xref: /dokuwiki/inc/Sitemap/Mapper.php (revision c1803f3da174a1e102253af2bf9c53f5f26de2e9)
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
11be5c1ea2SSatoshi Saharause dokuwiki\Extension\Event;
125a8d6e48SMichael Großeuse dokuwiki\HTTP\DokuHTTPClient;
1331667ec6SAndreas Gohruse dokuwiki\Logger;
144027a91aSSatoshi Saharause dokuwiki\Search\Indexer;
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 */
21*c1803f3dSSatoshi Saharaclass Mapper
22*c1803f3dSSatoshi Sahara{
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     */
36be5c1ea2SSatoshi Sahara    public static function generate()
37be5c1ea2SSatoshi Sahara    {
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($sitemap)) return false;
45432adb37SAndreas Gohr        } else {
46432adb37SAndreas Gohr            if (!is_writable(dirname($sitemap))) return false;
47432adb37SAndreas Gohr        }
48432adb37SAndreas Gohr
49432adb37SAndreas Gohr        if (@filesize($sitemap) &&
50be5c1ea2SSatoshi Sahara            @filemtime($sitemap) > (time()-($conf['sitemap']*86400)) // 60*60*24=86400
51be5c1ea2SSatoshi Sahara        ) {
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');
60432adb37SAndreas Gohr        $items = array();
61432adb37SAndreas Gohr
62432adb37SAndreas Gohr        // build the sitemap items
63432adb37SAndreas Gohr        foreach ($pages as $id) {
64432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
65432adb37SAndreas Gohr            if (isHiddenPage($id)) continue;
66432adb37SAndreas Gohr            if (auth_aclcheck($id,'',array()) < AUTH_READ) continue;
67432adb37SAndreas Gohr            $item = Item::createFromID($id);
68432adb37SAndreas Gohr            if ($item !== null)
69432adb37SAndreas Gohr                $items[] = $item;
70432adb37SAndreas Gohr        }
71432adb37SAndreas Gohr
72432adb37SAndreas Gohr        $eventData = array('items' => &$items, 'sitemap' => &$sitemap);
73be5c1ea2SSatoshi Sahara        $event = new Event('SITEMAP_GENERATE', $eventData);
74432adb37SAndreas Gohr        if ($event->advise_before(true)) {
75432adb37SAndreas Gohr            //save the new sitemap
76432adb37SAndreas Gohr            $event->result = io_saveFile($sitemap, 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 $items array The SitemapItems that shall be included in the sitemap.
87432adb37SAndreas Gohr     * @return string The sitemap XML.
88432adb37SAndreas Gohr     *
89432adb37SAndreas Gohr     * @author Michael Hamann
90432adb37SAndreas Gohr     */
91be5c1ea2SSatoshi Sahara    private static function getXML($items)
92be5c1ea2SSatoshi Sahara    {
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     */
113be5c1ea2SSatoshi Sahara    public static function getFilePath()
114be5c1ea2SSatoshi Sahara    {
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     */
130be5c1ea2SSatoshi Sahara    public static function sitemapIsCompressed()
131be5c1ea2SSatoshi Sahara    {
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     */
144be5c1ea2SSatoshi Sahara    public static function pingSearchEngines()
145be5c1ea2SSatoshi Sahara    {
146432adb37SAndreas Gohr        //ping search engines...
147198564abSMichael Große        $http = new DokuHTTPClient();
148432adb37SAndreas Gohr        $http->timeout = 8;
149432adb37SAndreas Gohr
150432adb37SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', array('do' => 'sitemap'), true, '&'));
151432adb37SAndreas Gohr        $ping_urls = array(
152432adb37SAndreas Gohr            'google'    => 'http://www.google.com/webmasters/sitemaps/ping?sitemap='.$encoded_sitemap_url,
153432adb37SAndreas Gohr            'microsoft' => 'http://www.bing.com/webmaster/ping.aspx?siteMap='.$encoded_sitemap_url,
154432adb37SAndreas Gohr            'yandex'    => 'http://blogs.yandex.ru/pings/?status=success&url='.$encoded_sitemap_url
155432adb37SAndreas Gohr        );
156432adb37SAndreas Gohr
157be5c1ea2SSatoshi Sahara        $data = array(
158be5c1ea2SSatoshi Sahara            'ping_urls' => $ping_urls,
159432adb37SAndreas Gohr            'encoded_sitemap_url' => $encoded_sitemap_url
160432adb37SAndreas Gohr        );
161be5c1ea2SSatoshi Sahara        $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