hipo_job_funnelback_reindex.inc

<?php
require_once SQ_SYSTEM_ROOT.'/core/hipo/hipo_job.inc';

class HIPO_Job_Funnelback_Reindex extends HIPO_Job
{


    function HIPO_Job_Funnelback_Reindex($code_name='')
    {
        $this->uses_trans = FALSE;
        $this->HIPO_Job($code_name);
        $GLOBALS['SQ_SYSTEM']->lm->includePackageStrings('funnelback');
        $this->_hipo_vars['job_dir'] = SQ_PACKAGES_PATH.'/funnelback/hipo_jobs';

    }//end constructor


    public static function paintConfig(&$o, $class, $write_access)
    {
        $o->openField(translate('fnb_hipo_reindex'));

        if ($write_access) {
            text_box($class.'[SQ_HIPO_REINDEX_FUNNELBACK]', SQ_HIPO_REINDEX_FUNNELBACK, 1);
        } else {
            echo SQ_HIPO_REINDEX_FUNNELBACK;
        }

        $o->closeField();

    }//end paintConfig()


    public static function getConfigVars()
    {
        return Array(
                'SQ_HIPO_REINDEX_FUNNELBACK'    => Array('editable' => 1, 'default' => 5),
               );

    }//end getConfigVars()


    function getThresholdPercentageRequired()
    {
        return 0;

    }//end getThresholdPercentageRequired()


    function getCodeName()
    {
        return parent::getCodeName().'-funnelback-reindex';

    }//end getCodeName()


    function getHipoName()
    {
        return translate('fnb_hipo_reindex');

    }//end getHipoName()


    function getInitialStepData()
    {
        return Array(
                Array(
                    'name'          => translate('fnb_hipo_reindex_step_name0'),
                    'function_call' => Array(
                                        'process_function'  => 'processReindex',
                                       ),
                    'running_mode'  => 'server',
                    'auto_step'     => TRUE,
                    'percent_done'  => 0,
                    'complete'      => FALSE,
                    'message'       => '',
                    'allow_cancel'  => TRUE,
                ),
               );

    }//end getInitialStepData()


    function freestyle()
    {
        // Put the HIPO in some context ;)
        if (array_get_index('contextid', $this->_running_vars)) {
            $this->_running_vars['contextid'] = $GLOBALS['SQ_SYSTEM']->getContextId();
        }

        while (!empty($this->_running_vars['todo_collections'])) {
            if (!$this->processReindex($this->_steps[0], get_class_lower($this))) {
                return FALSE;
            }
        }

        return TRUE;

    }//end freestyle()


    function prepare()
    {
        // Defaults
        $this->_running_vars['todo_collections'] = Array();
        $this->_running_vars['done_collections'] = Array();
        $fm = $GLOBALS['SQ_SYSTEM']->am->getSystemAsset('funnelback_manager');
        if (is_null($fm)) {
            trigger_localised_error('FNB0020', E_USER_WARNING);
            return FALSE;
        }//end if
        $collections = $fm->getCollections();

        // If we have not been told which context to re-index, assume we're in
        // it already (but this is not recommended if not freestyling, since it
        // could be stuck somewhere where the context you want is not set)
        if (array_get_index('contextid', $this->_running_vars)) {
            $this->_running_vars['contextid'] = $GLOBALS['SQ_SYSTEM']->getContextId();
        }

        // To reindex, get a collection or if none specified do the whole system
        if (!isset($this->_running_vars['collections']) || (is_null($this->_running_vars['collections']) || empty($this->_running_vars['collections']))) {
            // No URL specified, do the lot, fool!
            foreach ($collections as $collection_id => $collection_info) {
                // Meta collections not reindex here, so skip
                $collection_type = array_get_index($collection_info, 'type', 'web');
                if ($collection_type == 'meta') continue;

                $collection_valid = array_get_index($collection_info, 'valid', FALSE);
                if ($collection_valid === TRUE) {
                    $this->_running_vars['todo_collections'][] = Array($collection_id => $collection_info);
                }//end if
            }//end foreach
        } else {
            // Specified URL(s), do these ones, fool!
            if (!is_array($this->_running_vars['collections'])) {
                // If one specified, make it an array
                $this->_running_vars['collections'] = Array($this->_running_vars['collections']);
            }//end if

            foreach ($this->_running_vars['collections'] as $todo_collection) {
                // we have the collection, go thru each collection and find the correct entry (make sure the collection is valid)
                foreach ($collections as $collection_id => $collection_info) {
                    // Meta collections not reindex here, so skip
                    $collection_type = array_get_index($collection_info, 'type', 'web');
                    if ($collection_type == 'meta') continue;

                    $collection_valid = array_get_index($collection_info, 'valid', FALSE);
                    if ($collection_valid === TRUE && $collection_id == $todo_collection) {
                        $this->_running_vars['todo_collections'][] = Array($collection_id => $collection_info);
                    }//end if
                }//end foreach
            }//end foreach
        }//end if

        return parent::prepare();

    }//end prepare()


    function processReindex(&$step_data, $prefix)
    {
        $GLOBALS['SQ_SYSTEM']->changeContext(array_get_index($this->_running_vars, 'contextid', 0));

