"TYPO3 4.0.1: typo3_src-4.0.1/typo3/sysext/indexed_search/class.crawler.php Source File", "datetime" => "Sat Dec 2 19:22:32 2006", "date" => "2 Dec 2006", "doxygenversion" => "1.4.6", "projectname" => "TYPO3 4.0.1", "projectnumber" => "4.0.1" ); get_header($doxygen_vars); ?>

class.crawler.php

<?php
/***************************************************************
*  Copyright notice
*
*  (c) 2001-2005 Kasper Skaarhoj (kasperYYYY@typo3.com)
*  All rights reserved
*
*  This script is part of the TYPO3 project. The TYPO3 project is
*  free software; you can redistribute it and/or modify
*  it under the terms of the GNU General Public License as published by
*  the Free Software Foundation; either version 2 of the License, or
*  (at your option) any later version.
*
*  The GNU General Public License can be found at
*  http://www.gnu.org/copyleft/gpl.html.
*  A copy is found in the textfile GPL.txt and important notices to the license
*  from the author is found in LICENSE.txt distributed with these scripts.
*
*
*  This script is distributed in the hope that it will be useful,
*  but WITHOUT ANY WARRANTY; without even the implied warranty of
*  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
*  GNU General Public License for more details.
*
*  This copyright notice MUST APPEAR in all copies of the script!
***************************************************************/
# To make sure the backend charset is available:
require_once(PATH_site.TYPO3_mainDir.'sysext/lang/lang.php');
if (!is_object($GLOBALS['LANG']))       {
        $GLOBALS['LANG'] = t3lib_div::makeInstance('language');
        $GLOBALS['LANG']->init($GLOBALS['BE_USER']->uc['lang']);
}


class tx_indexedsearch_crawler {

                // Static:
        var $secondsPerExternalUrl = 3;         // Number of seconds to use as interval between queued indexing operations of URLs / files (types 2 & 3)

                // Internal, dynamic:
        var $instanceCounter = 0;               // Counts up for each added URL (type 3)

                // Internal, static:
        var $callBack = 'EXT:indexed_search/class.crawler.php:&tx_indexedsearch_crawler';               // The object reference to this class.

        function crawler_init(&$pObj){

                        // Select all indexing configuration which are waiting to be activated:
                $indexingConfigurations = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                        '*',
                        'index_config',
                        'hidden=0
                                AND (starttime=0 OR starttime<='.time().')
                                AND timer_next_indexing<'.time().'
                                AND set_id=0
                                '.t3lib_BEfunc::deleteClause('index_config')
                );

                        // For each configuration, check if it should be executed and if so, start:
                foreach($indexingConfigurations as $cfgRec)     {

                                // Generate a unique set-ID:
                        $setId = t3lib_div::md5int(microtime());

                                // Get next time:
                        $nextTime = $this->generateNextIndexingTime($cfgRec);

                                // Start process by updating index-config record:
                        $field_array = array (
                                'set_id' => $setId,
                                'timer_next_indexing' => $nextTime,
                                'session_data' => '',
                        );
                        $GLOBALS['TYPO3_DB']->exec_UPDATEquery('index_config','uid='.intval($cfgRec['uid']), $field_array);

                                // Based on configuration type:
                        switch($cfgRec['type']) {
                                case 1: // RECORDS:

                                                // Parameters:
                                        $params = array(
                                                'indexConfigUid' => $cfgRec['uid'],
                                                'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']'),
                                                'url' => 'Records (start)',     // Just for show.
                                        );
                                                //
                                        $pObj->addQueueEntry_callBack($setId,$params,$this->callBack,$cfgRec['pid']);
                                break;
                                case 2: // FILES:

                                                // Parameters:
                                        $params = array(
                                                'indexConfigUid' => $cfgRec['uid'],             // General
                                                'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']'),      // General
                                                'url' => $cfgRec['filepath'],   // Partly general... (for URL and file types)
                                                'depth' => 0    // Specific for URL and file types
                                        );

                                        $pObj->addQueueEntry_callBack($setId,$params,$this->callBack,$cfgRec['pid']);
                                break;
                                case 3: // External URL:

