xref: /dokuwiki/inc/TaskRunner.php (revision 3ad4c3cdc8d3bebaa2a9bc20d293c0af438f4923) 
1<?php
2
3namespace dokuwiki;
4
5use Doku_Event;
6use Sitemapper;
7use Subscription;
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 Doku_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 $media_changes   If the media changelog shall be trimmed instead of
87     *                              the page changelog
88     *
89     * @return bool
90     *
91     * @author Ben Coburn <btcoburn@silicodon.net>
92     */
93    protected function runTrimRecentChanges($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;
126                }                      // discard junk
127                if ($log['date'] < $trim_time) {
128                    $old_lines[$log['date'] . ".$i"] = $lines[$i];     // keep old lines for now (append .$i to prevent key collisions)
129                } else {
130                    $out_lines[$log['date'] . ".$i"] = $lines[$i];     // definitely keep these lines
131                }
132            }
133
134            if (count($lines) == count($out_lines)) {
135                // nothing to trim
136                @unlink($fn . '_tmp');
137                io_unlock($fn);
138                echo "runTrimRecentChanges($media_changes): finished" . NL;
139                return false;
140            }
141
142            // sort the final result, it shouldn't be necessary,
143            //   however the extra robustness in making the changelog cache self-correcting is worth it
144            ksort($out_lines);
145            $extra = $conf['recent'] - count($out_lines);        // do we need extra lines do bring us up to minimum
146            if ($extra > 0) {
147                ksort($old_lines);
148                $out_lines = array_merge(array_slice($old_lines, -$extra), $out_lines);
149            }
150
151            $eventData = [
152                'trimmedChangelogLines' => $out_lines,
153                'removedChangelogLines' => $extra > 0 ? array_slice($old_lines, 0, -$extra) : $old_lines,
154            ];
155            trigger_event('TRIM_RECENT_CHANGES', $eventData);
156            $out_lines = $eventData['trimmedChangelogLines'];
157
158            // save trimmed changelog
159            io_saveFile($fn . '_tmp', implode('', $out_lines));
160            @unlink($fn);
161            if (!rename($fn . '_tmp', $fn)) {
162                // rename failed so try another way...
163                io_unlock($fn);
164                io_saveFile($fn, implode('', $out_lines));
165                @unlink($fn . '_tmp');
166            } else {
167                io_unlock($fn);
168            }
169            echo "runTrimRecentChanges($media_changes): finished" . NL;
170            return true;
171        }
172
173        // nothing done
174        echo "runTrimRecentChanges($media_changes): finished" . NL;
175        return false;
176    }
177
178
179    /**
180     * Runs the indexer for the current page
181     *
182     * @author Andreas Gohr <andi@splitbrain.org>
183     */
184    protected function runIndexer()
185    {
186        global $ID;
187        global $conf;
188        print 'runIndexer(): started' . NL;
189
190        if (!$ID) {
191            return false;
192        }
193
194        // do the work
195        return idx_addPage($ID, true);
196    }
197
198    /**
199     * Builds a Google Sitemap of all public pages known to the indexer
200     *
201     * The map is placed in the root directory named sitemap.xml.gz - This
202     * file needs to be writable!
203     *
204     * @author Andreas Gohr
205     * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
206     */
207    protected function runSitemapper()
208    {
209        print 'runSitemapper(): started' . NL;
210        $result = Sitemapper::generate() && Sitemapper::pingSearchEngines();
211        print 'runSitemapper(): finished' . NL;
212        return $result;
213    }
214
215    /**
216     * Send digest and list mails for all subscriptions which are in effect for the
217     * current page
218     *
219     * @author Adrian Lang <lang@cosmocode.de>
220     */
221    protected function sendDigest()
222    {
223        global $conf;
224        global $ID;
225
226        echo 'sendDigest(): started' . NL;
227        if (!actionOK('subscribe')) {
228            echo 'sendDigest(): disabled' . NL;
229            return false;
230        }
231        $sub = new Subscription();
232        $sent = $sub->send_bulk($ID);
233
234        echo "sendDigest(): sent $sent mails" . NL;
235        echo 'sendDigest(): finished' . NL;
236        return (bool)$sent;
237    }
238}
239