xref: /dokuwiki/inc/Sitemap/Mapper.php (revision 8c7c53b0321a3cd3116b8d3b2ad27863a38dece7)
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
1124870174SAndreas 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 */
20*8c7c53b0SAndreas Gohrclass Mapper
21*8c7c53b0SAndreas Gohr{
22432adb37SAndreas Gohr    /**
23432adb37SAndreas Gohr     * Builds a Google Sitemap of all public pages known to the indexer
24432adb37SAndreas Gohr     *
25432adb37SAndreas Gohr     * The map is placed in the cache directory named sitemap.xml.gz - This
26432adb37SAndreas Gohr     * file needs to be writable!
27432adb37SAndreas Gohr     *
28432adb37SAndreas Gohr     * @author Michael Hamann
29432adb37SAndreas Gohr     * @author Andreas Gohr
30432adb37SAndreas Gohr     * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
31432adb37SAndreas Gohr     * @link   http://www.sitemaps.org/
32432adb37SAndreas Gohr     *
33432adb37SAndreas Gohr     * @return bool
34432adb37SAndreas Gohr     */
35432adb37SAndreas Gohr    public static function generate(){
36432adb37SAndreas Gohr        global $conf;
37432adb37SAndreas Gohr        if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
38432adb37SAndreas Gohr
39432adb37SAndreas Gohr        $sitemap = Mapper::getFilePath();
40432adb37SAndreas Gohr
41432adb37SAndreas Gohr        if (file_exists($sitemap)) {
42432adb37SAndreas Gohr            if(!is_writable($sitemap)) return false;
4324870174SAndreas Gohr        } elseif (!is_writable(dirname($sitemap))) {
4424870174SAndreas Gohr            return false;
45432adb37SAndreas Gohr        }
46432adb37SAndreas Gohr
47432adb37SAndreas Gohr        if(@filesize($sitemap) &&
48432adb37SAndreas Gohr           @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
4931667ec6SAndreas Gohr            Logger::debug('Sitemapper::generate(): Sitemap up to date');
50432adb37SAndreas Gohr            return false;
51432adb37SAndreas Gohr        }
52432adb37SAndreas Gohr
5331667ec6SAndreas Gohr        Logger::debug("Sitemapper::generate(): using $sitemap");
54432adb37SAndreas Gohr
55432adb37SAndreas Gohr        $pages = idx_get_indexer()->getPages();
5631667ec6SAndreas Gohr        Logger::debug('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
5724870174SAndreas Gohr        $items = [];
58432adb37SAndreas Gohr
59432adb37SAndreas Gohr        // build the sitemap items
60432adb37SAndreas Gohr        foreach($pages as $id){
61432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
62432adb37SAndreas Gohr            if(isHiddenPage($id)) continue;
6324870174SAndreas Gohr            if(auth_aclcheck($id, '', []) < AUTH_READ) continue;
64432adb37SAndreas Gohr            $item = Item::createFromID($id);
6524870174SAndreas Gohr            if ($item instanceof Item)
66432adb37SAndreas Gohr                $items[] = $item;
67432adb37SAndreas Gohr        }
68432adb37SAndreas Gohr
6924870174SAndreas Gohr        $eventData = ['items' => &$items, 'sitemap' => &$sitemap];
7024870174SAndreas Gohr        $event = new Event('SITEMAP_GENERATE', $eventData);
71432adb37SAndreas Gohr        if ($event->advise_before(true)) {
72432adb37SAndreas Gohr            //save the new sitemap
7324870174SAndreas Gohr            $event->result = io_saveFile($sitemap, (new Mapper())->getXML($items));
74432adb37SAndreas Gohr        }
75432adb37SAndreas Gohr        $event->advise_after();
76432adb37SAndreas Gohr
77432adb37SAndreas Gohr        return $event->result;
78432adb37SAndreas Gohr    }
79432adb37SAndreas Gohr
80432adb37SAndreas Gohr    /**
81432adb37SAndreas Gohr     * Builds the sitemap XML string from the given array auf SitemapItems.
82432adb37SAndreas Gohr     *
83432adb37SAndreas Gohr     * @param $items array The SitemapItems that shall be included in the sitemap.
84432adb37SAndreas Gohr     * @return string The sitemap XML.