                                                // Parameters:
                                        $params = array(
                                                'indexConfigUid' => $cfgRec['uid'],             // General
                                                'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']'),      // General
                                                'url' => $cfgRec['externalUrl'],        // Partly general... (for URL and file types)
                                                'depth' => 0    // Specific for URL and file types
                                        );

                                        $pObj->addQueueEntry_callBack($setId,$params,$this->callBack,$cfgRec['pid']);
                                break;
                                case 4: // Page tree

                                                // Parameters:
                                        $params = array(
                                                'indexConfigUid' => $cfgRec['uid'],             // General
                                                'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']'),      // General
                                                'url' => $cfgRec['alternative_source_pid'],     // Partly general... (for URL and file types and page tree (root))
                                                'depth' => 0    // Specific for URL and file types and page tree
                                        );

                                        $pObj->addQueueEntry_callBack($setId,$params,$this->callBack,$cfgRec['pid']);
                                break;
                                case 5: // Meta configuration, nothing to do:
                                        # NOOP
                                break;
                                default:
                                        if ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['crawler'][$cfgRec['type']])       {
                                                $hookObj = &t3lib_div::getUserObj($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['crawler'][$cfgRec['type']]);

                                                if (is_object($hookObj))        {

                                                                // Parameters:
                                                        $params = array(
                                                                'indexConfigUid' => $cfgRec['uid'],             // General
                                                                'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].'/CUSTOM]'),       // General
                                                                'url' => $hookObj->initMessage($message),
                                                        );

                                                        $pObj->addQueueEntry_callBack($setId,$params,$this->callBack,$cfgRec['pid']);
                                                }
                                        }
                                break;
                        }
                }

                        // Finally, look up all old index configurations which are finished and needs to be reset and done.
                $this->cleanUpOldRunningConfigurations();
        }

        function crawler_execute($params,&$pObj)        {

                        // Indexer configuration ID must exist:
                if ($params['indexConfigUid'])  {

                                // Load the indexing configuration record:
                        list($cfgRec) = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                                '*',
                                'index_config',
                                'uid='.intval($params['indexConfigUid'])
                        );

                        if (is_array($cfgRec))  {

                                        // Unpack session data:
                                $session_data = unserialize($cfgRec['session_data']);

                                        // Select which type:
                                switch($cfgRec['type']) {
                                        case 1: // Records:
                                                $this->crawler_execute_type1($cfgRec,$session_data,$params,$pObj);
                                        break;
                                        case 2: // Files
                                                $this->crawler_execute_type2($cfgRec,$session_data,$params,$pObj);
                                        break;
                                        case 3: // External URL:
                                                $this->crawler_execute_type3($cfgRec,$session_data,$params,$pObj);
                                        break;
                                        case 4: // Page tree:
                                                $this->crawler_execute_type4($cfgRec,$session_data,$params,$pObj);
                                        break;
                                        case 5: // Meta
                                                # NOOP (should never enter here!)
                                        break;
                                        default:
                                                if ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['crawler'][$cfgRec['type']])       {
                                                        $hookObj = &t3lib_div::getUserObj($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['crawler'][$cfgRec['type']]);

                                                        if (is_object($hookObj))        {
                                                                $this->pObj = &$pObj;   // For addQueueEntryForHook()
                                                                $hookObj->indexOperation($cfgRec,$session_data,$params,$this);
                                                        }
                                                }
                                        break;
                                }

                                        // Save process data which might be modified:
                                $field_array = array (
                                        'session_data' => serialize($session_data)
                                );
                                $GLOBALS['TYPO3_DB']->exec_UPDATEquery('index_config','uid='.intval($cfgRec['uid']), $field_array);
                        }
                }

                return array('log' => $params);
        }

        function crawler_execute_type1($cfgRec,&$session_data,$params,&$pObj)   {
                if ($cfgRec['table2index'] && isset($GLOBALS['TCA'][$cfgRec['table2index']]))   {

