xref: /dokuwiki/inc/Sitemap/Mapper.php (revision d4f83172d9533c4d84f450fe22ef630816b21d75)
1432adb37SAndreas Gohr<?php
2*d4f83172SAndreas Gohr
3432adb37SAndreas Gohr/**
4432adb37SAndreas Gohr * Sitemap handling functions
5432adb37SAndreas Gohr *
6432adb37SAndreas Gohr * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
7432adb37SAndreas Gohr * @author     Michael Hamann <michael@content-space.de>
8432adb37SAndreas Gohr */
9432adb37SAndreas Gohr
10432adb37SAndreas Gohrnamespace dokuwiki\Sitemap;
11432adb37SAndreas Gohr
1224870174SAndreas Gohruse dokuwiki\Extension\Event;
135a8d6e48SMichael Großeuse dokuwiki\HTTP\DokuHTTPClient;
1431667ec6SAndreas Gohruse dokuwiki\Logger;
15198564abSMichael Große
16432adb37SAndreas Gohr/**
17432adb37SAndreas Gohr * A class for building sitemaps and pinging search engines with the sitemap URL.
18432adb37SAndreas Gohr *
19432adb37SAndreas Gohr * @author Michael Hamann
20432adb37SAndreas Gohr */
218c7c53b0SAndreas Gohrclass Mapper
228c7c53b0SAndreas Gohr{
23432adb37SAndreas Gohr    /**
24432adb37SAndreas Gohr     * Builds a Google Sitemap of all public pages known to the indexer
25432adb37SAndreas Gohr     *
26432adb37SAndreas Gohr     * The map is placed in the cache directory named sitemap.xml.gz - This
27432adb37SAndreas Gohr     * file needs to be writable!
28432adb37SAndreas Gohr     *
29432adb37SAndreas Gohr     * @author Michael Hamann
30432adb37SAndreas Gohr     * @author Andreas Gohr
31432adb37SAndreas Gohr     * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
32432adb37SAndreas Gohr     * @link   http://www.sitemaps.org/
33432adb37SAndreas Gohr     *
34432adb37SAndreas Gohr     * @return bool
35432adb37SAndreas Gohr     */
36d868eb89SAndreas Gohr    public static function generate()
37d868eb89SAndreas Gohr    {
38432adb37SAndreas Gohr        global $conf;
39432adb37SAndreas Gohr        if ($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
40432adb37SAndreas Gohr
41432adb37SAndreas Gohr        $sitemap = Mapper::getFilePath();
42432adb37SAndreas Gohr
43432adb37SAndreas Gohr        if (file_exists($sitemap)) {
44432adb37SAndreas Gohr            if (!is_writable($sitemap)) return false;
4524870174SAndreas Gohr        } elseif (!is_writable(dirname($sitemap))) {
4624870174SAndreas Gohr            return false;
47432adb37SAndreas Gohr        }
48432adb37SAndreas Gohr
497d34963bSAndreas Gohr        if (
507d34963bSAndreas Gohr            @filesize($sitemap) &&
517d34963bSAndreas Gohr            @filemtime($sitemap) > (time() - ($conf['sitemap'] * 86400))
527d34963bSAndreas Gohr        ) { // 60*60*24=86400
5331667ec6SAndreas Gohr            Logger::debug('Sitemapper::generate(): Sitemap up to date');
54432adb37SAndreas Gohr            return false;
55432adb37SAndreas Gohr        }
56432adb37SAndreas Gohr
5731667ec6SAndreas Gohr        Logger::debug("Sitemapper::generate(): using $sitemap");
58432adb37SAndreas Gohr
59432adb37SAndreas Gohr        $pages = idx_get_indexer()->getPages();
6031667ec6SAndreas Gohr        Logger::debug('Sitemapper::generate(): creating sitemap using ' . count($pages) . ' pages');
6124870174SAndreas Gohr        $items = [];
62432adb37SAndreas Gohr
63432adb37SAndreas Gohr        // build the sitemap items
64432adb37SAndreas Gohr        foreach ($pages as $id) {
65432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
66432adb37SAndreas Gohr            if (isHiddenPage($id)) continue;
6724870174SAndreas Gohr            if (auth_aclcheck($id, '', []) < AUTH_READ) continue;
68432adb37SAndreas Gohr            $item = Item::createFromID($id);
6924870174SAndreas Gohr            if ($item instanceof Item)
70432adb37SAndreas Gohr                $items[] = $item;
71432adb37SAndreas Gohr        }
72432adb37SAndreas Gohr
7324870174SAndreas Gohr        $eventData = ['items' => &$items, 'sitemap' => &$sitemap];
7424870174SAndreas Gohr        $event = new Event('SITEMAP_GENERATE', $eventData);
75432adb37SAndreas Gohr        if ($event->advise_before(true)) {
76432adb37SAndreas Gohr            //save the new sitemap
7724870174SAndreas Gohr            $event->result = io_saveFile($sitemap, (new Mapper())->getXML($items));
78432adb37SAndreas Gohr        }
79432adb37SAndreas Gohr        $event->advise_after();
80432adb37SAndreas Gohr
81432adb37SAndreas Gohr        return $event->result;
82432adb37SAndreas Gohr    }
83432adb37SAndreas Gohr
84432adb37SAndreas Gohr    /**
85432adb37SAndreas Gohr     * Builds the sitemap XML string from the given array auf SitemapItems.
