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 11*be5c1ea2SSatoshi Saharause dokuwiki\Extension\Event; 125a8d6e48SMichael Großeuse dokuwiki\HTTP\DokuHTTPClient; 13*be5c1ea2SSatoshi Saharause dokuwiki\Search\PageIndex; 14198564abSMichael Große 15432adb37SAndreas Gohr/** 16432adb37SAndreas Gohr * A class for building sitemaps and pinging search engines with the sitemap URL. 17432adb37SAndreas Gohr * 18432adb37SAndreas Gohr * @author Michael Hamann 19432adb37SAndreas Gohr */ 20432adb37SAndreas Gohrclass Mapper { 21432adb37SAndreas Gohr /** 22432adb37SAndreas Gohr * Builds a Google Sitemap of all public pages known to the indexer 23432adb37SAndreas Gohr * 24432adb37SAndreas Gohr * The map is placed in the cache directory named sitemap.xml.gz - This 25432adb37SAndreas Gohr * file needs to be writable! 26432adb37SAndreas Gohr * 27432adb37SAndreas Gohr * @author Michael Hamann 28432adb37SAndreas Gohr * @author Andreas Gohr 29432adb37SAndreas Gohr * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html 30432adb37SAndreas Gohr * @link http://www.sitemaps.org/ 31432adb37SAndreas Gohr * 32432adb37SAndreas Gohr * @return bool 33432adb37SAndreas Gohr */ 34*be5c1ea2SSatoshi Sahara public static function generate() 35*be5c1ea2SSatoshi Sahara { 36432adb37SAndreas Gohr global $conf; 37432adb37SAndreas Gohr if ($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false; 38432adb37SAndreas Gohr 39432adb37SAndreas Gohr $sitemap = Mapper::getFilePath(); 40432adb37SAndreas Gohr 41432adb37SAndreas Gohr if (file_exists($sitemap)) { 42432adb37SAndreas Gohr if (!is_writable($sitemap)) return false; 43432adb37SAndreas Gohr } else { 44432adb37SAndreas Gohr if (!is_writable(dirname($sitemap))) return false; 45432adb37SAndreas Gohr } 46432adb37SAndreas Gohr 47432adb37SAndreas Gohr if (@filesize($sitemap) && 48*be5c1ea2SSatoshi Sahara @filemtime($sitemap) > (time()-($conf['sitemap']*86400)) // 60*60*24=86400 49*be5c1ea2SSatoshi Sahara ) { 50432adb37SAndreas Gohr dbglog('Sitemapper::generate(): Sitemap up to date'); 51432adb37SAndreas Gohr return false; 52432adb37SAndreas Gohr } 53432adb37SAndreas Gohr 54432adb37SAndreas Gohr dbglog("Sitemapper::generate(): using $sitemap"); 55432adb37SAndreas Gohr 56*be5c1ea2SSatoshi Sahara $PageIndex = PageIndex::getInstance(); 57*be5c1ea2SSatoshi Sahara $pages = $PageIndex->getPages(); 58432adb37SAndreas Gohr dbglog('Sitemapper::generate(): creating sitemap using '.count($pages).' pages'); 59432adb37SAndreas Gohr $items = array(); 60432adb37SAndreas Gohr 61432adb37SAndreas Gohr // build the sitemap items 62432adb37SAndreas Gohr foreach ($pages as $id) { 63432adb37SAndreas Gohr //skip hidden, non existing and restricted files 64432adb37SAndreas Gohr if (isHiddenPage($id)) continue; 65432adb37SAndreas Gohr if (auth_aclcheck($id,'',array()) < AUTH_READ) continue; 66432adb37SAndreas Gohr $item = Item::createFromID($id); 67432adb37SAndreas Gohr if ($item !== null) 68432adb37SAndreas Gohr $items[] = $item; 69432adb37SAndreas Gohr } 70432adb37SAndreas Gohr 71432adb37SAndreas Gohr $eventData = array('items' => &$items, 'sitemap' => &$sitemap); 72*be5c1ea2SSatoshi Sahara $event = new Event('SITEMAP_GENERATE', $eventData); 73432adb37SAndreas Gohr if ($event->advise_before(true)) { 74432adb37SAndreas Gohr //save the new sitemap 75432adb37SAndreas Gohr $event->result = io_saveFile($sitemap, Mapper::getXML($items)); 76432adb37SAndreas Gohr } 77432adb37SAndreas Gohr $event->advise_after(); 78432adb37SAndreas Gohr 79432adb37SAndreas Gohr return $event->result; 80432adb37SAndreas Gohr } 81432adb37SAndreas Gohr 82432adb37SAndreas Gohr /** 83432adb37SAndreas Gohr * Builds the sitemap XML string from the given array auf SitemapItems. 84432adb37SAndreas Gohr * 85432adb37SAndreas Gohr * @param $items array The SitemapItems that shall be included in the sitemap. 86432adb37SAndreas Gohr * @return string The sitemap XML. 87432adb37SAndreas Gohr * 88432adb37SAndreas Gohr * @author Michael Hamann 89432adb37SAndreas Gohr */ 90*be5c1ea2SSatoshi Sahara private static function getXML($items) 91*be5c1ea2SSatoshi Sahara { 92432adb37SAndreas Gohr ob_start(); 93432adb37SAndreas Gohr echo '<?xml version="1.0" encoding="UTF-8"?