application/hub/main/filter/task/crawler/class_CrawlerTaskHandlerInitializerFilter.php

   1 <?php
   2 /**
   3  * A TaskHandlerInitializer filter for crawler
   4  *
   5  * @author              Roland Haeder <webmaster@shipsimu.org>
   6  * @version             0.0.0
   7  * @copyright   Copyright (c) 2014 Crawler Developer Team
   8  * @license             GNU GPL 3.0 or any newer version
   9  * @link                http://www.shipsimu.org
  10  *
  11  * This program is free software: you can redistribute it and/or modify
  12  * it under the terms of the GNU General Public License as published by
  13  * the Free Software Foundation, either version 3 of the License, or
  14  * (at your option) any later version.
  15  *
  16  * This program is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19  * GNU General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU General Public License
  22  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  23  */
  24 class CrawlerTaskHandlerInitializerFilter extends BaseCrawlerFilter implements Filterable {
  25         /**
  26          * Protected constructor
  27          *
  28          * @return      void
  29          */
  30         protected function __construct () {
  31                 // Call parent constructor
  32                 parent::__construct(__CLASS__);
  33         }
  34
  35         /**
  36          * Creates an instance of this filter class
  37          *
  38          * @return      $filterInstance         An instance of this filter class
  39          */
  40         public static final function createCrawlerTaskHandlerInitializerFilter () {
  41                 // Get a new instance
  42                 $filterInstance = new CrawlerTaskHandlerInitializerFilter();
  43
  44                 // Return the instance
  45                 return $filterInstance;
  46         }
  47
  48         /**
  49          * Executes the filter with given request and response objects
  50          *
  51          * @param       $requestInstance        An instance of a class with an Requestable interface
  52          * @param       $responseInstance       An instance of a class with an Responseable interface
  53          * @return      void
  54          * @throws      FilterChainException    If the filter chain needs to be interrupted
  55          * @todo        10% done
  56          */
  57         public function execute (Requestable $requestInstance, Responseable $responseInstance) {
  58                 // Get crawler instance
  59                 //$crawlerInstance = Registry::getRegistry()->getInstance('crawler');
  60
  61                 // Get a new task handler instance
  62                 $handlerInstance = ObjectFactory::createObjectByConfiguredName('task_handler_class');
  63
  64                 /*
  65                  * Register all tasks:
  66                  *
  67                  * 1) "Communicator" for crawler->node communication
  68                  */
  69                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_node_communicator_task_class');
  70                 $handlerInstance->registerTask('crawler_node_communicator', $taskInstance);
  71
  72                 // 2) Local URL getter (gets URLs locally and adds them to the analyzer's input stack)
  73                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_local_url_getter_task_class');
  74                 $handlerInstance->registerTask('crawler_local_url_getter', $taskInstance);
  75
  76                 // 3) Remote URL getter (gets URLs locally for other nodes, also includes the crawled URL in local index)
  77                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_remote_url_getter_task_class');
  78                 $handlerInstance->registerTask('crawler_remote_url_getter', $taskInstance);
  79
  80                 // 4) Remote-crawl publisher (publishes crawl jobs for remote retrieval)
  81                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_remote_job_publisher_task_class');
  82                 $handlerInstance->registerTask('crawler_remote_job_publisher', $taskInstance);
  83
  84                 // 5) MIME sniffer
  85                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_mime_sniffer_task_class');
  86                 $handlerInstance->registerTask('crawler_mime_sniffer', $taskInstance);
  87
  88                 // 6) Document parser (converts document to meta format)
  89                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_document_parser_task_class');
  90                 $handlerInstance->registerTask('crawler_document_parser', $taskInstance);
  91
  92                 // 7) Document structure analyzer
  93                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_structure_analyzer_task_class');
  94                 $handlerInstance->registerTask('crawler_structure_analyzer', $taskInstance);
  95
  96                 // 8) Snippet extractor
  97                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_snippet_extractor_task_class');
  98                 $handlerInstance->registerTask('crawler_snippet_extractor', $taskInstance);
  99
 100                 // 9) Node ping
 101                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_ping_task_class');
 102                 $handlerInstance->registerTask('crawler_ping', $taskInstance);
 103
 104                 // 10) URL source: local start
 105                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_url_source_local_start_task_class');
 106                 $handlerInstance->registerTask('crawler_url_source_local_start', $taskInstance);
 107
 108                 // 11) URL source: uploaded list
 109                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_url_source_uploaded_list_task_class');
 110                 $handlerInstance->registerTask('crawler_url_source_uploaded_list', $taskInstance);
 111
 112                 // 12) URL source: RSS feed
 113                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_url_source_rss_start_task_class');
 114                 $handlerInstance->registerTask('crawler_url_source_rss_start', $taskInstance);
 115
 116                 // 13) URL source: found RSS/ATOM feed
 117                 $taskInstance = ObjectFactory::createObjectByConfiguredName('crawler_url_source_found_rss_task_class');
 118                 $handlerInstance->registerTask('crawler_url_source_found_rss', $taskInstance);
 119
 120                 // Put the task handler in registry
 121                 Registry::getRegistry()->addInstance('task_handler', $handlerInstance);
 122         }
 123 }
 124
 125 // [EOF]
 126 ?>