xref: /dokuwiki/inc/Sitemap/Mapper.php (revision 7d34963b3e75ea04c63ec066a6b7a692e123cb53)
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     */
35d868eb89SAndreas Gohr    public static function generate()
36d868eb89SAndreas 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
48*7d34963bSAndreas Gohr        if (
49*7d34963bSAndreas Gohr            @filesize($sitemap) &&
50*7d34963bSAndreas Gohr            @filemtime($sitemap) > (time()-($conf['sitemap']*86400))
51*7d34963bSAndreas Gohr        ) { // 60*60*24=86400
5231667ec6SAndreas Gohr            Logger::debug('Sitemapper::generate(): Sitemap up to date');
53432adb37SAndreas Gohr            return false;
54432adb37SAndreas Gohr        }
55432adb37SAndreas Gohr
5631667ec6SAndreas Gohr        Logger::debug("Sitemapper::generate(): using $sitemap");
57432adb37SAndreas Gohr
58432adb37SAndreas Gohr        $pages = idx_get_indexer()->getPages();
5931667ec6SAndreas Gohr        Logger::debug('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
6024870174SAndreas Gohr        $items = [];
61432adb37SAndreas Gohr
62432adb37SAndreas Gohr        // build the sitemap items
63432adb37SAndreas Gohr        foreach ($pages as $id) {
64432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
65432adb37SAndreas Gohr            if (isHiddenPage($id)) continue;
6624870174SAndreas Gohr            if (auth_aclcheck($id, '', []) < AUTH_READ) continue;
67432adb37SAndreas Gohr            $item = Item::createFromID($id);
6824870174SAndreas Gohr            if ($item instanceof Item)
69432adb37SAndreas Gohr                $items[] = $item;
70432adb37SAndreas Gohr        }
71432adb37SAndreas Gohr
7224870174SAndreas Gohr        $eventData = ['items' => &$items, 'sitemap' => &$sitemap];
7324870174SAndreas Gohr        $event = new Event('SITEMAP_GENERATE', $eventData);
74432adb37SAndreas Gohr        if ($event->advise_before(true)) {
75432adb37SAndreas Gohr            //save the new sitemap
7624870174SAndreas Gohr            $event->result = io_saveFile($sitemap, (new Mapper())->getXML($items));
77432adb37SAndreas Gohr        }
78432adb37SAndreas Gohr        $event->advise_after();
79432adb37SAndreas Gohr
80432adb37SAndreas Gohr        return $event->result;
81432adb37SAndreas Gohr    }
82432adb37SAndreas Gohr
83432adb37SAndreas Gohr    /**
84432adb37SAndreas Gohr     * Builds the sitemap XML string from the given array auf SitemapItems.
85432adb37SAndreas Gohr     *
86432adb37SAndreas Gohr     * @param $items array The SitemapItems that shall be included in the sitemap.
87432adb37SAndreas Gohr     * @return string The sitemap XML.
88432adb37SAndreas Gohr     *
89432adb37SAndreas Gohr     * @author Michael Hamann
90432adb37SAndreas Gohr     */
91d868eb89SAndreas Gohr    private function getXML($items)
92d868eb89SAndreas Gohr    {
93432adb37SAndreas Gohr        ob_start();
94432adb37SAndreas Gohr        echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
95432adb37SAndreas Gohr        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
96432adb37SAndreas Gohr        foreach ($items as $item) {
97432adb37SAndreas Gohr            /** @var Item $item */
98432adb37SAndreas Gohr            echo $item->toXML();
99432adb37SAndreas Gohr        }
100432adb37SAndreas Gohr        echo '</urlset>'.NL;
101432adb37SAndreas Gohr        $result = ob_get_contents();
102432adb37SAndreas Gohr        ob_end_clean();
103432adb37SAndreas Gohr        return $result;
104432adb37SAndreas Gohr    }
105432adb37SAndreas Gohr
106432adb37SAndreas Gohr    /**
107432adb37SAndreas Gohr     * Helper function for getting the path to the sitemap file.
