Remove a list of spam words from the user interest database in OJS

deleteInterests.php

<?php

/** This version works on 2.2.x versions of OJS.  
 * The ways interests are stored has changed a lot so tweaks will need to be made depending on the version of OJS
 
 The following method needs to be added to lib/pkp/classes/user/InterestDAO.inc.php
 
 	function getInterestObjects($userId) {
                $interests = $this->build($userId);
                $interestEntryDao =& DAORegistry::getDAO('InterestEntryDAO');
                $userInterests = $interestEntryDao->getByControlledVocabId($interests->getId());
                return $userInterests;
        }
 
 */
 

/**
 * @file deleteInterests.php
 *
 * Copyright (c) 2003-2012 John Willinsky
 * Distributed under the GNU GPL v2. For full terms see the file docs/COPYING.
 *
 * @class deleteInterests
 * @ingroup tools
 *
 * @brief CLI tool for removing (spammy) reviewing interests
 */

// $Id$


require(dirname(__FILE__) . '/bootstrap.inc.php');

class deleteInterests extends CommandLineTool {
	/** @var $inputFile string */
	var $inputFile;

	/**
	 * Constructor.
	 * @param $argv array command-line arguments
	 */
	function deleteInterests($argv = array()) {
		parent::CommandLineTool($argv);

		if (!isset($this->argv[0])) {
			$this->usage();
			exit(1);
		}

		$this->inputFile = $this->argv[0];
	}

	/**
	 * Print command usage information.
	 */
	function usage() {
		echo "OJS 2 interest removal tool\n"
			. "Use this tool to remove spammy user interest.\n\n"
			. "Usage: {$this->scriptName} [filename] \n"
			. "username1      The file with spam words to delete (comma separated).\n";
	}

	/**
	 * Execute the merge users command.
	 */
	function execute() {
		//load file into array
		ini_set('auto_detect_line_endings', true);
		$fileContents = file_get_contents($this->inputFile, true);
		$badInterests = array_map('trim', explode(',', $fileContents));

		$interestDao =& DAORegistry::getDAO('InterestDAO'); /* @var $interestDao InterestDAO */
		$interestEntryDao =& DAORegistry::getDAO('InterestEntryDAO'); /* @var $interestDao InterestDAO */

		$userDao =& DAORegistry::getDAO('UserDAO');
		$allUsers =& $userDao->getUsersByField();
		$counter = 0;
		while($user =& $allUsers->next()) {
			$userInterests = $interestDao->getInterestObjects($user->getId());
			while($interest =& $userInterests->next()) {
				if(strpos($interest->getInterest(), 'forums') !== false || strpos($interest->getInterest(), 'href') !== false || in_array($interest->getInterest(), $badInterests)) {
					$interestEntryDao->deleteObject($interest);
					printf("'%s' found and deleted.\n", $interest->getInterest());
					$counter++;
				}
				unset($interest);
			}
			unset($user);
			unset($userInterests);
		}

		// Iterate over all interests and check if they're in our spam list
		printf("Process completed: %s items removed.\n", $counter);

	}
}

$tool = new deleteInterests(isset($argv) ? $argv : array());
$tool->execute();
?>