inc/TaskRunner.php

   1 <?php
   2
   3 namespace dokuwiki;
   4
   5 use dokuwiki\Extension\Event;
   6 use dokuwiki\Sitemap\Mapper;
   7 use dokuwiki\Subscriptions\BulkSubscriptionSender;
   8 use dokuwiki\ChangeLog\ChangeLog;
   9
  10 /**
  11  * Class TaskRunner
  12  *
  13  * Run an asynchronous task.
  14  */
  15 class TaskRunner
  16 {
  17     /**
  18      * Run the next task
  19      *
  20      * @todo refactor to remove dependencies on globals
  21      * @triggers INDEXER_TASKS_RUN
  22      */
  23     public function run()
  24     {
  25         global $INPUT, $conf, $ID;
  26
  27         // keep running after browser closes connection
  28         @ignore_user_abort(true);
  29
  30         // check if user abort worked, if yes send output early
  31         $defer = !@ignore_user_abort() || $conf['broken_iua'];
  32         $output = $INPUT->has('debug') && $conf['allowdebug'];
  33         if (!$defer && !$output) {
  34             $this->sendGIF();
  35         }
  36
  37         $ID = cleanID($INPUT->str('id'));
  38
  39         // Catch any possible output (e.g. errors)
  40         if (!$output) {
  41             ob_start();
  42         } else {
  43             header('Content-Type: text/plain');
  44         }
  45
  46         // run one of the jobs
  47         $tmp = []; // No event data
  48         $evt = new Event('INDEXER_TASKS_RUN', $tmp);
  49         if ($evt->advise_before()) {
  50             if (
  51                 !(
  52                 $this->runIndexer() ||
  53                 $this->runSitemapper() ||
  54                 $this->sendDigest() ||
  55                 $this->runTrimRecentChanges() ||
  56                 $this->runTrimRecentChanges(true))
  57             ) {
  58                 $evt->advise_after();
  59             }
  60         }
  61
  62         if (!$output) {
  63             ob_end_clean();
  64             if ($defer) {
  65                 $this->sendGIF();
  66             }
  67         }
  68     }
  69
  70     /**
  71      * Just send a 1x1 pixel blank gif to the browser
  72      *
  73      * @author Andreas Gohr <andi@splitbrain.org>
  74      * @author Harry Fuecks <fuecks@gmail.com>
  75      */
  76     protected function sendGIF()
  77     {
  78         $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7');
  79         header('Content-Type: image/gif');
  80         header('Content-Length: ' . strlen($img));
  81         header('Connection: Close');
  82         echo $img;
  83         tpl_flush();
  84         // Browser should drop connection after this
  85         // Thinks it's got the whole image
  86     }
  87
  88     /**
  89      * Trims the recent changes cache (or imports the old changelog) as needed.
  90      *
  91      * @param bool $media_changes   If the media changelog shall be trimmed instead of
  92      *                              the page changelog
  93      *
  94      * @return bool
  95      * @triggers TASK_RECENTCHANGES_TRIM
  96      * @author Ben Coburn <btcoburn@silicodon.net>
  97      */
  98     protected function runTrimRecentChanges($media_changes = false)
  99     {
 100         global $conf;
 101
 102         echo "runTrimRecentChanges($media_changes): started" . NL;
 103
 104         $fn = ($media_changes ? $conf['media_changelog'] : $conf['changelog']);
 105
 106         // Trim the Recent Changes
 107         // Trims the recent changes cache to the last $conf['changes_days'] recent
 108         // changes or $conf['recent'] items, which ever is larger.
 109         // The trimming is only done once a day.
