xref: /dokuwiki/inc/Sitemap/Mapper.php (revision d868eb89f182718a31113373a6272670bd7f8012)
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 */
208c7c53b0SAndreas Gohrclass Mapper
218c7c53b0SAndreas 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     */
35*d868eb89SAndreas Gohr    public static function generate()
36*d868eb89SAndreas Gohr    {
37432adb37SAndreas Gohr        global $conf;
38432adb37SAndreas Gohr        if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
39432adb37SAndreas Gohr
40432adb37SAndreas Gohr        $sitemap = Mapper::getFilePath();
41432adb37SAndreas Gohr
42432adb37SAndreas Gohr        if (file_exists($sitemap)) {
43432adb37SAndreas Gohr            if(!is_writable($sitemap)) return false;
4424870174SAndreas Gohr        } elseif (!is_writable(dirname($sitemap))) {
4524870174SAndreas Gohr            return false;
46432adb37SAndreas Gohr        }
47432adb37SAndreas Gohr
48432adb37SAndreas Gohr        if(@filesize($sitemap) &&
49432adb37SAndreas Gohr           @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
5031667ec6SAndreas Gohr            Logger::debug('Sitemapper::generate(): Sitemap up to date');
51432adb37SAndreas Gohr            return false;
52432adb37SAndreas Gohr        }
53432adb37SAndreas Gohr
5431667ec6SAndreas Gohr        Logger::debug("Sitemapper::generate(): using $sitemap");
55432adb37SAndreas Gohr
56432adb37SAndreas Gohr        $pages = idx_get_indexer()->getPages();
5731667ec6SAndreas Gohr        Logger::debug('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
5824870174SAndreas Gohr        $items = [];
59432adb37SAndreas Gohr
60432adb37SAndreas Gohr        // build the sitemap items
61432adb37SAndreas Gohr        foreach($pages as $id){
62432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
63432adb37SAndreas Gohr            if(isHiddenPage($id)) continue;
6424870174SAndreas Gohr            if(auth_aclcheck($id, '', []) < AUTH_READ) continue;
65432adb37SAndreas Gohr            $item = Item::createFromID($id);
6624870174SAndreas Gohr            if ($item instanceof Item)
67432adb37SAndreas Gohr                $items[] = $item;
68432adb37SAndreas Gohr        }
69432adb37SAndreas Gohr
7024870174SAndreas Gohr        $eventData = ['items' => &$items, 'sitemap' => &$sitemap];
7124870174SAndreas Gohr        $event = new Event('SITEMAP_GENERATE', $eventData);
72432adb37SAndreas Gohr        if ($event->advise_before(true)) {
73432adb37SAndreas Gohr            //save the new sitemap
7424870174SAndreas Gohr            $event->result = io_saveFile($sitemap, (new Mapper())->getXML($items));
75432adb37SAndreas Gohr        }
76432adb37SAndreas Gohr        $event->advise_after();
77432adb37SAndreas Gohr
78432adb37SAndreas Gohr        return $event->result;
79432adb37SAndreas Gohr    }
80432adb37SAndreas Gohr
81432adb37SAndreas Gohr    /**
82432adb37SAndreas Gohr     * Builds the sitemap XML string from the given array auf SitemapItems.
83432adb37SAndreas Gohr     *
84432adb37SAndreas Gohr     * @param $items array The SitemapItems that shall be included in the sitemap.
85432adb37SAndreas Gohr     * @return string The sitemap XML.
86432adb37SAndreas Gohr     *
87432adb37SAndreas Gohr     * @author Michael Hamann
88432adb37SAndreas Gohr     */
89*d868eb89SAndreas Gohr    private function getXML($items)
90*d868eb89SAndreas Gohr    {
91432adb37SAndreas Gohr        ob_start();
92432adb37SAndreas Gohr        echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
93432adb37SAndreas Gohr        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
94432adb37SAndreas Gohr        foreach ($items as $item) {
95432adb37SAndreas Gohr            /** @var Item $item */
96432adb37SAndreas Gohr            echo $item->toXML();
97432adb37SAndreas Gohr        }
98432adb37SAndreas Gohr        echo '</urlset>'.NL;
99432adb37SAndreas Gohr        $result = ob_get_contents();
100432adb37SAndreas Gohr        ob_end_clean();
101432adb37SAndreas Gohr        return $result;
102432adb37SAndreas Gohr    }
103432adb37SAndreas Gohr
104432adb37SAndreas Gohr    /**
105432adb37SAndreas Gohr     * Helper function for getting the path to the sitemap file.
