xref: /dokuwiki/inc/Sitemap/Mapper.php (revision 24870174d2ee45460ba6bcfe5f5a0ae94715efd7)
1432adb37SAndreas Gohr<?php
2432adb37SAndreas Gohr/**
3432adb37SAndreas Gohr * Sitemap handling functions
4432adb37SAndreas Gohr *
5432adb37SAndreas Gohr * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6432adb37SAndreas Gohr * @author     Michael Hamann <michael@content-space.de>
7432adb37SAndreas Gohr */
8432adb37SAndreas Gohr
9432adb37SAndreas Gohrnamespace dokuwiki\Sitemap;
10432adb37SAndreas Gohr
11*24870174SAndreas Gohruse dokuwiki\Extension\Event;
125a8d6e48SMichael Großeuse dokuwiki\HTTP\DokuHTTPClient;
1331667ec6SAndreas Gohruse dokuwiki\Logger;
14198564abSMichael Große
15432adb37SAndreas Gohr/**
16432adb37SAndreas Gohr * A class for building sitemaps and pinging search engines with the sitemap URL.
17432adb37SAndreas Gohr *
18432adb37SAndreas Gohr * @author Michael Hamann
19432adb37SAndreas Gohr */
20432adb37SAndreas Gohrclass Mapper {
21432adb37SAndreas Gohr    /**
22432adb37SAndreas Gohr     * Builds a Google Sitemap of all public pages known to the indexer
23432adb37SAndreas Gohr     *
24432adb37SAndreas Gohr     * The map is placed in the cache directory named sitemap.xml.gz - This
25432adb37SAndreas Gohr     * file needs to be writable!
26432adb37SAndreas Gohr     *
27432adb37SAndreas Gohr     * @author Michael Hamann
28432adb37SAndreas Gohr     * @author Andreas Gohr
29432adb37SAndreas Gohr     * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
30432adb37SAndreas Gohr     * @link   http://www.sitemaps.org/
31432adb37SAndreas Gohr     *
32432adb37SAndreas Gohr     * @return bool
33432adb37SAndreas Gohr     */
34432adb37SAndreas Gohr    public static function generate(){
35432adb37SAndreas Gohr        global $conf;
36432adb37SAndreas Gohr        if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
37432adb37SAndreas Gohr
38432adb37SAndreas Gohr        $sitemap = Mapper::getFilePath();
39432adb37SAndreas Gohr
40432adb37SAndreas Gohr        if (file_exists($sitemap)) {
41432adb37SAndreas Gohr            if(!is_writable($sitemap)) return false;
42*24870174SAndreas Gohr        } elseif (!is_writable(dirname($sitemap))) {
43*24870174SAndreas Gohr            return false;
44432adb37SAndreas Gohr        }
45432adb37SAndreas Gohr
46432adb37SAndreas Gohr        if(@filesize($sitemap) &&
47432adb37SAndreas Gohr           @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
4831667ec6SAndreas Gohr            Logger::debug('Sitemapper::generate(): Sitemap up to date');
49432adb37SAndreas Gohr            return false;
50432adb37SAndreas Gohr        }
51432adb37SAndreas Gohr
5231667ec6SAndreas Gohr        Logger::debug("Sitemapper::generate(): using $sitemap");
53432adb37SAndreas Gohr
54432adb37SAndreas Gohr        $pages = idx_get_indexer()->getPages();
5531667ec6SAndreas Gohr        Logger::debug('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
56*24870174SAndreas Gohr        $items = [];
57432adb37SAndreas Gohr
58432adb37SAndreas Gohr        // build the sitemap items
59432adb37SAndreas Gohr        foreach($pages as $id){
60432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
61432adb37SAndreas Gohr            if(isHiddenPage($id)) continue;
62*24870174SAndreas Gohr            if(auth_aclcheck($id,'',[]) < AUTH_READ) continue;
63432adb37SAndreas Gohr            $item = Item::createFromID($id);
64*24870174SAndreas Gohr            if ($item instanceof Item)
65432adb37SAndreas Gohr                $items[] = $item;
66432adb37SAndreas Gohr        }
67432adb37SAndreas Gohr
68*24870174SAndreas Gohr        $eventData = ['items' => &$items, 'sitemap' => &$sitemap];
69*24870174SAndreas Gohr        $event = new Event('SITEMAP_GENERATE', $eventData);
70432adb37SAndreas Gohr        if ($event->advise_before(true)) {
71432adb37SAndreas Gohr            //save the new sitemap
72*24870174SAndreas Gohr            $event->result = io_saveFile($sitemap, (new Mapper())->getXML($items));
73432adb37SAndreas Gohr        }
74432adb37SAndreas Gohr        $event->advise_after();
75432adb37SAndreas Gohr
76432adb37SAndreas Gohr        return $event->result;
77432adb37SAndreas Gohr    }
78432adb37SAndreas Gohr
79432adb37SAndreas Gohr    /**
80432adb37SAndreas Gohr     * Builds the sitemap XML string from the given array auf SitemapItems.