86432adb37SAndreas Gohr     *
87432adb37SAndreas Gohr     * @param $items array The SitemapItems that shall be included in the sitemap.
88432adb37SAndreas Gohr     * @return string The sitemap XML.
89432adb37SAndreas Gohr     *
90432adb37SAndreas Gohr     * @author Michael Hamann
91432adb37SAndreas Gohr     */
92d868eb89SAndreas Gohr    private function getXML($items)
93d868eb89SAndreas Gohr    {
94432adb37SAndreas Gohr        ob_start();
95432adb37SAndreas Gohr        echo '<?xml version="1.0" encoding="UTF-8"?>' . NL;
96432adb37SAndreas Gohr        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . NL;
97432adb37SAndreas Gohr        foreach ($items as $item) {
98432adb37SAndreas Gohr            /** @var Item $item */
99432adb37SAndreas Gohr            echo $item->toXML();
100432adb37SAndreas Gohr        }
101432adb37SAndreas Gohr        echo '</urlset>' . NL;
102432adb37SAndreas Gohr        $result = ob_get_contents();
103432adb37SAndreas Gohr        ob_end_clean();
104432adb37SAndreas Gohr        return $result;
105432adb37SAndreas Gohr    }
106432adb37SAndreas Gohr
107432adb37SAndreas Gohr    /**
108432adb37SAndreas Gohr     * Helper function for getting the path to the sitemap file.
109432adb37SAndreas Gohr     *
110432adb37SAndreas Gohr     * @return string The path to the sitemap file.
111432adb37SAndreas Gohr     *
112432adb37SAndreas Gohr     * @author Michael Hamann
113432adb37SAndreas Gohr     */
114d868eb89SAndreas Gohr    public static function getFilePath()
115d868eb89SAndreas Gohr    {
116432adb37SAndreas Gohr        global $conf;
117432adb37SAndreas Gohr
118432adb37SAndreas Gohr        $sitemap = $conf['cachedir'] . '/sitemap.xml';
119432adb37SAndreas Gohr        if (self::sitemapIsCompressed()) {
120432adb37SAndreas Gohr            $sitemap .= '.gz';
121432adb37SAndreas Gohr        }
122432adb37SAndreas Gohr
123432adb37SAndreas Gohr        return $sitemap;
124432adb37SAndreas Gohr    }
125432adb37SAndreas Gohr
126432adb37SAndreas Gohr    /**
127432adb37SAndreas Gohr     * Helper function for checking if the sitemap is compressed
128432adb37SAndreas Gohr     *
129432adb37SAndreas Gohr     * @return bool If the sitemap file is compressed
130432adb37SAndreas Gohr     */
131d868eb89SAndreas Gohr    public static function sitemapIsCompressed()
132d868eb89SAndreas Gohr    {
133432adb37SAndreas Gohr        global $conf;
134432adb37SAndreas Gohr        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
135432adb37SAndreas Gohr    }
136432adb37SAndreas Gohr
137432adb37SAndreas Gohr    /**
138432adb37SAndreas Gohr     * Pings search engines with the sitemap url. Plugins can add or remove
139432adb37SAndreas Gohr     * urls to ping using the SITEMAP_PING event.
140432adb37SAndreas Gohr     *
141432adb37SAndreas Gohr     * @author Michael Hamann
142432adb37SAndreas Gohr     *
143432adb37SAndreas Gohr     * @return bool
144432adb37SAndreas Gohr     */
145d868eb89SAndreas Gohr    public static function pingSearchEngines()
146d868eb89SAndreas Gohr    {
147432adb37SAndreas Gohr        //ping search engines...
148198564abSMichael Große        $http = new DokuHTTPClient();
149432adb37SAndreas Gohr        $http->timeout = 8;
150432adb37SAndreas Gohr
15124870174SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', ['do' => 'sitemap'], true, '&'));
15224870174SAndreas Gohr        $ping_urls = [
1531ef741dbSNicolas Friedli            'google'    => 'https://www.google.com/ping?sitemap=' . $encoded_sitemap_url,
1541560cda4SAndreas Gohr            'yandex'    => 'https://webmaster.yandex.com/ping?sitemap=' . $encoded_sitemap_url
15524870174SAndreas Gohr        ];
156432adb37SAndreas Gohr
15724870174SAndreas Gohr        $data = [
15824870174SAndreas Gohr            'ping_urls' => $ping_urls,
159432adb37SAndreas Gohr            'encoded_sitemap_url' => $encoded_sitemap_url
16024870174SAndreas Gohr        ];
16124870174SAndreas Gohr        $event = new Event('SITEMAP_PING', $data);
162432adb37SAndreas Gohr        if ($event->advise_before(true)) {
163432adb37SAndreas Gohr            foreach ($data['ping_urls'] as $name => $url) {
16431667ec6SAndreas Gohr                Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
165432adb37SAndreas Gohr                $resp = $http->get($url);
16631667ec6SAndreas Gohr                if ($http->error) {
16731667ec6SAndreas Gohr                    Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
16831667ec6SAndreas Gohr                }
169432adb37SAndreas Gohr            }
170432adb37SAndreas Gohr        }
171432adb37SAndreas Gohr        $event->advise_after();
172432adb37SAndreas Gohr
173432adb37SAndreas Gohr        return true;
174432adb37SAndreas Gohr    }
175432adb37SAndreas Gohr}
176