106432adb37SAndreas Gohr     *
107432adb37SAndreas Gohr     * @return string The path to the sitemap file.
108432adb37SAndreas Gohr     *
109432adb37SAndreas Gohr     * @author Michael Hamann
110432adb37SAndreas Gohr     */
111*d868eb89SAndreas Gohr    public static function getFilePath()
112*d868eb89SAndreas Gohr    {
113432adb37SAndreas Gohr        global $conf;
114432adb37SAndreas Gohr
115432adb37SAndreas Gohr        $sitemap = $conf['cachedir'].'/sitemap.xml';
116432adb37SAndreas Gohr        if (self::sitemapIsCompressed()) {
117432adb37SAndreas Gohr            $sitemap .= '.gz';
118432adb37SAndreas Gohr        }
119432adb37SAndreas Gohr
120432adb37SAndreas Gohr        return $sitemap;
121432adb37SAndreas Gohr    }
122432adb37SAndreas Gohr
123432adb37SAndreas Gohr    /**
124432adb37SAndreas Gohr     * Helper function for checking if the sitemap is compressed
125432adb37SAndreas Gohr     *
126432adb37SAndreas Gohr     * @return bool If the sitemap file is compressed
127432adb37SAndreas Gohr     */
128*d868eb89SAndreas Gohr    public static function sitemapIsCompressed()
129*d868eb89SAndreas Gohr    {
130432adb37SAndreas Gohr        global $conf;
131432adb37SAndreas Gohr        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
132432adb37SAndreas Gohr    }
133432adb37SAndreas Gohr
134432adb37SAndreas Gohr    /**
135432adb37SAndreas Gohr     * Pings search engines with the sitemap url. Plugins can add or remove
136432adb37SAndreas Gohr     * urls to ping using the SITEMAP_PING event.
137432adb37SAndreas Gohr     *
138432adb37SAndreas Gohr     * @author Michael Hamann
139432adb37SAndreas Gohr     *
140432adb37SAndreas Gohr     * @return bool
141432adb37SAndreas Gohr     */
142*d868eb89SAndreas Gohr    public static function pingSearchEngines()
143*d868eb89SAndreas Gohr    {
144432adb37SAndreas Gohr        //ping search engines...
145198564abSMichael Große        $http = new DokuHTTPClient();
146432adb37SAndreas Gohr        $http->timeout = 8;
147432adb37SAndreas Gohr
14824870174SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', ['do' => 'sitemap'], true, '&'));
14924870174SAndreas Gohr        $ping_urls = [
1501ef741dbSNicolas Friedli            'google'    => 'https://www.google.com/ping?sitemap='.$encoded_sitemap_url,
1511560cda4SAndreas Gohr            'yandex'    => 'https://webmaster.yandex.com/ping?sitemap='.$encoded_sitemap_url
15224870174SAndreas Gohr        ];
153432adb37SAndreas Gohr
15424870174SAndreas Gohr        $data = [
15524870174SAndreas Gohr            'ping_urls' => $ping_urls,
156432adb37SAndreas Gohr            'encoded_sitemap_url' => $encoded_sitemap_url
15724870174SAndreas Gohr        ];
15824870174SAndreas Gohr        $event = new Event('SITEMAP_PING', $data);
159432adb37SAndreas Gohr        if ($event->advise_before(true)) {
160432adb37SAndreas Gohr            foreach ($data['ping_urls'] as $name => $url) {
16131667ec6SAndreas Gohr                Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
162432adb37SAndreas Gohr                $resp = $http->get($url);
16331667ec6SAndreas Gohr                if($http->error) {
16431667ec6SAndreas Gohr                    Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
16531667ec6SAndreas Gohr                }
166432adb37SAndreas Gohr            }
167432adb37SAndreas Gohr        }
168432adb37SAndreas Gohr        $event->advise_after();
169432adb37SAndreas Gohr
170432adb37SAndreas Gohr        return true;
171432adb37SAndreas Gohr    }
172432adb37SAndreas Gohr}
173432adb37SAndreas Gohr
174