                                // Init session data array if not already:
                        if (!is_array($session_data))   {
                                $session_data = array(
                                        'uid' => 0
                                );
                        }

                                // Init:
                        $pid = intval($cfgRec['alternative_source_pid']) ? intval($cfgRec['alternative_source_pid']) : $cfgRec['pid'];
                        $numberOfRecords = $cfgRec['recordsbatch'] ? t3lib_div::intInRange($cfgRec['recordsbatch'],1) : 100;

                                // Get root line:
                        $rl = $this->getUidRootLineForClosestTemplate($cfgRec['pid']);

                                // Select
                        $recs = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                                                '*',
                                                $cfgRec['table2index'],
                                                'pid = '.intval($pid).'
                                                        AND uid > '.intval($session_data['uid']).
                                                        t3lib_BEfunc::deleteClause($cfgRec['table2index']),
                                                '',
                                                'uid',
                                                $numberOfRecords
                                        );

                                // Traverse:
                        if (count($recs))       {
                                foreach($recs as $r)    {

                                                // Index single record:
                                        $this->indexSingleRecord($r,$cfgRec,$rl);

                                                // Update the UID we last processed:
                                        $session_data['uid'] = $r['uid'];
                                }

                                        // Finally, set entry for next indexing of batch of records:
                                $nparams = array(
                                        'indexConfigUid' => $cfgRec['uid'],
                                        'url' => 'Records from UID#'.($r['uid']+1).'-?',
                                        'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']')
                                );
                                $pObj->addQueueEntry_callBack($cfgRec['set_id'],$nparams,$this->callBack,$cfgRec['pid']);
                        }
                }
        }

        function crawler_execute_type2($cfgRec,&$session_data,$params,&$pObj)   {

                        // Prepare path, making it absolute and checking:
                $readpath = $params['url'];
                if (!t3lib_div::isAbsPath($readpath))   {
                        $readpath = t3lib_div::getFileAbsFileName($readpath);
                }

                if (t3lib_div::isAllowedAbsPath($readpath))     {
                        if (@is_file($readpath))        {       // If file, index it!

                                        // Get root line (need to provide this when indexing external files)
                                $rl = $this->getUidRootLineForClosestTemplate($cfgRec['pid']);

                                        // Load indexer if not yet.
                                $this->loadIndexerClass();

                                        // (Re)-Indexing file on page.
                                $indexerObj = &t3lib_div::makeInstance('tx_indexedsearch_indexer');
                                $indexerObj->backend_initIndexer($cfgRec['pid'], 0, 0, '', $rl);
                                $indexerObj->backend_setFreeIndexUid($cfgRec['uid'], $cfgRec['set_id']);
                                $indexerObj->hash['phash'] = -1;        // EXPERIMENT - but to avoid phash_t3 being written to file sections (otherwise they are removed when page is reindexed!!!)

                                        // Index document:
                                $indexerObj->indexRegularDocument(substr($readpath,strlen(PATH_site)), TRUE);
                        } elseif (@is_dir($readpath)) { // If dir, read content and create new pending items for log:

                                        // Select files and directories in path:
                                $extList = implode(',',t3lib_div::trimExplode(',',$cfgRec['extensions'],1));
                                $fileArr = array();
                                $files = t3lib_div::getAllFilesAndFoldersInPath($fileArr,$readpath,$extList,0,0);

                                $directoryList = t3lib_div::get_dirs($readpath);
                                if (is_array($directoryList) && $params['depth'] < $cfgRec['depth'])    {
                                        foreach ($directoryList as $subdir)     {
                                                if ((string)$subdir!='')        {
                                                        $files[]= $readpath.$subdir.'/';
                                                }
                                        }
                                }
                                $files = t3lib_div::removePrefixPathFromList($files,PATH_site);

