xref: /dokuwiki/lib/exe/indexer.php (revision 40b33eff69a8a44da0ab6962d50f2027bcfc33b6) 
1<?php
2/**
3 * DokuWiki indexer
4 *
5 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6 * @author     Andreas Gohr <andi@splitbrain.org>
7 */
8if(!defined('DOKU_INC')) define('DOKU_INC',realpath(dirname(__FILE__).'/../../').'/');
9define('DOKU_DISABLE_GZIP_OUTPUT',1);
10require_once(DOKU_INC.'inc/init.php');
11require_once(DOKU_INC.'inc/auth.php');
12require_once(DOKU_INC.'inc/events.php');
13session_write_close();  //close session
14if(!defined('NL')) define('NL',"\n");
15
16// keep running after browser closes connection
17@ignore_user_abort(true);
18
19// check if user abort worked, if yes send output early
20if(@ignore_user_abort()){
21    sendGIF(); // send gif
22    $defer = false;
23}else{
24    $defer = true;
25}
26
27// Catch any possible output (e.g. errors)
28if(!$_REQUEST['debug']) ob_start();
29
30// run one of the jobs
31runIndexer() or metaUpdate() or runSitemapper() or runTrimRecentChanges();
32if($defer) sendGIF();
33
34if(!$_REQUEST['debug']) ob_end_clean();
35exit;
36
37// --------------------------------------------------------------------
38
39/**
40 * Trims the recent changes cache (or imports the old changelog) as needed.
41 *
42 * @author Ben Coburn <btcoburn@silicodon.net>
43 */
44function runTrimRecentChanges() {
45    global $conf;
46
47    // Import old changelog (if needed)
48    // Uses the imporoldchangelog plugin to upgrade the changelog automaticaly.
49    // FIXME: Remove this from runTrimRecentChanges when it is no longer needed.
50    if (isset($conf['changelog_old']) &&
51        @file_exists($conf['changelog_old']) && !@file_exists($conf['changelog']) &&
52        !@file_exists($conf['changelog'].'_importing') && !@file_exists($conf['changelog'].'_tmp')) {
53            $tmp = array(); // no event data
54            trigger_event('TEMPORARY_CHANGELOG_UPGRADE_EVENT', $tmp);
55            return true;
56    }
57
58    // Trim the Recent Changes
59    // Trims the recent changes cache to the last $conf['changes_days'] recent
60    // changes or $conf['recent'] items, which ever is larger.
61    // The trimming is only done once a day.
62    if (@file_exists($conf['changelog']) &&
63        (filectime($conf['changelog'])+86400)<time() &&
64        !@file_exists($conf['changelog'].'_tmp')) {
65
66            io_lock($conf['changelog']);
67            $lines = file($conf['changelog']);
68            if (count($lines)<$conf['recent']) {
69                // nothing to trim
70                io_unlock($conf['changelog']);
71                return true;
72            }
73
74            io_saveFile($conf['changelog'].'_tmp', ''); // presave tmp as 2nd lock
75            $trim_time = time() - $conf['recent_days']*86400;
76            $out_lines = array();
77
78            for ($i=0; $i<count($lines); $i++) {
79              $log = parseChangelogLine($lines[$i]);
80              if ($log === false || $log['date'] < $trim_time) continue;  // discard old lines
81              $out_lines[$log['date']] = $lines[$i];                      // preserve the rest
82            }
83
84            ksort($out_lines);                                            // sort lines, just in case!
85            if (count($out_lines) > $conf['recent']) {
86              $out_lines = array_slice($out_lines,-$conf['recent']);      // trim list to one page
87            }
88
89            // save trimmed changelog
90            io_saveFile($conf['changelog'].'_tmp', implode('', $out_lines));
91            @unlink($conf['changelog']);
92            if (!rename($conf['changelog'].'_tmp', $conf['changelog'])) {
93                // rename failed so try another way...
94                io_unlock($conf['changelog']);
95                io_saveFile($conf['changelog'], implode('', $out_lines));
96                @unlink($conf['changelog'].'_tmp');
97            } else {
98                io_unlock($conf['changelog']);
99            }
100            return true;
101    }
102
103    // nothing done
104    return false;
105}
106
107/**
108 * Runs the indexer for the current page
109 *
110 * @author Andreas Gohr <andi@splitbrain.org>
111 */
112function runIndexer(){
113    global $conf;
114    print "runIndexer(): started".NL;
115
116    $ID = cleanID($_REQUEST['id']);
117    if(!$ID) return false;
118
119    // check if indexing needed
120    $last = @filemtime(metaFN($ID,'.indexed'));
121    if($last > @filemtime(wikiFN($ID))){
122        print "runIndexer(): index for $ID up to date".NL;
123        return false;
124    }
125
126    // try to aquire a lock
127    $lock = $conf['lockdir'].'/_indexer.lock';
128    while(!@mkdir($lock,$conf['dmode'])){
129        usleep(50);
130        if(time()-@filemtime($lock) > 60*5){
131            // looks like a stale lock - remove it
132            @rmdir($lock);
133            print "runIndexer(): stale lock removed".NL;
134        }else{
135            print "runIndexer(): indexer locked".NL;
136            return false;
137        }
138    }
139    if($conf['dperm']) chmod($lock, $conf['dperm']);
140
141    require_once(DOKU_INC.'inc/indexer.php');
142
143    // do the work
144    idx_addPage($ID);
145
146    // we're finished - save and free lock
147    io_saveFile(metaFN($ID,'.indexed'),' ');
148    @rmdir($lock);
149    print "runIndexer(): finished".NL;
150    return true;
151}
152
153/**
154 * Will render the metadata for the page if not exists yet
155 *
156 * This makes sure pages which are created from outside DokuWiki will
157 * gain their data when viewed for the first time.
