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