xref: /dokuwiki/inc/Sitemap/Mapper.php (revision 432adb37131e09da3a60af09344c423727a7e2ad)
1*432adb37SAndreas Gohr<?php
2*432adb37SAndreas Gohr/**
3*432adb37SAndreas Gohr * Sitemap handling functions
4*432adb37SAndreas Gohr *
5*432adb37SAndreas Gohr * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6*432adb37SAndreas Gohr * @author     Michael Hamann <michael@content-space.de>
7*432adb37SAndreas Gohr */
8*432adb37SAndreas Gohr
9*432adb37SAndreas Gohrnamespace dokuwiki\Sitemap;
10*432adb37SAndreas Gohr
11*432adb37SAndreas Gohr/**
12*432adb37SAndreas Gohr * A class for building sitemaps and pinging search engines with the sitemap URL.
13*432adb37SAndreas Gohr *
14*432adb37SAndreas Gohr * @author Michael Hamann
15*432adb37SAndreas Gohr */
16*432adb37SAndreas Gohrclass Mapper {
17*432adb37SAndreas Gohr    /**
18*432adb37SAndreas Gohr     * Builds a Google Sitemap of all public pages known to the indexer
19*432adb37SAndreas Gohr     *
20*432adb37SAndreas Gohr     * The map is placed in the cache directory named sitemap.xml.gz - This
21*432adb37SAndreas Gohr     * file needs to be writable!
22*432adb37SAndreas Gohr     *
23*432adb37SAndreas Gohr     * @author Michael Hamann
24*432adb37SAndreas Gohr     * @author Andreas Gohr
25*432adb37SAndreas Gohr     * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
26*432adb37SAndreas Gohr     * @link   http://www.sitemaps.org/
27*432adb37SAndreas Gohr     *
28*432adb37SAndreas Gohr     * @return bool
29*432adb37SAndreas Gohr     */
30*432adb37SAndreas Gohr    public static function generate(){
31*432adb37SAndreas Gohr        global $conf;
32*432adb37SAndreas Gohr        if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
33*432adb37SAndreas Gohr
34*432adb37SAndreas Gohr        $sitemap = Mapper::getFilePath();
35*432adb37SAndreas Gohr
36*432adb37SAndreas Gohr        if(file_exists($sitemap)){
37*432adb37SAndreas Gohr            if(!is_writable($sitemap)) return false;
38*432adb37SAndreas Gohr        }else{
39*432adb37SAndreas Gohr            if(!is_writable(dirname($sitemap))) return false;
40*432adb37SAndreas Gohr        }
41*432adb37SAndreas Gohr
42*432adb37SAndreas Gohr        if(@filesize($sitemap) &&
43*432adb37SAndreas Gohr           @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
44*432adb37SAndreas Gohr            dbglog('Sitemapper::generate(): Sitemap up to date');
45*432adb37SAndreas Gohr            return false;
46*432adb37SAndreas Gohr        }
47*432adb37SAndreas Gohr
48*432adb37SAndreas Gohr        dbglog("Sitemapper::generate(): using $sitemap");
49*432adb37SAndreas Gohr
50*432adb37SAndreas Gohr        $pages = idx_get_indexer()->getPages();
51*432adb37SAndreas Gohr        dbglog('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
52*432adb37SAndreas Gohr        $items = array();
53*432adb37SAndreas Gohr
54*432adb37SAndreas Gohr        // build the sitemap items
55*432adb37SAndreas Gohr        foreach($pages as $id){
56*432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
57*432adb37SAndreas Gohr            if(isHiddenPage($id)) continue;
58*432adb37SAndreas Gohr            if(auth_aclcheck($id,'',array()) < AUTH_READ) continue;
59*432adb37SAndreas Gohr            $item = Item::createFromID($id);
60*432adb37SAndreas Gohr            if ($item !== null)
61*432adb37SAndreas Gohr                $items[] = $item;
62*432adb37SAndreas Gohr        }
63*432adb37SAndreas Gohr
64*432adb37SAndreas Gohr        $eventData = array('items' => &$items, 'sitemap' => &$sitemap);
65*432adb37SAndreas Gohr        $event = new \Doku_Event('SITEMAP_GENERATE', $eventData);
66*432adb37SAndreas Gohr        if ($event->advise_before(true)) {
67*432adb37SAndreas Gohr            //save the new sitemap
68*432adb37SAndreas Gohr            $event->result = io_saveFile($sitemap, Mapper::getXML($items));
69*432adb37SAndreas Gohr        }
70*432adb37SAndreas Gohr        $event->advise_after();
71*432adb37SAndreas Gohr
72*432adb37SAndreas Gohr        return $event->result;
73*432adb37SAndreas Gohr    }
74*432adb37SAndreas Gohr
75*432adb37SAndreas Gohr    /**
76*432adb37SAndreas Gohr     * Builds the sitemap XML string from the given array auf SitemapItems.
77*432adb37SAndreas Gohr     *
78*432adb37SAndreas Gohr     * @param $items array The SitemapItems that shall be included in the sitemap.
79*432adb37SAndreas Gohr     * @return string The sitemap XML.