158 */
159function metaUpdate(){
160    print "metaUpdate(): started".NL;
161
162    $ID = cleanID($_REQUEST['id']);
163    if(!$ID) return false;
164    $file = metaFN($ID, '.meta');
165    echo "meta file: $file".NL;
166
167    // rendering needed?
168    if (@file_exists($file)) return false;
169    if (!@file_exists(wikiFN($ID))) return false;
170
171    require_once(DOKU_INC.'inc/common.php');
172    require_once(DOKU_INC.'inc/parserutils.php');
173    global $conf;
174
175
176    // gather some additional info from changelog
177    $info = io_grep($conf['changelog'],
178                    '/^(\d+)\t(\d+\.\d+\.\d+\.\d+)\t'.preg_quote($ID,'/').'\t([^\t]+)\t([^\t\n]+)/',
179                    0,true);
180
181    $meta = array();
182    if(count($info)){
183        $meta['date']['created'] = $info[0][1];
184        foreach($info as $item){
185            if($item[4] != '*'){
186                $meta['date']['modified'] = $item[1];
187                if($item[3]){
188                    $meta['contributor'][$item[3]] = $item[3];
189                }
190            }
191        }
192    }
193
194    $meta = p_render_metadata($ID, $meta);
195    io_saveFile($file, serialize($meta));
196
197    echo "metaUpdate(): finished".NL;
198    return 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 */
210function runSitemapper(){
211    global $conf;
212    print "runSitemapper(): started".NL;
213    if(!$conf['sitemap']) return false;
214
215    if($conf['usegzip']){
216        $sitemap = 'sitemap.xml.gz';
217    }else{
218        $sitemap = 'sitemap.xml';
219    }
220    print "runSitemapper(): using $sitemap".NL;
221
222    if(!is_writable(DOKU_INC.$sitemap)) return false;
223    if(@filesize(DOKU_INC.$sitemap) &&
224       @filemtime(DOKU_INC.$sitemap) > (time()-($conf['sitemap']*60*60*24))){
225       print 'runSitemapper(): Sitemap up to date'.NL;
226       return false;
227    }
228
229    $pages = file($conf['cachedir'].'/page.idx');
230    print 'runSitemapper(): creating sitemap using '.count($pages).' pages'.NL;
231
232    // build the sitemap
233    ob_start();
234    print '<?xml version="1.0" encoding="UTF-8"?>'.NL;
235    print '<urlset xmlns="http://www.google.com/schemas/sitemap/0.84">'.NL;
236    foreach($pages as $id){
237        $id = trim($id);
238        $file = wikiFN($id);
239
240        //skip hidden, non existing and restricted files
241        if(isHiddenPage($id)) continue;
242        $date = @filemtime($file);
243        if(!$date) continue;
244        if(auth_aclcheck($id,'','') < AUTH_READ) continue;
245
246        print '  <url>'.NL;
247        print '    <loc>'.wl($id,'',true).'</loc>'.NL;
248        print '    <lastmod>'.date_iso8601($date).'</lastmod>'.NL;
249        print '  </url>'.NL;
250    }
251    print '</urlset>'.NL;
252    $data = ob_get_contents();
253    ob_end_clean();
254
255    //save the new sitemap
256    io_saveFile(DOKU_INC.$sitemap,$data);
257
258    print 'runSitemapper(): pinging google'.NL;
259    //ping google
260    $url  = 'http://www.google.com/webmasters/sitemaps/ping?sitemap=';
261    $url .= urlencode(DOKU_URL.$sitemap);
262    $http = new DokuHTTPClient();
263    $http->get($url);
264    if($http->error) print 'runSitemapper(): '.$http->error.NL;
265
266    print 'runSitemapper(): finished'.NL;
267    return true;
268}
269
270/**
271 * Formats a timestamp as ISO 8601 date
272 *
273 * @author <ungu at terong dot com>
274 * @link http://www.php.net/manual/en/function.date.php#54072
275 */
276function date_iso8601($int_date) {
277   //$int_date: current date in UNIX timestamp
278   $date_mod = date('Y-m-d\TH:i:s', $int_date);
279   $pre_timezone = date('O', $int_date);
280   $time_zone = substr($pre_timezone, 0, 3).":".substr($pre_timezone, 3, 2);
281   $date_mod .= $time_zone;
282   return $date_mod;
283}
284
285/**
286 * Just send a 1x1 pixel blank gif to the browser
287 *
288 * @author Andreas Gohr <andi@splitbrain.org>
289 * @author Harry Fuecks <fuecks@gmail.com>
290 */
291function sendGIF(){
292    if($_REQUEST['debug']){
293        header('Content-Type: text/plain');
294        return;
295    }
296    $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7');
297    header('Content-Type: image/gif');
298    header('Content-Length: '.strlen($img));
299    header('Connection: Close');
300    print $img;
301    flush();
302    // Browser should drop connection after this
303    // Thinks it's got the whole image
304}
305
306//Setup VIM: ex: et ts=4 enc=utf-8 :
307// No trailing PHP closing tag - no output please!
308// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php
309