 110         if (
 111             file_exists($fn) &&
 112             (@filemtime($fn . '.trimmed') + 86400) < time() &&
 113             !file_exists($fn . '_tmp')
 114         ) {
 115             @touch($fn . '.trimmed');
 116             io_lock($fn);
 117             $lines = file($fn);
 118             if (count($lines) <= $conf['recent']) {
 119                 // nothing to trim
 120                 io_unlock($fn);
 121                 echo "runTrimRecentChanges($media_changes): finished" . NL;
 122                 return false;
 123             }
 124
 125             io_saveFile($fn . '_tmp', '');          // presave tmp as 2nd lock
 126             $trim_time = time() - $conf['recent_days'] * 86400;
 127             $out_lines = [];
 128             $old_lines = [];
 129             $counter = count($lines);
 130             for ($i = 0; $i < $counter; $i++) {
 131                 $log = ChangeLog::parseLogLine($lines[$i]);
 132                 if ($log === false) {
 133                     continue; // discard junk
 134                 }
 135
 136                 if ($log['date'] < $trim_time) {
 137                     // keep old lines for now (append .$i to prevent key collisions)
 138                     $old_lines[$log['date'] . ".$i"] = $lines[$i];
 139                 } else {
 140                     // definitely keep these lines
 141                     $out_lines[$log['date'] . ".$i"] = $lines[$i];
 142                 }
 143             }
 144
 145             if (count($lines) === count($out_lines)) {
 146                 // nothing to trim
 147                 @unlink($fn . '_tmp');
 148                 io_unlock($fn);
 149                 echo "runTrimRecentChanges($media_changes): finished" . NL;
 150                 return false;
 151             }
 152
 153             // sort the final result, it shouldn't be necessary,
 154             //   however the extra robustness in making the changelog cache self-correcting is worth it
 155             ksort($out_lines);
 156             $extra = $conf['recent'] - count($out_lines);        // do we need extra lines do bring us up to minimum
 157             if ($extra > 0) {
 158                 ksort($old_lines);
 159                 $out_lines = array_merge(array_slice($old_lines, -$extra), $out_lines);
 160             }
 161
 162             $eventData = [
 163                 'isMedia' => $media_changes,
 164                 'trimmedChangelogLines' => $out_lines,
 165                 'removedChangelogLines' => $extra > 0 ? array_slice($old_lines, 0, -$extra) : $old_lines,
 166             ];
 167             Event::createAndTrigger('TASK_RECENTCHANGES_TRIM', $eventData);
 168             $out_lines = $eventData['trimmedChangelogLines'];
 169
 170             // save trimmed changelog
 171             io_saveFile($fn . '_tmp', implode('', $out_lines));
 172             @unlink($fn);
 173             if (!rename($fn . '_tmp', $fn)) {
 174                 // rename failed so try another way...
 175                 io_unlock($fn);
 176                 io_saveFile($fn, implode('', $out_lines));
 177                 @unlink($fn . '_tmp');
 178             } else {
 179                 io_unlock($fn);
 180             }
 181             echo "runTrimRecentChanges($media_changes): finished" . NL;
 182             return true;
 183         }
 184
 185         // nothing done
 186         echo "runTrimRecentChanges($media_changes): finished" . NL;
 187         return false;
 188     }
 189
 190
 191     /**
 192      * Runs the indexer for the current page
 193      *
 194      * @author Andreas Gohr <andi@splitbrain.org>
 195      */
 196     protected function runIndexer()
 197     {
 198         global $ID;
 199         echo 'runIndexer(): started' . NL;
 200
 201         if ((string) $ID === '') {
 202             return false;
 203         }
 204
 205         // do the work
 206         return idx_addPage($ID, true);
 207     }
 208
 209     /**
 210      * Builds a Google Sitemap of all public pages known to the indexer
 211      *
 212      * The map is placed in the root directory named sitemap.xml.gz - This
 213      * file needs to be writable!
 214      *
 215      * @author Andreas Gohr
 216      * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
 217      */
 218     protected function runSitemapper()
 219     {
 220         echo 'runSitemapper(): started' . NL;
 221         $result = Mapper::generate() && Mapper::pingSearchEngines();
 222         echo 'runSitemapper(): finished' . NL;
 223         return $result;
 224     }
 225
 226     /**
 227      * Send digest and list mails for all subscriptions which are in effect for the
 228      * current page
 229      *
 230      * @author Adrian Lang <lang@cosmocode.de>
 231      */
 232     protected function sendDigest()
 233     {
 234         global $ID;
 235
 236         echo 'sendDigest(): started' . NL;
 237         if (!actionOK('subscribe')) {
 238             echo 'sendDigest(): disabled' . NL;
 239             return false;
 240         }
 241         $sub = new BulkSubscriptionSender();
 242         $sent = $sub->sendBulk($ID);
 243
 244         echo "sendDigest(): sent $sent mails" . NL;
 245         echo 'sendDigest(): finished' . NL;
 246         return (bool)$sent;
 247     }
 248 }