                                        // traverse the items and create log entries:
                                foreach($files as $path)        {
                                        $this->instanceCounter++;
                                        if ($path!==$params['url'])     {
                                                        // Parameters:
                                                $nparams = array(
                                                        'indexConfigUid' => $cfgRec['uid'],
                                                        'url' => $path,
                                                        'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']'),
                                                        'depth' => $params['depth']+1
                                                );
                                                $pObj->addQueueEntry_callBack($cfgRec['set_id'],$nparams,$this->callBack,$cfgRec['pid'],time()+$this->instanceCounter*$this->secondsPerExternalUrl);
                                        }
                                }
                        }
                }
        }

        function crawler_execute_type3($cfgRec,&$session_data,$params,&$pObj)   {

                        // Init session data array if not already:
                if (!is_array($session_data))   {
                        $session_data = array(
                                'urlLog' => array($params['url'])
                        );
                }

                        // Index the URL:
                $rl = $this->getUidRootLineForClosestTemplate($cfgRec['pid']);
                $subUrls = $this->indexExtUrl($params['url'], $cfgRec['pid'], $rl, $cfgRec['uid'], $cfgRec['set_id']);

                        // Add more elements to log now:
                if ($params['depth'] < $cfgRec['depth'])        {
                        foreach($subUrls as $url)       {
                                if ($url = $this->checkUrl($url,$session_data['urlLog'],$cfgRec['externalUrl']))        {
                                        if (!$this->checkDeniedSuburls($url, $cfgRec['url_deny']))      {
                                                $this->instanceCounter++;
                                                $session_data['urlLog'][] = $url;

                                                        // Parameters:
                                                $nparams = array(
                                                        'indexConfigUid' => $cfgRec['uid'],
                                                        'url' => $url,
                                                        'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']'),
                                                        'depth' => $params['depth']+1
                                                );
                                                $pObj->addQueueEntry_callBack($cfgRec['set_id'],$nparams,$this->callBack,$cfgRec['pid'],time()+$this->instanceCounter*$this->secondsPerExternalUrl);
                                        }
                                }
                        }
                }
        }

        function crawler_execute_type4($cfgRec,&$session_data,$params,&$pObj)   {

                        // Base page uid:
                $pageUid = intval($params['url']);

                        // Get array of URLs from page:
                $pageRow = t3lib_BEfunc::getRecord('pages',$pageUid);
                $res = $pObj->getUrlsForPageRow($pageRow);

                $duplicateTrack = array();      // Registry for duplicates
                $downloadUrls = array();        // Dummy.

                        // Submit URLs:
                if (count($res))        {
                        foreach($res as $paramSetKey => $vv)    {
                                $urlList = $pObj->urlListFromUrlArray($vv,$pageRow,time(),30,1,0,$duplicateTrack,$downloadUrls,array('tx_indexedsearch_reindex'));
                        }
                }

                        // Add subpages to log now:
                if ($params['depth'] < $cfgRec['depth'])        {

                                // Subpages selected
                        $recs = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                                'uid,title',
                                'pages',
                                'pid = '.intval($pageUid).
                                        t3lib_BEfunc::deleteClause('pages')
                        );

                                // Traverse subpages and add to queue:
                        if (count($recs))       {
                                foreach($recs as $r)    {
                                        $this->instanceCounter++;
                                        $url = 'pages:'.$r['uid'].': '.$r['title'];
                                        $session_data['urlLog'][] = $url;

                                                        // Parameters:
                                        $nparams = array(
                                                'indexConfigUid' => $cfgRec['uid'],
                                                'url' => $r['uid'],
                                                'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']'),
                                                'depth' => $params['depth']+1
                                        );
                                        $pObj->addQueueEntry_callBack($cfgRec['set_id'],$nparams,$this->callBack,$cfgRec['pid'],time()+$this->instanceCounter*$this->secondsPerExternalUrl);
                                }
                        }
                }
        }

        function cleanUpOldRunningConfigurations()      {

                        // Lookup running index configurations:
                $runningIndexingConfigurations = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                        'uid,set_id',
                        'index_config',
                        'set_id!=0'.t3lib_BEfunc::deleteClause('index_config')
                );

                        // For each running configuration, look up how many log entries there are which are scheduled for execution and if none, clear the "set_id" (means; Processing was DONE)
                foreach($runningIndexingConfigurations as $cfgRec)      {

