xref: /dokuwiki/lib/exe/indexer.php (revision 8ed45f3a4c95bf8c64968ec2afb15dcd1f7d17b3) 
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__).'/../../').'/');
9require_once(DOKU_INC.'inc/init.php');
10require_once(DOKU_INC.'inc/auth.php');
11session_write_close();  //close session
12if(!defined('NL')) define('NL',"\n");
13
14// keep running after browser closes connection
15@ignore_user_abort(true);
16
17// send gif
18sendGIF();
19
20// Catch any possible output (e.g. errors)
21if(!$_REQUEST['debug']) ob_start();
22
23// run one of the jobs
24runIndexer() or runSitemapper();
25
26if(!$_REQUEST['debug']) ob_end_clean();
27exit;
28
29// --------------------------------------------------------------------
30
31/**
32 * Runs the indexer for the current page
33 *
34 * @author Andreas Gohr <andi@splitbrain.org>
35 */
36function runIndexer(){
37    global $conf;
38    print "runIndexer(): started".NL;
39
40    $ID = cleanID($_REQUEST['id']);
41    if(!$ID) return false;
42
43    // check if indexing needed
44    $last = @filemtime(metaFN($ID,'.indexed'));
45    if($last > @filemtime(wikiFN($ID))){
46        print "runIndexer(): index for $ID up to date".NL;
47        return false;
48    }
49
50    // try to aquire a lock
51    $lock = $conf['lockdir'].'/_indexer.lock';
52    while(!@mkdir($lock,$conf['dmode'])){
53        usleep(50);
54        if(time()-@filemtime($lock) > 60*5){
55            // looks like a stale lock - remove it
56            @rmdir($lock);
57            print "runIndexer(): stale lock removed".NL;
58        }else{
59            print "runIndexer(): indexer locked".NL;
60            return false;
61        }
62    }
63    if(isset($conf['dmask'])) { chmod($lock, $conf['dmask']); }
64
65    require_once(DOKU_INC.'inc/indexer.php');
66
67    // do the work
68    idx_addPage($ID);
69
70    // we're finished - save and free lock
71    io_saveFile(metaFN($ID,'.indexed'),' ');
72    @rmdir($lock);
73    print "runIndexer(): finished".NL;
74    return true;
75}
76
77/**
78 * Builds a Google Sitemap of all public pages known to the indexer
79 *
80 * The map is placed in the root directory named sitemap.xml.gz - This
81 * file needs to be writable!
82 *
83 * @author Andreas Gohr
84 * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
85 */
86function runSitemapper(){
87    global $conf;
88    print "runSitemapper(): started".NL;
89    if(!$conf['sitemap']) return false;
90
91    if($conf['usegzip']){
92        $sitemap = DOKU_INC.'sitemap.xml.gz';
93    }else{
94        $sitemap = DOKU_INC.'sitemap.xml';
95    }
96    print "runSitemapper(): using $sitemap".NL;
97
98    if(!is_writable($sitemap)) return false;
99    if(@filesize($sitemap) &&
100       @filemtime($sitemap) > (time()-($conf['sitemap']*60*60*24))){
101       print 'runSitemapper(): Sitemap up to date'.NL;
102       return false;
103    }
104
105    $pages = file($conf['cachedir'].'/page.idx');
106    print 'runSitemapper(): creating sitemap using '.count($pages).' pages'.NL;
107
108    // build the sitemap
109    ob_start();
110    print '<?xml version="1.0" encoding="UTF-8"?>'.NL;
111    print '<urlset xmlns="http://www.google.com/schemas/sitemap/0.84">'.NL;
112    foreach($pages as $id){
113        $id = trim($id);
114        $file = wikiFN($id);
115
116        //skip hidden, non existing and restricted files
117        if(isHiddenPage($id)) return false;
118        $date = @filemtime($file);
119        if(!$date) continue;
120        if(auth_aclcheck($id,'','') < AUTH_READ) continue;
121
122        print '  <url>'.NL;
123        print '    <loc>'.wl($id,'',true).'</loc>'.NL;
124        print '    <lastmod>'.date_iso8601($date).'</lastmod>'.NL;
125        print '  </url>'.NL;
126    }
127    print '</urlset>'.NL;
128    $data = ob_get_contents();
129    ob_end_clean();
130
131    //save the new sitemap
132    io_saveFile($sitemap,$data);
133
134    print 'runSitemapper(): pinging google'.NL;
135    //ping google
136    $url  = 'http://www.google.com/webmasters/sitemaps/ping?sitemap=';
137    $url .= urlencode(DOKU_URL.$sitemap);
138    $http = new DokuHTTPClient();
139    $http->get($url);
140    if($http->error) print 'runSitemapper(): '.$http->error.NL;
141
142    print 'runSitemapper(): finished'.NL;
143    return true;
144}
145
146/**
147 * Formats a timestamp as ISO 8601 date
148 *
149 * @author <ungu at terong dot com>
150 * @link http://www.php.net/manual/en/function.date.php#54072
151 */
152function date_iso8601($int_date) {
153   //$int_date: current date in UNIX timestamp
154   $date_mod = date('Y-m-d\TH:i:s', $int_date);
155   $pre_timezone = date('O', $int_date);
156   $time_zone = substr($pre_timezone, 0, 3).":".substr($pre_timezone, 3, 2);
157   $date_mod .= $time_zone;
158   return $date_mod;
159}
160
161/**
162 * Just send a 1x1 pixel blank gif to the browser
163 *
164 * @author Andreas Gohr <andi@splitbrain.org>
165 * @author Harry Fuecks <fuecks@gmail.com>
166 */
167function sendGIF(){
168    if($_REQUEST['debug']){
169        header('Content-Type: text/plain');
170        return;
171    }
172    $img = base64_decode('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAEALAAAAAABAAEAAAIBTAA7');
173    header('Content-Type: image/gif');
174    header('Content-Length: '.strlen($img));
175    header('Connection: Close');
176    print $img;
177    flush();
178    // Browser should drop connection after this
179    // Thinks it's got the whole image
180}
181
182//Setup VIM: ex: et ts=4 enc=utf-8 :
183// No trailing PHP closing tag - no output please!
184// See Note at http://www.php.net/manual/en/language.basic-syntax.instruction-separation.php
185