123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165 |
- <?php
- /**
- * Sitemap handling functions
- *
- * @license GPL 2 (http://www.gnu.org/licenses/gpl.html)
- * @author Michael Hamann <michael@content-space.de>
- */
- namespace dokuwiki\Sitemap;
- use dokuwiki\HTTP\DokuHTTPClient;
- use dokuwiki\Logger;
- /**
- * A class for building sitemaps and pinging search engines with the sitemap URL.
- *
- * @author Michael Hamann
- */
- class Mapper {
- /**
- * Builds a Google Sitemap of all public pages known to the indexer
- *
- * The map is placed in the cache directory named sitemap.xml.gz - This
- * file needs to be writable!
- *
- * @author Michael Hamann
- * @author Andreas Gohr
- * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html
- * @link http://www.sitemaps.org/
- *
- * @return bool
- */
- public static function generate(){
- global $conf;
- if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
- $sitemap = Mapper::getFilePath();
- if(file_exists($sitemap)){
- if(!is_writable($sitemap)) return false;
- }else{
- if(!is_writable(dirname($sitemap))) return false;
- }
- if(@filesize($sitemap) &&
- @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
- Logger::debug('Sitemapper::generate(): Sitemap up to date');
- return false;
- }
- Logger::debug("Sitemapper::generate(): using $sitemap");
- $pages = idx_get_indexer()->getPages();
- Logger::debug('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
- $items = array();
- // build the sitemap items
- foreach($pages as $id){
- //skip hidden, non existing and restricted files
- if(isHiddenPage($id)) continue;
- if(auth_aclcheck($id,'',array()) < AUTH_READ) continue;
- $item = Item::createFromID($id);
- if ($item !== null)
- $items[] = $item;
- }
- $eventData = array('items' => &$items, 'sitemap' => &$sitemap);
- $event = new \dokuwiki\Extension\Event('SITEMAP_GENERATE', $eventData);
- if ($event->advise_before(true)) {
- //save the new sitemap
- $event->result = io_saveFile($sitemap, Mapper::getXML($items));
- }
- $event->advise_after();
- return $event->result;
- }
- /**
- * Builds the sitemap XML string from the given array auf SitemapItems.
- *
- * @param $items array The SitemapItems that shall be included in the sitemap.
- * @return string The sitemap XML.
- *
- * @author Michael Hamann
- */
- private static function getXML($items) {
- ob_start();
- echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
- echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
- foreach ($items as $item) {
- /** @var Item $item */
- echo $item->toXML();
- }
- echo '</urlset>'.NL;
- $result = ob_get_contents();
- ob_end_clean();
- return $result;
- }
- /**
- * Helper function for getting the path to the sitemap file.
- *
- * @return string The path to the sitemap file.
- *
- * @author Michael Hamann
- */
- public static function getFilePath() {
- global $conf;
- $sitemap = $conf['cachedir'].'/sitemap.xml';
- if (self::sitemapIsCompressed()) {
- $sitemap .= '.gz';
- }
- return $sitemap;
- }
- /**
- * Helper function for checking if the sitemap is compressed
- *
- * @return bool If the sitemap file is compressed
- */
- public static function sitemapIsCompressed() {
- global $conf;
- return $conf['compression'] === 'bz2' || $conf['compression'] === 'gz';
- }
- /**
- * Pings search engines with the sitemap url. Plugins can add or remove
- * urls to ping using the SITEMAP_PING event.
- *
- * @author Michael Hamann
- *
- * @return bool
- */
- public static function pingSearchEngines() {
- //ping search engines...
- $http = new DokuHTTPClient();
- $http->timeout = 8;
- $encoded_sitemap_url = urlencode(wl('', array('do' => 'sitemap'), true, '&'));
- $ping_urls = array(
- 'google' => 'https://www.google.com/ping?sitemap='.$encoded_sitemap_url,
- 'yandex' => 'https://webmaster.yandex.com/ping?sitemap='.$encoded_sitemap_url
- );
- $data = array('ping_urls' => $ping_urls,
- 'encoded_sitemap_url' => $encoded_sitemap_url
- );
- $event = new \dokuwiki\Extension\Event('SITEMAP_PING', $data);
- if ($event->advise_before(true)) {
- foreach ($data['ping_urls'] as $name => $url) {
- Logger::debug("Sitemapper::PingSearchEngines(): pinging $name");
- $resp = $http->get($url);
- if($http->error) {
- Logger::debug("Sitemapper:pingSearchengines(): $http->error", $resp);
- }
- }
- }
- $event->advise_after();
- return true;
- }
- }
|