1<?php 2/** 3 * Sitemap handling functions 4 * 5 * @license GPL 2 (http://www.gnu.org/licenses/gpl.html) 6 * @author Michael Hamann <michael@content-space.de> 7 */ 8 9namespace dokuwiki\Sitemap; 10 11use dokuwiki\Extension\Event; 12use dokuwiki\HTTP\DokuHTTPClient; 13use dokuwiki\Search\Indexer; 14 15/** 16 * A class for building sitemaps and pinging search engines with the sitemap URL. 17 * 18 * @author Michael Hamann 19 */ 20class Mapper { 21 /** 22 * Builds a Google Sitemap of all public pages known to the indexer 23 * 24 * The map is placed in the cache directory named sitemap.xml.gz - This 25 * file needs to be writable! 26 * 27 * @author Michael Hamann 28 * @author Andreas Gohr 29 * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html 30 * @link http://www.sitemaps.org/ 31 * 32 * @return bool 33 */ 34 public static function generate() 35 { 36 global $conf; 37 if ($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false; 38 39 $sitemap = Mapper::getFilePath(); 40 41 if (file_exists($sitemap)) { 42 if (!is_writable($sitemap)) return false; 43 } else { 44 if (!is_writable(dirname($sitemap))) return false; 45 } 46 47 if (@filesize($sitemap) && 48 @filemtime($sitemap) > (time()-($conf['sitemap']*86400)) // 60*60*24=86400 49 ) { 50 dbglog('Sitemapper::generate(): Sitemap up to date'); 51 return false; 52 } 53 54 dbglog("Sitemapper::generate(): using $sitemap"); 55 56 $pages = (new Indexer())->getPages(); 57 dbglog('Sitemapper::generate(): creating sitemap using '.count($pages).' pages'); 58 $items = array(); 59 60 // build the sitemap items 61 foreach ($pages as $id) { 62 //skip hidden, non existing and restricted files 63 if (isHiddenPage($id)) continue; 64 if (auth_aclcheck($id,'',array()) < AUTH_READ) continue; 65 $item = Item::createFromID($id); 66 if ($item !== null) 67 $items[] = $item; 68 } 69 70 $eventData = array('items' => &$items, 'sitemap' => &$sitemap); 71 $event = new Event('SITEMAP_GENERATE', $eventData); 72 if ($event->advise_before(true)) { 73 //save the new sitemap 74 $event->result = io_saveFile($sitemap, Mapper::getXML($items)); 75 } 76 $event->advise_after(); 77 78 return $event->result; 79 } 80 81 /** 82 * Builds the sitemap XML string from the given array auf SitemapItems. 83 * 84 * @param $items array The SitemapItems that shall be included in the sitemap. 85 * @return string The sitemap XML. 86 * 87 * @author Michael Hamann 88 */ 89 private static function getXML($items) 90 { 91 ob_start(); 92 echo '<?xml version="1.0" encoding="UTF-8"?>'.NL; 93 echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL; 94 foreach ($items as $item) { 95 /** @var Item $item */ 96 echo $item->toXML(); 97 } 98 echo '</urlset>'.NL; 99 $result = ob_get_contents(); 100 ob_end_clean(); 101 return $result; 102 } 103 104 /** 105 * Helper function for getting the path to the sitemap file. 106 * 107 * @return string The path to the sitemap file. 108 * 109 * @author Michael Hamann 110 */ 111 public static function getFilePath() 112 { 113 global $conf; 114 115 $sitemap = $conf['cachedir'].'/sitemap.xml'; 116 if (self::sitemapIsCompressed()) { 117 $sitemap .= '.gz'; 118 } 119 120 return $sitemap; 121 } 122 123 /** 124 * Helper function for checking if the sitemap is compressed 125 * 126 * @return bool If the sitemap file is compressed 127 */ 128 public static function sitemapIsCompressed() 129 { 130 global $conf; 131 return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz'; 132 } 133 134 /** 135 * Pings search engines with the sitemap url. Plugins can add or remove 136 * urls to ping using the SITEMAP_PING event. 137 * 138 * @author Michael Hamann 139 * 140 * @return bool 141 */ 142 public static function pingSearchEngines() 143 { 144 //ping search engines... 145 $http = new DokuHTTPClient(); 146 $http->timeout = 8; 147 148 $encoded_sitemap_url = urlencode(wl('', array('do' => 'sitemap'), true, '&')); 149 $ping_urls = array( 150 'google' => 'http://www.google.com/webmasters/sitemaps/ping?sitemap='.$encoded_sitemap_url, 151 'microsoft' => 'http://www.bing.com/webmaster/ping.aspx?siteMap='.$encoded_sitemap_url, 152 'yandex' => 'http://blogs.yandex.ru/pings/?status=success&url='.$encoded_sitemap_url 153 ); 154 155 $data = array( 156 'ping_urls' => $ping_urls, 157 'encoded_sitemap_url' => $encoded_sitemap_url 158 ); 159 $event = new Event('SITEMAP_PING', $data); 160 if ($event->advise_before(true)) { 161 foreach ($data['ping_urls'] as $name => $url) { 162 dbglog("Sitemapper::PingSearchEngines(): pinging $name"); 163 $resp = $http->get($url); 164 if ($http->error) dbglog("Sitemapper:pingSearchengines(): $http->error"); 165 dbglog('Sitemapper:pingSearchengines(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp))); 166 } 167 } 168 $event->advise_after(); 169 170 return true; 171 } 172} 173