                                // Look for ended processes:
                        list($queued_items) = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                                'count(*) AS count',
                                'tx_crawler_queue',
                                'set_id='.intval($cfgRec['set_id']).' AND exec_time=0'
                        );

                        if (!$queued_items['count'])    {

                                        // Lookup old phash rows:
                                $oldPhashRows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                                        'phash',
                                        'index_phash',
                                        'freeIndexUid='.intval($cfgRec['uid']).' AND freeIndexSetId!='.$cfgRec['set_id']
                                );

                                foreach($oldPhashRows as $pHashRow)     {
                                                // Removing old registrations for all tables (code copied from class.tx_indexedsearch_modfunc1.php)
                                        $tableArr = explode(',','index_phash,index_rel,index_section,index_grlist,index_fulltext,index_debug');
                                        foreach($tableArr as $table)    {
                                                $GLOBALS['TYPO3_DB']->exec_DELETEquery($table, 'phash='.intval($pHashRow['phash']));
                                        }
                                }

                                        // End process by updating index-config record:
                                $field_array = array (
                                        'set_id' => 0,
                                        'session_data' => '',
                                );
                                $GLOBALS['TYPO3_DB']->exec_UPDATEquery('index_config','uid='.intval($cfgRec['uid']), $field_array);
                        }
                }
        }







        /*****************************************
         *
         * Helper functions
         *
         *****************************************/

        function checkUrl($url,$urlLog,$baseUrl)        {
                $url = ereg_replace('\/\/$','/',$url);
                list($url) = explode('#',$url);

                if (!strstr($url,'../'))        {
                        if (t3lib_div::isFirstPartOfStr($url,$baseUrl)) {
                                if (!in_array($url,$urlLog))    {
                                        return $url;
                                }
                        }
                }
        }

        function indexExtUrl($url, $pageId, $rl, $cfgUid, $setId)       {

                        // Load indexer if not yet.
                $this->loadIndexerClass();

                        // Index external URL:
                $indexerObj = &t3lib_div::makeInstance('tx_indexedsearch_indexer');
                $indexerObj->backend_initIndexer($pageId, 0, 0, '', $rl);
                $indexerObj->backend_setFreeIndexUid($cfgUid, $setId);
                $indexerObj->hash['phash'] = -1;        // To avoid phash_t3 being written to file sections (otherwise they are removed when page is reindexed!!!)

                $indexerObj->indexExternalUrl($url);
                $url_qParts = parse_url($url);

                        // Get URLs on this page:
                $subUrls = array();
                $list = $indexerObj->extractHyperLinks($indexerObj->indexExternalUrl_content);

                                                // Traverse links:
                foreach($list as $count => $linkInfo)   {

                                // Decode entities:
                        $subUrl = t3lib_div::htmlspecialchars_decode($linkInfo['href']);

                        $qParts = parse_url($subUrl);
                        if (!$qParts['scheme']) {
                                $subUrl = $url_qParts['scheme'].'://'.$url_qParts['host'].'/'.t3lib_div::resolveBackPath($subUrl);
                        }

                        $subUrls[] = $subUrl;
                }

                return $subUrls;
        }

        function indexSingleRecord($r,$cfgRec,$rl=NULL) {

                        // Load indexer if not yet.
                $this->loadIndexerClass();


                        // Init:
                $rl = is_array($rl) ? $rl : $this->getUidRootLineForClosestTemplate($cfgRec['pid']);
                $fieldList = t3lib_div::trimExplode(',',$cfgRec['fieldlist'],1);
                $languageField = $GLOBALS['TCA'][$cfgRec['table2index']]['ctrl']['languageField'];
                $sys_language_uid = $languageField ? $r[$languageField] : 0;

                        // (Re)-Indexing a row from a table:
                $indexerObj = &t3lib_div::makeInstance('tx_indexedsearch_indexer');
                parse_str(str_replace('###UID###',$r['uid'],$cfgRec['get_params']),$GETparams);
                $indexerObj->backend_initIndexer($cfgRec['pid'], 0, $sys_language_uid, '', $rl, $GETparams, $cfgRec['chashcalc'] ? TRUE : FALSE);
                $indexerObj->backend_setFreeIndexUid($cfgRec['uid'], $cfgRec['set_id']);
                $indexerObj->forceIndexing = TRUE;