        $am =& $GLOBALS['SQ_SYSTEM']->am;
        $fm = $am->getSystemAsset('funnelback_manager');
        if (is_null($fm)) {
            return FALSE;
        }//end if

        // get next url to reindex, if none to do just flow through and finish the job neat and tidy ;)
        if (!empty($this->_running_vars['todo_collections'])) {
            reset($this->_running_vars['todo_collections']);
            $next = key($this->_running_vars['todo_collections']);
            $collection_id   = key($this->_running_vars['todo_collections'][$next]);
            $collection_info = $this->_running_vars['todo_collections'][$next][$collection_id];
            $url = array_get_index($collection_info, 'url', Array());
            $assetid = array_get_index($collection_info, 'asset', 0);
            $asset = $GLOBALS['SQ_SYSTEM']->am->getAsset($assetid, '', TRUE);
            unset($this->_running_vars['todo_collections'][$next]);
        
            if (!empty($url) && !is_null($asset)) {
                // Let people know what you are thinking ;-)
                $step_data['message'] = translate('fnb_hipo_reindexing_url', htmlspecialchars($asset->attr('name')));

                // Build the parameters
                $output = '';
                $xml = $fm->generateXMLCache($url, $assetid, FALSE);
                $log_file = SQ_LOG_PATH.'/funnelback.log';
                $bin_path = $fm->getFunnelbackPath();
                $coll_path = $fm->getCollectionDataPath($collection_id);
                $coll_log_path = $fm->getCollectionDataPath($collection_id, 'log');
                $data_path = $fm->getCollectionDataPath($collection_id, 'idx_new');
                $live_path = $fm->getCollectionDataPath($collection_id, 'idx');
                if (empty($data_path)) return FALSE;
                $options = $fm->getIndexingOptions();

                // Check for permissions and show a fail if cannot write
                if (!is_writable($data_path)) {
                    trigger_localised_error('FNB0030', E_USER_WARNING);
                    return FALSE;
                }//end if

                // Create the XML on the filesystem
                file_put_contents($data_path.'/sitemap.xml', $xml);
                // Calculate max line length of metadata summary
                preg_match('/[0-9]+/', exec('wc -L '.$data_path.'/sitemap.xml'), $matches);
                $maxLineLength = isset($matches[0]) ? $matches[0] : NULL;

                // Apply login credentials
                $user_id = $fm->attr('user');
                if (!empty($user_id)) {
                    $user = $GLOBALS['SQ_SYSTEM']->am->getAsset($user_id, '', TRUE);
                    if (!is_null($user)) {
                        $username = $user->attr('username');
                        $password = $fm->attr('password');
                        file_put_contents($data_path.'/creds', $username.' '.$password);
                        $options[] = '-CF='.$data_path.'/creds';
                        $options[] = '-loginPrefix='.SQ_CONF_LOGIN_SUFFIX;
                    }//end if
                }//end if

                // Add in the click log if exists
                if (file_exists($coll_log_path.'/click.log')) {
                    $options[] = '-F'.$coll_log_path.'/click.log';
                }//end if

                // Make sure we ONLY index with at least one key (public assets will have a 'public' key)
                $options[] = '-security_level=1';

                // Build the command line
                $command = $bin_path.'/bin/padre-iw';
                $command .= ' -secmap '.escapeshellarg($data_path.'/sitemap.xml');
                $command .= ' '.escapeshellarg($collection_id);     // Index Stem
                $command .= ' '.implode(' ', $options);             // Command line Options
                $command .= isset($maxLineLength) ? ' -mdsfml'.$maxLineLength : ''; // Max line length of metadata summary file

                // Start the reindex process
                $status = $fm->runCommand($command, $output, NULL, $log_file, $data_path, Array('SITE_SEARCH_ROOT'=>$coll_path));

                // Regardless of status, remove the credential file
                if (file_exists($data_path.'/creds')) {
                    unlink($data_path.'/creds');
                }//end if

                if ($status === 0) {
                    // Successfully completed
                    $make_live_cmd = 'rm -rf idx_old; mv idx idx_old; mv idx_new idx';
                    $final_status = $fm->runCommand($make_live_cmd, $output, NULL, $log_file, $coll_path);
                    if ($status !== 0) {
                        // Failed making live
                        trigger_localised_error('FNB0032', E_USER_WARNING);
                        return FALSE;
                    }//end if
                } else {
                    // Fail....
                    trigger_localised_error('FNB0031', E_USER_WARNING);
                    return FALSE;
                }//end if

                // Forget some information
                $GLOBALS['SQ_SYSTEM']->am->forgetAsset($asset, TRUE);
            }//end if
        
            // add this assetid to the done array
            $this->_running_vars['done_collections'][] = $collection_id;
        }//end if
    
        $GLOBALS['SQ_SYSTEM']->restoreContext();
    
        if (empty($this->_running_vars['todo_collections'])) {
            $step_data['percent_done'] = 100;
            $step_data['complete']     = TRUE;
        } else {
            $total = count($this->_running_vars['todo_collections']) + count($this->_running_vars['done_collections']);
            $step_data['percent_done'] =(count($this->_running_vars['done_collections']) / $total) * 100;
            $step_data['complete']     = FALSE;
        }

        return TRUE;

    }//end processReindex()


}//end class

?>