); ?> min. <?php echo htmlspecialchars($process->getRuntime()) % 60 ?> sec.</td>
+				<td><?php echo htmlspecialchars(floor($process->getRuntime() / 60)); ?> min. <?php echo htmlspecialchars($process->getRuntime()) % 60 ?> sec.</td>
 				<td><?php echo htmlspecialchars($this->asDate($process->getTTL())); ?></td>
 				<td><?php echo htmlspecialchars($process->countItemsProcessed()); ?></td>
 				<td><?php echo htmlspecialchars($process->countItemsAssigned()); ?></td>
-				<td><?php echo htmlspecialchars($process->countItemsToProcess()+$process->countItemsProcessed()); ?></td>
+				<td><?php echo htmlspecialchars($process->countItemsToProcess() + $process->countItemsProcessed()); ?></td>
 				<td>
 				<?php if ($process->getState() == 'running'): ?>
 					<div class="crawlerprocessprogress" style="width: 200px;">

template/pagination.php

Spacing +2 added lines, -2 removed lines

@@ -3,12 +3,12 @@
 } ?>
 
 Page:
-<?php for ($currentPageOffset 	= 0; $currentPageOffset < $this->getTotalPagesCount(); $currentPageOffset++) {
+<?php for ($currentPageOffset = 0; $currentPageOffset < $this->getTotalPagesCount(); $currentPageOffset++) {
     ?>
 	<a href="index.php?offset=<?php echo htmlspecialchars($currentPageOffset * $this->getPerPage()); ?>">
 		<?php echo	htmlspecialchars($this->getLabelForPageOffset($currentPageOffset)); ?>
 	</a>
-	<?php if ($currentPageOffset+1 < $this->getTotalPagesCount()) {
+	<?php if ($currentPageOffset + 1 < $this->getTotalPagesCount()) {
         ?>
 	|
 	<?php

domain/process/class.tx_crawler_domain_process_collection.php

Spacing +6 added lines, -6 removed lines

@@ -50,8 +50,8 @@ 
      */
     public function offsetGet($index)
     {
-        if (! parent::offsetExists($index)) {
-            throw new Exception('Index "' . var_export($index, true) . '" for tx_crawler_domain_process are not available');
+        if (!parent::offsetExists($index)) {
+            throw new Exception('Index "'.var_export($index, true).'" for tx_crawler_domain_process are not available');
         }
         return parent::offsetGet($index);
     }
@@ -66,7 +66,7 @@ 
      */
     public function offsetSet($index, $subject)
     {
-        if (! $subject instanceof tx_crawler_domain_process) {
+        if (!$subject instanceof tx_crawler_domain_process) {
             throw new InvalidArgumentException('Wrong parameter type given, "tx_crawler_domain_process" expected!');
         }
         parent::offsetSet($index, $subject);
@@ -80,7 +80,7 @@ 
      */
     public function append($subject)
     {
-        if (! $subject instanceof tx_crawler_domain_process) {
+        if (!$subject instanceof tx_crawler_domain_process) {
             throw new InvalidArgumentException('Wrong parameter type given, "tx_crawler_domain_process" expected!');
         }
         parent::append($subject);
@@ -92,9 +92,9 @@ 
      */
     public function getProcessIds()
     {
-        $result=[];
+        $result = [];
         foreach ($this->getIterator() as $value) {
-            $result[]=$value->getProcess_id();
+            $result[] = $value->getProcess_id();
         }
         return $result;
     }

domain/process/class.tx_crawler_domain_process_manager.php

Spacing +17 added lines, -17 removed lines

@@ -67,8 +67,8 @@ 
      */
     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']);
@@ -84,7 +84,7 @@ 
     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 = $this->queueRepository->countAllPendingItems();
@@ -102,7 +102,7 @@ 
             }
             if ($currentPendingItems == 0) {
                 if ($this->verbose) {
-                    echo 'Finished...' . chr(10);
+                    echo 'Finished...'.chr(10);
                 }
                 break;
             }
@@ -114,16 +114,16 @@ 
             }
             sleep(1);
             if ($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);
         }
     }
 
@@ -143,17 +143,17 @@ 
     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);
-        $startProcessCount = min([$availableProcessesCount,$requiredProcessesCount]);
+        $startProcessCount = min([$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) {
@@ -174,13 +174,13 @@ 
      */
     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;
                 }
@@ -197,9 +197,9 @@ 
      */
     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'), '/');
+        $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;
     }