                $theContent = '';
                foreach($fieldList as $k => $v) {
                        if (!$k)        {
                                $theTitle = $r[$v];
                        } else {
                                $theContent.= $r[$v].' ';
                        }
                }

                        // Indexing the record as a page (but with parameters set, see ->backend_setFreeIndexUid())
                $indexerObj->backend_indexAsTYPO3Page(
                        strip_tags($theTitle),
                        '',
                        '',
                        strip_tags($theContent),
                        $GLOBALS['LANG']->charSet,      // Requires that
                        $r[$GLOBALS['TCA'][$cfgRec['table2index']]['ctrl']['tstamp']],
                        $r[$GLOBALS['TCA'][$cfgRec['table2index']]['ctrl']['crdate']],
                        $r['uid']
                );

                #echo print_r($indexerObj->internal_log);
                #echo print_r($indexerObj->contentParts);
        }

        function loadIndexerClass()     {
                global $TYPO3_CONF_VARS;
                require_once(t3lib_extMgm::extPath('indexed_search').'class.indexer.php');
        }

        function getUidRootLineForClosestTemplate($id)  {
                global $TYPO3_CONF_VARS;

                require_once (PATH_t3lib."class.t3lib_page.php");
                require_once (PATH_t3lib."class.t3lib_tstemplate.php");
                require_once (PATH_t3lib."class.t3lib_tsparser_ext.php");



                $tmpl = t3lib_div::makeInstance("t3lib_tsparser_ext");
                $tmpl->tt_track = 0;    // Do not log time-performance information
                $tmpl->init();

                                // Gets the rootLine
                $sys_page = t3lib_div::makeInstance("t3lib_pageSelect");
                $rootLine = $sys_page->getRootLine($id);
                $tmpl->runThroughTemplates($rootLine,0);        // This generates the constants/config + hierarchy info for the template.

                        // Root line uids
                $rootline_uids = array();
                foreach($tmpl->rootLine as $rlkey => $rldat)    {
                        $rootline_uids[$rlkey] = $rldat['uid'];
                }

                return $rootline_uids;
        }

        function generateNextIndexingTime($cfgRec)      {
                $currentTime = time();

                        // Now, find a midnight time to use for offset calculation. This has to differ depending on whether we have frequencies within a day or more than a day; Less than a day, we don't care which day to use for offset, more than a day we want to respect the currently entered day as offset regardless of when the script is run - thus the day-of-week used in case "Weekly" is selected will be respected
                if ($cfgRec['timer_frequency']<=24*3600)        {
                        $aMidNight = mktime (0,0,0)-1*24*3600;
                } else {
                        $lastTime = $cfgRec['timer_next_indexing']?$cfgRec['timer_next_indexing']:time();
                        $aMidNight = mktime (0,0,0, date('m',$lastTime), date('d',$lastTime), date('y',$lastTime));
                }

                        // Find last offset time plus frequency in seconds:
                $lastSureOffset = $aMidNight+t3lib_div::intInRange($cfgRec['timer_offset'],0,86400);
                $frequencySeconds = t3lib_div::intInRange($cfgRec['timer_frequency'],1);

                        // Now, find out how many blocks of the length of frequency there is until the next time:
                $frequencyBlocksUntilNextTime = ceil(($currentTime-$lastSureOffset)/$frequencySeconds);