85432adb37SAndreas Gohr     *
86432adb37SAndreas Gohr     * @author Michael Hamann
87432adb37SAndreas Gohr     */
8824870174SAndreas Gohr    private function getXML($items) {
89432adb37SAndreas Gohr        ob_start();
90432adb37SAndreas Gohr        echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
91432adb37SAndreas Gohr        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
92432adb37SAndreas Gohr        foreach ($items as $item) {
93432adb37SAndreas Gohr            /** @var Item $item */
94432adb37SAndreas Gohr            echo $item->toXML();
95432adb37SAndreas Gohr        }
96432adb37SAndreas Gohr        echo '</urlset>'.NL;
97432adb37SAndreas Gohr        $result = ob_get_contents();
98432adb37SAndreas Gohr        ob_end_clean();
99432adb37SAndreas Gohr        return $result;
100432adb37SAndreas Gohr    }
101432adb37SAndreas Gohr
102432adb37SAndreas Gohr    /**
103432adb37SAndreas Gohr     * Helper function for getting the path to the sitemap file.
104432adb37SAndreas Gohr     *
105432adb37SAndreas Gohr     * @return string The path to the sitemap file.
106432adb37SAndreas Gohr     *
107432adb37SAndreas Gohr     * @author Michael Hamann
108432adb37SAndreas Gohr     */
109432adb37SAndreas Gohr    public static function getFilePath() {
110432adb37SAndreas Gohr        global $conf;
111432adb37SAndreas Gohr
112432adb37SAndreas Gohr        $sitemap = $conf['cachedir'].'/sitemap.xml';
113432adb37SAndreas Gohr        if (self::sitemapIsCompressed()) {
114432adb37SAndreas Gohr            $sitemap .= '.gz';
115432adb37SAndreas Gohr        }
116432adb37SAndreas Gohr
117432adb37SAndreas Gohr        return $sitemap;
118432adb37SAndreas Gohr    }
119432adb37SAndreas Gohr
120432adb37SAndreas Gohr    /**
121432adb37SAndreas Gohr     * Helper function for checking if the sitemap is compressed
122432adb37SAndreas Gohr     *
123432adb37SAndreas Gohr     * @return bool If the sitemap file is compressed
124432adb37SAndreas Gohr     */
125432adb37SAndreas Gohr    public static function sitemapIsCompressed() {
126432adb37SAndreas Gohr        global $conf;
127432adb37SAndreas Gohr        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
128432adb37SAndreas Gohr    }
129432adb37SAndreas Gohr
130432adb37SAndreas Gohr    /**
131432adb37SAndreas Gohr     * Pings search engines with the sitemap url. Plugins can add or remove
132432adb37SAndreas Gohr     * urls to ping using the SITEMAP_PING event.
133432adb37SAndreas Gohr     *
134432adb37SAndreas Gohr     * @author Michael Hamann
135432adb37SAndreas Gohr     *
136432adb37SAndreas Gohr     * @return bool
137432adb37SAndreas Gohr     */
138432adb37SAndreas Gohr    public static function pingSearchEngines() {
139432adb37SAndreas Gohr        //ping search engines...
140198564abSMichael Große        $http = new DokuHTTPClient();
141432adb37SAndreas Gohr        $http->timeout = 8;
142432adb37SAndreas Gohr
14324870174SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', ['do' => 'sitemap'], true, '&'));
14424870174SAndreas Gohr        $ping_urls = [
1451ef741dbSNicolas Friedli            'google'    => 'https://www.google.com/ping?sitemap='.$encoded_sitemap_url,
1461560cda4SAndreas Gohr            'yandex'    => 'https://webmaster.yandex.com/ping?sitemap='.$encoded_sitemap_url
14724870174SAndreas Gohr        ];
148432adb37SAndreas Gohr
14924870174SAndreas Gohr        $data = [
15024870174SAndreas Gohr            'ping_urls' => $ping_urls,
151432adb37SAndreas Gohr            'encoded_sitemap_url' => $encoded_sitemap_url
15224870174SAndreas Gohr        ];
15324870174SAndreas Gohr        $event = new Event('SITEMAP_PING', $data);
154432adb37SAndreas Gohr        if ($event->advise_before(true)) {
155432adb37SAndreas Gohr            foreach ($data['ping_urls'] as $name => $url) {
15631667ec6SAndreas Gohr                Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
157432adb37SAndreas Gohr                $resp = $http->get($url);
15831667ec6SAndreas Gohr                if($http->error) {
15931667ec6SAndreas Gohr                    Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
16031667ec6SAndreas Gohr                }
161432adb37SAndreas Gohr            }
162432adb37SAndreas Gohr        }
163432adb37SAndreas Gohr        $event->advise_after();
164432adb37SAndreas Gohr
165432adb37SAndreas Gohr        return true;
166432adb37SAndreas Gohr    }
167432adb37SAndreas Gohr}
168432adb37SAndreas Gohr
169