80*432adb37SAndreas Gohr     *
81*432adb37SAndreas Gohr     * @author Michael Hamann
82*432adb37SAndreas Gohr     */
83*432adb37SAndreas Gohr    private static function getXML($items) {
84*432adb37SAndreas Gohr        ob_start();
85*432adb37SAndreas Gohr        echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
86*432adb37SAndreas Gohr        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
87*432adb37SAndreas Gohr        foreach ($items as $item) {
88*432adb37SAndreas Gohr            /** @var Item $item */
89*432adb37SAndreas Gohr            echo $item->toXML();
90*432adb37SAndreas Gohr        }
91*432adb37SAndreas Gohr        echo '</urlset>'.NL;
92*432adb37SAndreas Gohr        $result = ob_get_contents();
93*432adb37SAndreas Gohr        ob_end_clean();
94*432adb37SAndreas Gohr        return $result;
95*432adb37SAndreas Gohr    }
96*432adb37SAndreas Gohr
97*432adb37SAndreas Gohr    /**
98*432adb37SAndreas Gohr     * Helper function for getting the path to the sitemap file.
99*432adb37SAndreas Gohr     *
100*432adb37SAndreas Gohr     * @return string The path to the sitemap file.
101*432adb37SAndreas Gohr     *
102*432adb37SAndreas Gohr     * @author Michael Hamann
103*432adb37SAndreas Gohr     */
104*432adb37SAndreas Gohr    public static function getFilePath() {
105*432adb37SAndreas Gohr        global $conf;
106*432adb37SAndreas Gohr
107*432adb37SAndreas Gohr        $sitemap = $conf['cachedir'].'/sitemap.xml';
108*432adb37SAndreas Gohr        if (self::sitemapIsCompressed()) {
109*432adb37SAndreas Gohr            $sitemap .= '.gz';
110*432adb37SAndreas Gohr        }
111*432adb37SAndreas Gohr
112*432adb37SAndreas Gohr        return $sitemap;
113*432adb37SAndreas Gohr    }
114*432adb37SAndreas Gohr
115*432adb37SAndreas Gohr    /**
116*432adb37SAndreas Gohr     * Helper function for checking if the sitemap is compressed
117*432adb37SAndreas Gohr     *
118*432adb37SAndreas Gohr     * @return bool If the sitemap file is compressed
119*432adb37SAndreas Gohr     */
120*432adb37SAndreas Gohr    public static function sitemapIsCompressed() {
121*432adb37SAndreas Gohr        global $conf;
122*432adb37SAndreas Gohr        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
123*432adb37SAndreas Gohr    }
124*432adb37SAndreas Gohr
125*432adb37SAndreas Gohr    /**
126*432adb37SAndreas Gohr     * Pings search engines with the sitemap url. Plugins can add or remove
127*432adb37SAndreas Gohr     * urls to ping using the SITEMAP_PING event.
128*432adb37SAndreas Gohr     *
129*432adb37SAndreas Gohr     * @author Michael Hamann
130*432adb37SAndreas Gohr     *
131*432adb37SAndreas Gohr     * @return bool
132*432adb37SAndreas Gohr     */
133*432adb37SAndreas Gohr    public static function pingSearchEngines() {
134*432adb37SAndreas Gohr        //ping search engines...
135*432adb37SAndreas Gohr        $http = new \DokuHTTPClient();
136*432adb37SAndreas Gohr        $http->timeout = 8;
137*432adb37SAndreas Gohr
138*432adb37SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', array('do' => 'sitemap'), true, '&'));
139*432adb37SAndreas Gohr        $ping_urls = array(
140*432adb37SAndreas Gohr            'google'    => 'http://www.google.com/webmasters/sitemaps/ping?sitemap='.$encoded_sitemap_url,
141*432adb37SAndreas Gohr            'microsoft' => 'http://www.bing.com/webmaster/ping.aspx?siteMap='.$encoded_sitemap_url,
142*432adb37SAndreas Gohr            'yandex'    => 'http://blogs.yandex.ru/pings/?status=success&url='.$encoded_sitemap_url
143*432adb37SAndreas Gohr        );
144*432adb37SAndreas Gohr
145*432adb37SAndreas Gohr        $data = array('ping_urls' => $ping_urls,
146*432adb37SAndreas Gohr                            'encoded_sitemap_url' => $encoded_sitemap_url
147*432adb37SAndreas Gohr        );
148*432adb37SAndreas Gohr        $event = new \Doku_Event('SITEMAP_PING', $data);
149*432adb37SAndreas Gohr        if ($event->advise_before(true)) {
150*432adb37SAndreas Gohr            foreach ($data['ping_urls'] as $name => $url) {
151*432adb37SAndreas Gohr                dbglog("Sitemapper::PingSearchEngines(): pinging $name");
152*432adb37SAndreas Gohr                $resp = $http->get($url);
153*432adb37SAndreas Gohr                if($http->error) dbglog("Sitemapper:pingSearchengines(): $http->error");
154*432adb37SAndreas Gohr                dbglog('Sitemapper:pingSearchengines(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp)));
155*432adb37SAndreas Gohr            }
156*432adb37SAndreas Gohr        }
157*432adb37SAndreas Gohr        $event->advise_after();
158*432adb37SAndreas Gohr
159*432adb37SAndreas Gohr        return true;
160*432adb37SAndreas Gohr    }
161*432adb37SAndreas Gohr}
162*432adb37SAndreas Gohr
163