xref: /dokuwiki/inc/Sitemap/Mapper.php (revision 8c7c53b0321a3cd3116b8d3b2ad27863a38dece7) 
1<?php
2/**
3 * Sitemap handling functions
4 *
5 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6 * @author     Michael Hamann <michael@content-space.de>
7 */
8
9namespace dokuwiki\Sitemap;
10
11use dokuwiki\Extension\Event;
12use dokuwiki\HTTP\DokuHTTPClient;
13use dokuwiki\Logger;
14
15/**
16 * A class for building sitemaps and pinging search engines with the sitemap URL.
17 *
18 * @author Michael Hamann
19 */
20class Mapper
21{
22    /**
23     * Builds a Google Sitemap of all public pages known to the indexer
24     *
25     * The map is placed in the cache directory named sitemap.xml.gz - This
26     * file needs to be writable!
27     *
28     * @author Michael Hamann
29     * @author Andreas Gohr
30     * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
31     * @link   http://www.sitemaps.org/
32     *
33     * @return bool
34     */
35    public static function generate(){
36        global $conf;
37        if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
38
39        $sitemap = Mapper::getFilePath();
40
41        if (file_exists($sitemap)) {
42            if(!is_writable($sitemap)) return false;
43        } elseif (!is_writable(dirname($sitemap))) {
44            return false;
45        }
46
47        if(@filesize($sitemap) &&
48           @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
49            Logger::debug('Sitemapper::generate(): Sitemap up to date');
50            return false;
51        }
52
53        Logger::debug("Sitemapper::generate(): using $sitemap");
54
55        $pages = idx_get_indexer()->getPages();
56        Logger::debug('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
57        $items = [];
58
59        // build the sitemap items
60        foreach($pages as $id){
61            //skip hidden, non existing and restricted files
62            if(isHiddenPage($id)) continue;
63            if(auth_aclcheck($id, '', []) < AUTH_READ) continue;
64            $item = Item::createFromID($id);
65            if ($item instanceof Item)
66                $items[] = $item;
67        }
68
69        $eventData = ['items' => &$items, 'sitemap' => &$sitemap];
70        $event = new Event('SITEMAP_GENERATE', $eventData);
71        if ($event->advise_before(true)) {
72            //save the new sitemap
73            $event->result = io_saveFile($sitemap, (new Mapper())->getXML($items));
74        }
75        $event->advise_after();
76
77        return $event->result;
78    }
79
80    /**
81     * Builds the sitemap XML string from the given array auf SitemapItems.
82     *
83     * @param $items array The SitemapItems that shall be included in the sitemap.
84     * @return string The sitemap XML.
85     *
86     * @author Michael Hamann
87     */
88    private function getXML($items) {
89        ob_start();
90        echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
91        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
92        foreach ($items as $item) {
93            /** @var Item $item */
94            echo $item->toXML();
95        }
96        echo '</urlset>'.NL;
97        $result = ob_get_contents();
98        ob_end_clean();
99        return $result;
100    }
101
102    /**
103     * Helper function for getting the path to the sitemap file.
104     *
105     * @return string The path to the sitemap file.
106     *
107     * @author Michael Hamann
108     */
109    public static function getFilePath() {
110        global $conf;
111
112        $sitemap = $conf['cachedir'].'/sitemap.xml';
113        if (self::sitemapIsCompressed()) {
114            $sitemap .= '.gz';
115        }
116
117        return $sitemap;
118    }
119
120    /**
121     * Helper function for checking if the sitemap is compressed
122     *
123     * @return bool If the sitemap file is compressed
124     */
125    public static function sitemapIsCompressed() {
126        global $conf;
127        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
128    }
129
130    /**
131     * Pings search engines with the sitemap url. Plugins can add or remove
132     * urls to ping using the SITEMAP_PING event.
133     *
134     * @author Michael Hamann
135     *
136     * @return bool
137     */
138    public static function pingSearchEngines() {
139        //ping search engines...
140        $http = new DokuHTTPClient();
141        $http->timeout = 8;
142
143        $encoded_sitemap_url = urlencode(wl('', ['do' => 'sitemap'], true, '&'));
144        $ping_urls = [
145            'google'    => 'https://www.google.com/ping?sitemap='.$encoded_sitemap_url,
146            'yandex'    => 'https://webmaster.yandex.com/ping?sitemap='.$encoded_sitemap_url
147        ];
148
149        $data = [
150            'ping_urls' => $ping_urls,
151            'encoded_sitemap_url' => $encoded_sitemap_url
152        ];
153        $event = new Event('SITEMAP_PING', $data);
154        if ($event->advise_before(true)) {
155            foreach ($data['ping_urls'] as $name => $url) {
156                Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
157                $resp = $http->get($url);
158                if($http->error) {
159                    Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
160                }
161            }
162        }
163        $event->advise_after();
164
165        return true;
166    }
167}
168
169