                        // Set next time to the offset + the frequencyblocks multiplied with the frequency length in seconds.
                $nextTime = $lastSureOffset + $frequencyBlocksUntilNextTime*$frequencySeconds;

                return $nextTime;
/*              echo print_r(array(
                        $cfgRec['timer_offset'],
                        $frequencySeconds,
                        date('d-m-Y H:i:s',$currentTime),
                        date('d-m-Y H:i:s',$aMidNight),
                        date('d-m-Y H:i:s',$lastSureOffset),
                        date('d-m-Y H:i:s',$nextTime)
                ));
*/      }

        function checkDeniedSuburls($url, $url_deny)    {
                if (trim($url_deny))    {
                        $url_denyArray = t3lib_div::trimExplode(chr(10),$url_deny,1);
                        foreach($url_denyArray as $testurl)     {
                                if (t3lib_div::isFirstPartOfStr($url,$testurl)) {
                                        echo $url.' 
                                        return TRUE;
                                }
                        }
                }
                return FALSE;
        }

        function addQueueEntryForHook($cfgRec, $title)  {

                $nparams = array(
                        'indexConfigUid' => $cfgRec['uid'],             // This must ALWAYS be the cfgRec uid!
                        'url' => $title,
                        'procInstructions' => array('[Index Cfg UID#'.$cfgRec['uid'].']')       // Also just for information. Its good style to show that its an indexing configuration that added the entry.
                );
                $this->pObj->addQueueEntry_callBack($cfgRec['set_id'],$nparams,$this->callBack,$cfgRec['pid']);
        }







        /*************************
         *
         * Hook functions for TCEmain (indexing of records)
         *
         *************************/

        function processDatamap_afterDatabaseOperations($status, $table, $id, $fieldArray, &$pObj) {

                        // Check if any fields are actually updated:
                if (count($fieldArray)) {

                                // Translate new ids.
                        if ($status=='new')     {
                                $id = $pObj->substNEWwithIDs[$id];
                        }

                                // Get full record and if exists, search for indexing configurations:
                        $currentRecord = t3lib_BEfunc::getRecord($table,$id);
                        if (is_array($currentRecord))   {

                                        // Select all (not running) indexing configurations of type "record" (1) and which points to this table and is located on the same page as the record or pointing to the right source PID
                                $indexingConfigurations = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
                                        '*',
                                        'index_config',
                                        'hidden=0
                                                AND (starttime=0 OR starttime<='.time().')
                                                AND set_id=0
                                                AND type=1
                                                AND table2index='.$GLOBALS['TYPO3_DB']->fullQuoteStr($table,'index_config').'
                                                AND (
                                                                (alternative_source_pid='.$GLOBALS['TYPO3_DB']->fullQuoteStr('','index_config').' AND pid='.intval($currentRecord['pid']).')
                                                                OR (alternative_source_pid='.$GLOBALS['TYPO3_DB']->fullQuoteStr($currentRecord['pid'],'index_config').')
                                                        )
                                                AND records_indexonchange=1
                                                '.t3lib_BEfunc::deleteClause('index_config')
                                );

                                foreach($indexingConfigurations as $cfgRec)     {
                                        $this->indexSingleRecord($currentRecord,$cfgRec);
                                }
                        }
                }
        }
}


class tx_indexedsearch_files {

        function crawler_execute($params,&$pObj)        {

                        // Load indexer if not yet.
                $this->loadIndexerClass();

                if (is_array($params['conf']))  {

                                // Initialize the indexer class:
                        $indexerObj = &t3lib_div::makeInstance('tx_indexedsearch_indexer');
                        $indexerObj->conf = $params['conf'];
                        $indexerObj->init();

                                // Index document:
                        if ($params['alturl'])  {
                                $fI = pathinfo($params['document']);
                                $ext = strtolower($fI['extension']);
                                $indexerObj->indexRegularDocument($params['alturl'], TRUE, $params['document'], $ext);
                        } else {
                                $indexerObj->indexRegularDocument($params['document'], TRUE);
                        }

                                // Return OK:
                        return array('content' => array());
                }
        }

        function loadIndexerClass()     {
                global $TYPO3_CONF_VARS;
                require_once(t3lib_extMgm::extPath('indexed_search').'class.indexer.php');
        }
}


if (defined('TYPO3_MODE') && $TYPO3_CONF_VARS[TYPO3_MODE]['XCLASS']['ext/indexed_search/class.crawler.php'])    {
        include_once($TYPO3_CONF_VARS[TYPO3_MODE]['XCLASS']['ext/indexed_search/class.crawler.php']);
}
?>