xref: /dokuwiki/inc/Sitemap/Mapper.php (revision 31667ec67823633de149106f2f0de3f1dbc5fd1e)
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
115a8d6e48SMichael Großeuse dokuwiki\HTTP\DokuHTTPClient;
12*31667ec6SAndreas Gohruse dokuwiki\Logger;
13198564abSMichael Große
14432adb37SAndreas Gohr/**
15432adb37SAndreas Gohr * A class for building sitemaps and pinging search engines with the sitemap URL.
16432adb37SAndreas Gohr *
17432adb37SAndreas Gohr * @author Michael Hamann
18432adb37SAndreas Gohr */
19432adb37SAndreas Gohrclass Mapper {
20432adb37SAndreas Gohr    /**
21432adb37SAndreas Gohr     * Builds a Google Sitemap of all public pages known to the indexer
22432adb37SAndreas Gohr     *
23432adb37SAndreas Gohr     * The map is placed in the cache directory named sitemap.xml.gz - This
24432adb37SAndreas Gohr     * file needs to be writable!
25432adb37SAndreas Gohr     *
26432adb37SAndreas Gohr     * @author Michael Hamann
27432adb37SAndreas Gohr     * @author Andreas Gohr
28432adb37SAndreas Gohr     * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
29432adb37SAndreas Gohr     * @link   http://www.sitemaps.org/
30432adb37SAndreas Gohr     *
31432adb37SAndreas Gohr     * @return bool
32432adb37SAndreas Gohr     */
33432adb37SAndreas Gohr    public static function generate(){
34432adb37SAndreas Gohr        global $conf;
35432adb37SAndreas Gohr        if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
36432adb37SAndreas Gohr
37432adb37SAndreas Gohr        $sitemap = Mapper::getFilePath();
38432adb37SAndreas Gohr
39432adb37SAndreas Gohr        if(file_exists($sitemap)){
40432adb37SAndreas Gohr            if(!is_writable($sitemap)) return false;
41432adb37SAndreas Gohr        }else{
42432adb37SAndreas Gohr            if(!is_writable(dirname($sitemap))) return false;
43432adb37SAndreas Gohr        }
44432adb37SAndreas Gohr
45432adb37SAndreas Gohr        if(@filesize($sitemap) &&
46432adb37SAndreas Gohr           @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
47*31667ec6SAndreas Gohr            Logger::debug('Sitemapper::generate(): Sitemap up to date');
48432adb37SAndreas Gohr            return false;
49432adb37SAndreas Gohr        }
50432adb37SAndreas Gohr
51*31667ec6SAndreas Gohr        Logger::debug("Sitemapper::generate(): using $sitemap");
52432adb37SAndreas Gohr
53432adb37SAndreas Gohr        $pages = idx_get_indexer()->getPages();
54*31667ec6SAndreas Gohr        Logger::debug('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
55432adb37SAndreas Gohr        $items = array();
56432adb37SAndreas Gohr
57432adb37SAndreas Gohr        // build the sitemap items
58432adb37SAndreas Gohr        foreach($pages as $id){
59432adb37SAndreas Gohr            //skip hidden, non existing and restricted files
60432adb37SAndreas Gohr            if(isHiddenPage($id)) continue;
61432adb37SAndreas Gohr            if(auth_aclcheck($id,'',array()) < AUTH_READ) continue;
62432adb37SAndreas Gohr            $item = Item::createFromID($id);
63432adb37SAndreas Gohr            if ($item !== null)
64432adb37SAndreas Gohr                $items[] = $item;
65432adb37SAndreas Gohr        }
66432adb37SAndreas Gohr
67432adb37SAndreas Gohr        $eventData = array('items' => &$items, 'sitemap' => &$sitemap);
68e1d9dcc8SAndreas Gohr        $event = new \dokuwiki\Extension\Event('SITEMAP_GENERATE', $eventData);
69432adb37SAndreas Gohr        if ($event->advise_before(true)) {
70432adb37SAndreas Gohr            //save the new sitemap
71432adb37SAndreas Gohr            $event->result = io_saveFile($sitemap, Mapper::getXML($items));
72432adb37SAndreas Gohr        }
73432adb37SAndreas Gohr        $event->advise_after();
74432adb37SAndreas Gohr
75432adb37SAndreas Gohr        return $event->result;
76432adb37SAndreas Gohr    }
77432adb37SAndreas Gohr
78432adb37SAndreas Gohr    /**
79432adb37SAndreas Gohr     * Builds the sitemap XML string from the given array auf SitemapItems.
80432adb37SAndreas Gohr     *
81432adb37SAndreas Gohr     * @param $items array The SitemapItems that shall be included in the sitemap.
82432adb37SAndreas Gohr     * @return string The sitemap XML.
