root/websites/planet.php.gr.jp/trunk/app/action_cli/Crawler.php @ 13538

<?php
/**
 *  Crawler.php
 *
 *  @author     halt feits <halt.feits at gmail.com>
 *  @package    Delphinus
 *  @version    $Id: skel.action.php,v 1.4 2005/01/04 12:53:26 fujimoto Exp $
 */

/**
 *  Crawler����若����茖� *
 *  @author     halt feits <halt.feits at gmail.com>
 *  @access     public
 *  @package    Delphinus
 */
class Delphinus_CLI_Form_Crawler extends Ethna_ActionForm
{
    /**
     *  @access private
     *  @var    array   ����若��ゅ�臂�     */
    var $form = array(
        /*
        'sample' => array(
            'name'          => '�泣����',      // 茵�ず��            'required'      => true,            // 綽������激���true/false)
            'min'           => null,            // �����            'max'           => null,            // ��ぇ��            'regexp'        => null,            // ���腮��絎�罩ｈ�茵��
            'custom'        => null,            // �＜�����������с���            'filter'        => null,            // �ュ��ゅ�����ｃ��帥�����с�
            'form_type'     => FORM_TYPE_TEXT   // ����若���            'type'          => VAR_TYPE_INT,    // �ュ��ゅ�
        ),
        */
    );
}

/**
 *  Crawler�≪��激��潟�絎��
 *
 *  @author     halt feits <halt.feits at gmail.com>
 *  @access     public
 *  @package    Delphinus
 */
class Delphinus_CLI_Action_Crawler extends Ethna_ActionClass
{
    /**
     *  Crawler�≪��激��潟������     *
     *  @access public
     *  @return string      �欠Щ��罩ｅ幻腟�����null, ���腟�����false)
     */
    function prepare()
    {
        $Controller = $this->backend->getController();
        if ( $Controller->getCLI() ) {
            return null;
        }

        return false;

    }

    /**
     *  Crawler�≪��激��潟�絎��
     *
     *  @access public
     *  @return string  �欠Щ��     */
    function perform()
    {
        $cache =& $this->backend->plugin->getPlugin('Cachemanager', 'Localfile');
        $cache->clear('index');

        print("Start Crawl Feeds<br>\n");
        $this->crawlRSS();
        $Controller = $this->backend->getController();
        $path_bin = $Controller->getDirectory('bin');
        exec("php {$path_bin}/rss_rebuild.php");
        return null;
    }

    /**
     * crawlRSS
     *
     * @access protected
     * @param void
     */
    function crawlRSS()
    {
        require_once 'XML/Feed/Parser.php';

        $DB = $this->backend->getDB();
        $rss_list = $DB->getRssList();
        $allow_category = $this->config->get('allow_category');

        foreach( $rss_list as $rss){

            print("Fetch:{$rss['url']}<br>\n");

            try {
                $feed = new XML_Feed_Parser(file_get_contents($rss['rss_url']));
            } catch(XML_Feed_Parser_Exception $e) {
                print('Feed invalid: ' . $e->getMessage());
            }

            foreach($feed as $entry){

                if (is_string($allow_category) && ($entry->category !== false) && (strtoupper($entry->category) != $allow_category)) {
                    print("Parge:{$entry->category}\n");
                    continue;
                }

                $item['title'] = $entry->title;
                if ($entry->date !== false) {
                    $item['date'] = date('Y-m-d H:i:s', $entry->date);
                } else {
                    $item['date'] = date('Y-m-d H:i:s', $entry->pubdate);
                }
                $item['link'] = $entry->link;
                if ($entry->description !== false) {
                    $item['description'] = $entry->description;
                } else {
                    $item['description'] = $entry->content;
                }

                if ( !$DB->existsEntryFromLink($item['link']) ) {
                    $DB->setEntry($rss['id'], $item);
                } else {
                    print('Delete Entry' . "\n");
                    $DB->deleteEntry($item['link']);
                    $DB->setEntry($rss['id'], $item);
                }
            }
        }

        return true;
    }

}
?>