>'.NL; 94432adb37SAndreas Gohr echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL; 95432adb37SAndreas Gohr foreach ($items as $item) { 96432adb37SAndreas Gohr /** @var Item $item */ 97432adb37SAndreas Gohr echo $item->toXML(); 98432adb37SAndreas Gohr } 99432adb37SAndreas Gohr echo '</urlset>'.NL; 100432adb37SAndreas Gohr $result = ob_get_contents(); 101432adb37SAndreas Gohr ob_end_clean(); 102432adb37SAndreas Gohr return $result; 103432adb37SAndreas Gohr } 104432adb37SAndreas Gohr 105432adb37SAndreas Gohr /** 106432adb37SAndreas Gohr * Helper function for getting the path to the sitemap file. 107432adb37SAndreas Gohr * 108432adb37SAndreas Gohr * @return string The path to the sitemap file. 109432adb37SAndreas Gohr * 110432adb37SAndreas Gohr * @author Michael Hamann 111432adb37SAndreas Gohr */ 112*be5c1ea2SSatoshi Sahara public static function getFilePath() 113*be5c1ea2SSatoshi Sahara { 114432adb37SAndreas Gohr global $conf; 115432adb37SAndreas Gohr 116432adb37SAndreas Gohr $sitemap = $conf['cachedir'].'/sitemap.xml'; 117432adb37SAndreas Gohr if (self::sitemapIsCompressed()) { 118432adb37SAndreas Gohr $sitemap .= '.gz'; 119432adb37SAndreas Gohr } 120432adb37SAndreas Gohr 121432adb37SAndreas Gohr return $sitemap; 122432adb37SAndreas Gohr } 123432adb37SAndreas Gohr 124432adb37SAndreas Gohr /** 125432adb37SAndreas Gohr * Helper function for checking if the sitemap is compressed 126432adb37SAndreas Gohr * 127432adb37SAndreas Gohr * @return bool If the sitemap file is compressed 128432adb37SAndreas Gohr */ 129*be5c1ea2SSatoshi Sahara public static function sitemapIsCompressed() 130*be5c1ea2SSatoshi Sahara { 131432adb37SAndreas Gohr global $conf; 132432adb37SAndreas Gohr return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz'; 133432adb37SAndreas Gohr } 134432adb37SAndreas Gohr 135432adb37SAndreas Gohr /** 136432adb37SAndreas Gohr * Pings search engines with the sitemap url. Plugins can add or remove 137432adb37SAndreas Gohr * urls to ping using the SITEMAP_PING event. 138432adb37SAndreas Gohr * 139432adb37SAndreas Gohr * @author Michael Hamann 140432adb37SAndreas Gohr * 141432adb37SAndreas Gohr * @return bool 142432adb37SAndreas Gohr */ 143*be5c1ea2SSatoshi Sahara public static function pingSearchEngines() 144*be5c1ea2SSatoshi Sahara { 145432adb37SAndreas Gohr //ping search engines... 146198564abSMichael Große $http = new DokuHTTPClient(); 147432adb37SAndreas Gohr $http->timeout = 8; 148432adb37SAndreas Gohr 149432adb37SAndreas Gohr $encoded_sitemap_url = urlencode(wl('', array('do' => 'sitemap'), true, '&')); 150432adb37SAndreas Gohr $ping_urls = array( 151432adb37SAndreas Gohr 'google' => 'http://www.google.com/webmasters/sitemaps/ping?sitemap='.$encoded_sitemap_url, 152432adb37SAndreas Gohr 'microsoft' => 'http://www.bing.com/webmaster/ping.aspx?siteMap='.$encoded_sitemap_url, 153432adb37SAndreas Gohr 'yandex' => 'http://blogs.yandex.ru/pings/?status=success&url='.$encoded_sitemap_url 154432adb37SAndreas Gohr ); 155432adb37SAndreas Gohr 156*be5c1ea2SSatoshi Sahara $data = array( 157*be5c1ea2SSatoshi Sahara 'ping_urls' => $ping_urls, 158432adb37SAndreas Gohr 'encoded_sitemap_url' => $encoded_sitemap_url 159432adb37SAndreas Gohr ); 160*be5c1ea2SSatoshi Sahara $event = new Event('SITEMAP_PING', $data); 161432adb37SAndreas Gohr if ($event->advise_before(true)) { 162432adb37SAndreas Gohr foreach ($data['ping_urls'] as $name => $url) { 163432adb37SAndreas Gohr dbglog("Sitemapper::PingSearchEngines(): pinging $name"); 164432adb37SAndreas Gohr $resp = $http->get($url); 165432adb37SAndreas Gohr if ($http->error) dbglog("Sitemapper:pingSearchengines(): $http->error"); 166432adb37SAndreas Gohr dbglog('Sitemapper:pingSearchengines(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp))); 167432adb37SAndreas Gohr } 168432adb37SAndreas Gohr } 169432adb37SAndreas Gohr $event->advise_after(); 170432adb37SAndreas Gohr 171432adb37SAndreas Gohr return true; 172432adb37SAndreas Gohr } 173432adb37SAndreas Gohr} 174