git.mxchange.org Git - hub.git/blob

   1 <?php
   2 // Own namespace
   3 namespace Org\Shipsimu\Hub\Filter\Initializer\Crawler\Handler\Task;
   4
   5 // Import application-specific stuff
   6 use Org\Shipsimu\Hub\Crawler\Filter\Task\BaseCrawlerTaskFilter;
   7
   8 // Import framework stuff
   9 use Org\Mxchange\CoreFramework\Bootstrap\FrameworkBootstrap;
  10 use Org\Mxchange\CoreFramework\Factory\Object\ObjectFactory;
  11 use Org\Mxchange\CoreFramework\Filter\Filterable;
  12 use Org\Mxchange\CoreFramework\Registry\GenericRegistry;
  13 use Org\Mxchange\CoreFramework\Request\Requestable;
  14 use Org\Mxchange\CoreFramework\Response\Responseable;
  15
  16 /**
  17  * A TaskHandlerInitializer filter for crawler
  18  *
  19  * @author              Roland Haeder <webmaster@shipsimu.org>
  20  * @version             0.0.0
  21  * @copyright   Copyright (c) 2014 - 2020 Crawler Developer Team
  22  * @license             GNU GPL 3.0 or any newer version
  23  * @link                http://www.shipsimu.org
  24  *
  25  * This program is free software: you can redistribute it and/or modify
  26  * it under the terms of the GNU General Public License as published by
  27  * the Free Software Foundation, either version 3 of the License, or
  28  * (at your option) any later version.
  29  *
  30  * This program is distributed in the hope that it will be useful,
  31  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  32  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  33  * GNU General Public License for more details.
  34  *
  35  * You should have received a copy of the GNU General Public License
  36  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  37  */
  38 class CrawlerDistributedTaskHandlerInitializerFilter extends BaseCrawlerTaskFilter implements Filterable {
  39         /**
  40          * Protected constructor
  41          *
  42          * @return      void
  43          */
  44         private function __construct () {
  45                 // Call parent constructor
  46                 parent::__construct(__CLASS__);
  47         }
  48
  49         /**
  50          * Creates an instance of this filter class
  51          *
  52          * @return      $filterInstance         An instance of this filter class
  53          */
  54         public static final function createCrawlerDistributedTaskHandlerInitializerFilter () {
  55                 // Get a new instance
  56                 $filterInstance = new CrawlerDistributedTaskHandlerInitializerFilter();
  57
  58                 // Return the instance
  59                 return $filterInstance;
  60         }
  61
  62         /**
  63          * Executes the filter with given request and response objects
  64          *
  65          * @param       $requestInstance        An instance of a class with an Requestable interface
  66          * @param       $responseInstance       An instance of a class with an Responseable interface
  67          * @return      void
  68          * @throws      FilterChainException    If the filter chain needs to be interrupted
  69          * @todo        10% done
  70          */
  71         public function execute (Requestable $requestInstance, Responseable $responseInstance) {
  72                 // Run parent method first
  73                 parent::execute($requestInstance, $responseInstance);
  74
  75                 // Get task handler instance
  76                 $handlerInstance = GenericRegistry::getRegistry()->getInstance('task_handler');
  77
  78                 /*
  79                  * Register all tasks for a distributed crawler:
  80                  *
  81                  * 1) "Communicator" for crawler->node communication
  82                  */
  83                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_node_communicator_task_class');
  84                 $handlerInstance->registerTask('crawler_node_communicator', $taskInstance);
  85
  86                 // 2) Local URL crawler (gets URLs locally and adds them to the analyzer's input stack)
  87                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_local_url_crawler_task_class');
  88                 $handlerInstance->registerTask('crawler_local_url_crawler', $taskInstance);
  89
  90                 // 3) Remote URL crawler (gets URLs locally for other nodes, also includes the crawled URL in local index)
  91                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_remote_url_crawler_task_class');
  92                 $handlerInstance->registerTask('crawler_remote_url_crawler', $taskInstance);
  93
  94                 // 4) Remote-crawl publisher (publishes crawl jobs for remote retrieval)
  95                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_remote_job_publisher_task_class');
  96                 $handlerInstance->registerTask('crawler_remote_job_publisher', $taskInstance);
  97
  98                 // 5) Node ping
  99                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_ping_task_class');
 100                 $handlerInstance->registerTask('crawler_ping', $taskInstance);
 101
 102                 // 6) URL sources
 103                 foreach (explode(':', FrameworkBootstrap::getConfigurationInstance()->getConfigEntry('crawler_url_stacks')) as $stack) {
 104                         // Init task instance
 105                         $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_url_source_' . $stack . '_task_class');
 106
 107                         // And register it
 108                         $handlerInstance->registerTask('crawler_url_source_' . $stack, $taskInstance);
 109                 }
 110         }
 111
 112 }