108432adb37SAndreas Gohr     *
109432adb37SAndreas Gohr     * @return string The path to the sitemap file.
110432adb37SAndreas Gohr     *
111432adb37SAndreas Gohr     * @author Michael Hamann
112432adb37SAndreas Gohr     */
113d868eb89SAndreas Gohr    public static function getFilePath()
114d868eb89SAndreas Gohr    {
115432adb37SAndreas Gohr        global $conf;
116432adb37SAndreas Gohr
117432adb37SAndreas Gohr        $sitemap = $conf['cachedir'].'/sitemap.xml';
118432adb37SAndreas Gohr        if (self::sitemapIsCompressed()) {
119432adb37SAndreas Gohr            $sitemap .= '.gz';
120432adb37SAndreas Gohr        }
121432adb37SAndreas Gohr
122432adb37SAndreas Gohr        return $sitemap;
123432adb37SAndreas Gohr    }
124432adb37SAndreas Gohr
125432adb37SAndreas Gohr    /**
126432adb37SAndreas Gohr     * Helper function for checking if the sitemap is compressed
127432adb37SAndreas Gohr     *
128432adb37SAndreas Gohr     * @return bool If the sitemap file is compressed
129432adb37SAndreas Gohr     */
130d868eb89SAndreas Gohr    public static function sitemapIsCompressed()
131d868eb89SAndreas Gohr    {
132432adb37SAndreas Gohr        global $conf;
133432adb37SAndreas Gohr        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
134432adb37SAndreas Gohr    }
135432adb37SAndreas Gohr
136432adb37SAndreas Gohr    /**
137432adb37SAndreas Gohr     * Pings search engines with the sitemap url. Plugins can add or remove
138432adb37SAndreas Gohr     * urls to ping using the SITEMAP_PING event.
139432adb37SAndreas Gohr     *
140432adb37SAndreas Gohr     * @author Michael Hamann
141432adb37SAndreas Gohr     *
142432adb37SAndreas Gohr     * @return bool
143432adb37SAndreas Gohr     */
144d868eb89SAndreas Gohr    public static function pingSearchEngines()
145d868eb89SAndreas Gohr    {
146432adb37SAndreas Gohr        //ping search engines...
147198564abSMichael Große        $http = new DokuHTTPClient();
148432adb37SAndreas Gohr        $http->timeout = 8;
149432adb37SAndreas Gohr
15024870174SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', ['do' => 'sitemap'], true, '&'));
15124870174SAndreas Gohr        $ping_urls = [
1521ef741dbSNicolas Friedli            'google'    => 'https://www.google.com/ping?sitemap='.$encoded_sitemap_url,
1531560cda4SAndreas Gohr            'yandex'    => 'https://webmaster.yandex.com/ping?sitemap='.$encoded_sitemap_url
15424870174SAndreas Gohr        ];
155432adb37SAndreas Gohr
15624870174SAndreas Gohr        $data = [
15724870174SAndreas Gohr            'ping_urls' => $ping_urls,
158432adb37SAndreas Gohr            'encoded_sitemap_url' => $encoded_sitemap_url
15924870174SAndreas Gohr        ];
16024870174SAndreas Gohr        $event = new Event('SITEMAP_PING', $data);
161432adb37SAndreas Gohr        if ($event->advise_before(true)) {
162432adb37SAndreas Gohr            foreach ($data['ping_urls'] as $name => $url) {
16331667ec6SAndreas Gohr                Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
164432adb37SAndreas Gohr                $resp = $http->get($url);
16531667ec6SAndreas Gohr                if ($http->error) {
16631667ec6SAndreas Gohr                    Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
16731667ec6SAndreas Gohr                }
168432adb37SAndreas Gohr            }
169432adb37SAndreas Gohr        }
170432adb37SAndreas Gohr        $event->advise_after();
171432adb37SAndreas Gohr
172432adb37SAndreas Gohr        return true;
173432adb37SAndreas Gohr    }
174432adb37SAndreas Gohr}
175