application/hub/main/source/urls/class_CrawlerUploadedListUrlSource.php

   1 <?php
   2 /**
   3  * A UploadedList URL source class for crawlers
   4  *
   5  * @author              Roland Haeder <webmaster@ship-simu.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.ship-simu.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 CrawlerUploadedListUrlSource extends BaseUrlSource implements UrlSource, Registerable {
  25         /**
  26          * "Cached" CSV path
  27          */
  28         private $csvFilePath = '';
  29
  30         /**
  31          * Last CSV file instance
  32          */
  33         private $lastCsvFileInstance = NULL;
  34
  35         /**
  36          * Stack for pushing data from this clas to another
  37          */
  38         private $stackSourceInstance = NULL;
  39
  40         /**
  41          * Stack name for a CSV file
  42          */
  43         const STACK_NAME_CSV_FILE = 'csv_file';
  44
  45         /**
  46          * "Imported" CSV files
  47          */
  48         private $csvFileImported = array();
  49
  50         /**
  51          * Protected constructor
  52          *
  53          * @return      void
  54          */
  55         protected function __construct () {
  56                 // Call parent constructor
  57                 parent::__construct(__CLASS__);
  58
  59                 // "Cache" CSV path for faster usage
  60                 $this->csvFilePath = $this->getConfigInstance()->getConfigEntry('base_path') . '/' . $this->getConfigInstance()->getConfigEntry('csv_file_path');
  61
  62                 // Initialize directory instance
  63                 $directoryInstance = ObjectFactory::createObjectByConfiguredName('directory_class', array($this->csvFilePath));
  64
  65                 // Set it here
  66                 $this->setDirectoryInstance($directoryInstance);
  67
  68                 // Init stack instance
  69                 $this->stackSourceInstance = ObjectFactory::createObjectByConfiguredName('crawler_uploaded_list_url_source_stack_class');
  70
  71                 // Init stack
  72                 $this->getStackSourceInstance()->initStack(self::STACK_NAME_CSV_FILE);
  73         }
  74
  75         /**
  76          * Creates an instance of this class
  77          *
  78          * @return      $sourceInstance         An instance of a Source class
  79          */
  80         public final static function createCrawlerUploadedListUrlSource () {
  81                 // Get new instance
  82                 $sourceInstance = new CrawlerUploadedListUrlSource();
  83
  84                 // Init source
  85                 $sourceInstance->initSource('crawler', 'uploaded_list');
  86
  87                 // Return the prepared instance
  88                 return $sourceInstance;
  89         }
  90
  91         /**
  92          * Checks whether a CSV file is found
  93          *
  94          * @return      $isFound        Whether a CSV file is found
  95          */
  96         private function isCsvFileFound () {
  97                 // Is the instance valid?
  98                 if (!$this->getDirectoryInstance()->getDirectoryIteratorInstance()->valid()) {
  99                         // Then rewind it
 100                         $this->getDirectoryInstance()->getDirectoryIteratorInstance()->rewind();
 101                 } // END - if
 102
 103                 // Read next entry
 104                 $directoryEntry = $this->getDirectoryInstance()->readDirectoryExcept(array_merge(array('.htaccess', '.', '..'), $this->csvFileImported));
 105
 106                 // The read entry has not to be empty and extension must be '.csv'
 107                 if ((empty($directoryEntry)) || (substr($directoryEntry, -4, 4) != '.csv')) {
 108                         // Skip further processing
 109                         return FALSE;
 110                 } // END - if
 111
 112                 // Initialize CSV file instance
 113                 $this->lastCsvFileInstance = ObjectFactory::createObjectByConfiguredName('csv_file_class', array($this->csvFilePath . '/' . $directoryEntry));
 114
 115                 // Found an entry
 116                 return TRUE;
 117         }
 118
 119         /**
 120          * Initializes the import of the CSV file which is being processed by other task
 121          *
 122          * @return      void
 123          * @throws      NullPointerException    If lastCsvFileInstance is not set
 124          */
 125         private function importCsvFile () {
 126                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__)->debugOutput('CRAWLER-SOURCE [' . __METHOD__ . ':' . __LINE__ . ']: CALLED!');
 127
 128                 // Is the instance set?
 129                 if (is_null($this->lastCsvFileInstance)) {
 130                         // This should not happen
 131                         throw new NullPointerException($this, self::EXCEPTION_IS_NULL_POINTER);
 132                 } // END - if
 133
 134                 // Stack this file
 135                 $this->getStackSourceInstance()->pushNamed(self::STACK_NAME_CSV_FILE, $this->lastCsvFileInstance);
 136
 137                 // ... and mark it as "imported"
 138                 array_push($this->csvFileImported, basename($this->lastCsvFileInstance->getFileName()));
 139
 140                 // ... and finally NULL it (to save some RAM)
 141                 $this->lastCsvFileInstance = NULL;
 142
 143                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__)->debugOutput('CRAWLER-SOURCE [' . __METHOD__ . ':' . __LINE__ . ']: EXIT!');
 144         }
 145
 146         /**
 147          * Getter for stackSourceInstance variable
 148          *
 149          * @return      $stackSourceInstance    An instance of an additional stack
 150          */
 151         public final function stackSourceInstance () {
 152                 return $this->stackSourceInstance;
 153         }
 154
 155         /**
 156          * Processes entries in the stack.
 157          *
 158          * @return      void
 159          * @todo        ~10% done
 160          */
 161         public function processStack () {
 162                 // Does the stack have some entries left?
 163                 if (!$this->isUrlStackEmpty()) {
 164                         // Handle next entry
 165                         $this->processNextEntry();
 166                 } elseif ($this->isCsvFileFound()) {
 167                         // A CSV file has been found and can maybe be imported.
 168                         $this->importCsvFile();
 169                 }
 170
 171                 $this->partialStub('Please implement this method.');
 172         }
 173 }
 174
 175 // [EOF]
 176 ?>