83432adb37SAndreas Gohr     *
84432adb37SAndreas Gohr     * @author Michael Hamann
85432adb37SAndreas Gohr     */
86432adb37SAndreas Gohr    private static function getXML($items) {
87432adb37SAndreas Gohr        ob_start();
88432adb37SAndreas Gohr        echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
89432adb37SAndreas Gohr        echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
90432adb37SAndreas Gohr        foreach ($items as $item) {
91432adb37SAndreas Gohr            /** @var Item $item */
92432adb37SAndreas Gohr            echo $item->toXML();
93432adb37SAndreas Gohr        }
94432adb37SAndreas Gohr        echo '</urlset>'.NL;
95432adb37SAndreas Gohr        $result = ob_get_contents();
96432adb37SAndreas Gohr        ob_end_clean();
97432adb37SAndreas Gohr        return $result;
98432adb37SAndreas Gohr    }
99432adb37SAndreas Gohr
100432adb37SAndreas Gohr    /**
101432adb37SAndreas Gohr     * Helper function for getting the path to the sitemap file.
102432adb37SAndreas Gohr     *
103432adb37SAndreas Gohr     * @return string The path to the sitemap file.
104432adb37SAndreas Gohr     *
105432adb37SAndreas Gohr     * @author Michael Hamann
106432adb37SAndreas Gohr     */
107432adb37SAndreas Gohr    public static function getFilePath() {
108432adb37SAndreas Gohr        global $conf;
109432adb37SAndreas Gohr
110432adb37SAndreas Gohr        $sitemap = $conf['cachedir'].'/sitemap.xml';
111432adb37SAndreas Gohr        if (self::sitemapIsCompressed()) {
112432adb37SAndreas Gohr            $sitemap .= '.gz';
113432adb37SAndreas Gohr        }
114432adb37SAndreas Gohr
115432adb37SAndreas Gohr        return $sitemap;
116432adb37SAndreas Gohr    }
117432adb37SAndreas Gohr
118432adb37SAndreas Gohr    /**
119432adb37SAndreas Gohr     * Helper function for checking if the sitemap is compressed
120432adb37SAndreas Gohr     *
121432adb37SAndreas Gohr     * @return bool If the sitemap file is compressed
122432adb37SAndreas Gohr     */
123432adb37SAndreas Gohr    public static function sitemapIsCompressed() {
124432adb37SAndreas Gohr        global $conf;
125432adb37SAndreas Gohr        return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
126432adb37SAndreas Gohr    }
127432adb37SAndreas Gohr
128432adb37SAndreas Gohr    /**
129432adb37SAndreas Gohr     * Pings search engines with the sitemap url. Plugins can add or remove
130432adb37SAndreas Gohr     * urls to ping using the SITEMAP_PING event.
131432adb37SAndreas Gohr     *
132432adb37SAndreas Gohr     * @author Michael Hamann
133432adb37SAndreas Gohr     *
134432adb37SAndreas Gohr     * @return bool
135432adb37SAndreas Gohr     */
136432adb37SAndreas Gohr    public static function pingSearchEngines() {
137432adb37SAndreas Gohr        //ping search engines...
138198564abSMichael Große        $http = new DokuHTTPClient();
139432adb37SAndreas Gohr        $http->timeout = 8;
140432adb37SAndreas Gohr
141432adb37SAndreas Gohr        $encoded_sitemap_url = urlencode(wl('', array('do' => 'sitemap'), true, '&'));
142432adb37SAndreas Gohr        $ping_urls = array(
143432adb37SAndreas Gohr            'google'    => 'http://www.google.com/webmasters/sitemaps/ping?sitemap='.$encoded_sitemap_url,
144432adb37SAndreas Gohr            'microsoft' => 'http://www.bing.com/webmaster/ping.aspx?siteMap='.$encoded_sitemap_url,
145432adb37SAndreas Gohr            'yandex'    => 'http://blogs.yandex.ru/pings/?status=success&url='.$encoded_sitemap_url
146432adb37SAndreas Gohr        );
147432adb37SAndreas Gohr
148432adb37SAndreas Gohr        $data = array('ping_urls' => $ping_urls,
149432adb37SAndreas Gohr                            'encoded_sitemap_url' => $encoded_sitemap_url
150432adb37SAndreas Gohr        );
151e1d9dcc8SAndreas Gohr        $event = new \dokuwiki\Extension\Event('SITEMAP_PING', $data);
152432adb37SAndreas Gohr        if ($event->advise_before(true)) {
153432adb37SAndreas Gohr            foreach ($data['ping_urls'] as $name => $url) {
154*31667ec6SAndreas Gohr                Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
155432adb37SAndreas Gohr                $resp = $http->get($url);
156*31667ec6SAndreas Gohr                if($http->error) {
157*31667ec6SAndreas Gohr                    Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
158*31667ec6SAndreas Gohr                }
159432adb37SAndreas Gohr            }
160432adb37SAndreas Gohr        }
161432adb37SAndreas Gohr        $event->advise_after();
162432adb37SAndreas Gohr
163432adb37SAndreas Gohr        return true;
164432adb37SAndreas Gohr    }
165432adb37SAndreas Gohr}
166432adb37SAndreas Gohr
167