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