81432adb37SAndreas Gohr     *
82432adb37SAndreas Gohr     * @param $items array The SitemapItems that shall be included in the sitemap.
83432adb37SAndreas Gohr     * @return string The sitemap XML.
84432adb37SAndreas Gohr     *
85432adb37SAndreas Gohr     * @author Michael Hamann
86432adb37SAndreas Gohr     */
87*24870174SAndreas Gohr    private function getXML($items) {
88432adb37SAndreas Gohr        ob_start();
89432adb37SAndreas Gohr        echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
90432adb37SAndreas Gohr        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
91432adb37SAndreas Gohr        foreach ($items as $item) {
92432adb37SAndreas Gohr            /** @var Item $item */
93432adb37SAndreas Gohr            echo $item->toXML();
94432adb37SAndreas Gohr        }
95432adb37SAndreas Gohr        echo '</urlset>'.NL;
96432adb37SAndreas Gohr        $result = ob_get_contents();
97432adb37SAndreas Gohr        ob_end_clean();
98432adb37SAndreas Gohr        return $result;
99432adb37SAndreas Gohr    }
100432adb37SAndreas Gohr
101432adb37SAndreas Gohr    /**
102432adb37SAndreas Gohr     * Helper function for getting the path to the sitemap file.
103432adb37SAndreas Gohr     *
104432adb37SAndreas Gohr     * @return string The path to the sitemap file.
105432adb37SAndreas Gohr     *
106432adb37SAndreas Gohr     * @author Michael Hamann
107432adb37SAndreas Gohr     */
108432adb37SAndreas Gohr    public static function getFilePath() {
109432adb37SAndreas Gohr        global $conf;
110432adb37SAndreas Gohr
111432adb37SAndreas Gohr        $sitemap = $conf['cachedir'].'/sitemap.xml';
112432adb37SAndreas Gohr        if (self::sitemapIsCompressed()) {
113432adb37SAndreas Gohr            $sitemap .= '.gz';
114432adb37SAndreas Gohr        }
115432adb37SAndreas Gohr
116432adb37SAndreas Gohr        return $sitemap;
117432adb37SAndreas Gohr    }
118432adb37SAndreas Gohr
119432adb37SAndreas Gohr    /**
120432adb37SAndreas Gohr     * Helper function for checking if the sitemap is compressed
121432adb37SAndreas Gohr     *
122432adb37SAndreas Gohr     * @return bool If the sitemap file is compressed
123432adb37SAndreas Gohr     */
124432adb37SAndreas Gohr    public static function sitemapIsCompressed() {
125432adb37SAndreas Gohr        global $conf;
126432adb37SAndreas Gohr        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
127432adb37SAndreas Gohr    }
128432adb37SAndreas Gohr
129432adb37SAndreas Gohr    /**
130432adb37SAndreas Gohr     * Pings search engines with the sitemap url. Plugins can add or remove
131432adb37SAndreas Gohr     * urls to ping using the SITEMAP_PING event.
132432adb37SAndreas Gohr     *
133432adb37SAndreas Gohr     * @author Michael Hamann
134432adb37SAndreas Gohr     *
135432adb37SAndreas Gohr     * @return bool
136432adb37SAndreas Gohr     */
137432adb37SAndreas Gohr    public static function pingSearchEngines() {
138432adb37SAndreas Gohr        //ping search engines...
139198564abSMichael Große        $http = new DokuHTTPClient();
140432adb37SAndreas Gohr        $http->timeout = 8;
141432adb37SAndreas Gohr
142*24870174SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', ['do' => 'sitemap'], true, '&'));
143*24870174SAndreas Gohr        $ping_urls = [
1441ef741dbSNicolas Friedli            'google'    => 'https://www.google.com/ping?sitemap='.$encoded_sitemap_url,
1451560cda4SAndreas Gohr            'yandex'    => 'https://webmaster.yandex.com/ping?sitemap='.$encoded_sitemap_url
146*24870174SAndreas Gohr        ];
147432adb37SAndreas Gohr
148*24870174SAndreas Gohr        $data = [
149*24870174SAndreas Gohr            'ping_urls' => $ping_urls,
150432adb37SAndreas Gohr            'encoded_sitemap_url' => $encoded_sitemap_url
151*24870174SAndreas Gohr        ];
152*24870174SAndreas Gohr        $event = new Event('SITEMAP_PING', $data);
153432adb37SAndreas Gohr        if ($event->advise_before(true)) {
154432adb37SAndreas Gohr            foreach ($data['ping_urls'] as $name => $url) {
15531667ec6SAndreas Gohr                Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
156432adb37SAndreas Gohr                $resp = $http->get($url);
15731667ec6SAndreas Gohr                if($http->error) {
15831667ec6SAndreas Gohr                    Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
15931667ec6SAndreas Gohr                }
160432adb37SAndreas Gohr            }
161432adb37SAndreas Gohr        }
162432adb37SAndreas Gohr        $event->advise_after();
163432adb37SAndreas Gohr
164432adb37SAndreas Gohr        return true;
165432adb37SAndreas Gohr    }
166432adb37SAndreas Gohr}
167432adb37SAndreas Gohr
168