1 <?php
  2 /**
  3  * Initial implementation of Sitemap support.
  4  * GoogleSitemap should handle requests to 'sitemap.xml'
  5  * the other two classes are used to render the sitemap.
  6  * 
  7  * You can notify ("ping") Google about a changed sitemap
  8  * automatically whenever a new page is published or unpublished.
  9  * By default, Google is not notified, and will pick up your new
 10  * sitemap whenever the GoogleBot visits your website.
 11  * 
 12  * Enabling notification of Google after every publish (in your _config.php):
 13  * <example
 14  * GoogleSitemap::enable_google_notificaton();
 15  * </example>
 16  * 
 17  * @see http://www.google.com/support/webmasters/bin/answer.py?hl=en&answer=34609
 18  * 
 19  * @package googlesitemaps
 20  */
 21 class GoogleSitemap extends Controller {
 22     
 23     static $cache_expire = 86400;
 24     
 25     static function local_sitemap_path() {
 26         return ASSETS_PATH . '/sitemap.xml';
 27     }
 28 
 29     static function clear_cache() {
 30         $file = self::local_sitemap_path();
 31         if (@file_exists($file))
 32             unlink($file);
 33     }
 34     
 35     /**
 36      * @var boolean
 37      */
 38     protected static $enabled = true;
 39     
 40     /**
 41      * @var boolean
 42      */
 43     protected static $google_notification_enabled = false;
 44     
 45     /**
 46      * Использовать флаг ShowInSiteMap(он есть не во всех версиях Webylon) @var boolean
 47      */
 48     protected static $use_show_in_sitemap = true;
 49     
 50     //Установка использования флага ShowInSiteMap
 51     static function set_use_show_in_sitemap($val = false) {
 52         self::$use_show_in_sitemap = $val;
 53     }
 54     
 55     public function Items($start=0) {
 56         $count = 1000;
 57         Versioned::reading_stage('Live');
 58         $filter = '';
 59         if(self::$use_show_in_sitemap) {
 60             $filter .= '"ShowInSiteMap" = 1';
 61         }
 62         
 63         $pages = DataObject::get('SiteTree', $filter, '', '', "{$start},{$count}");
 64         if (!$pages) return false;
 65         $newPages = new DataObjectSet();
 66         while($page = $pages->shift()) {
 67             // Only include pages from this host and pages which are not an instance of ErrorPage 
 68             // We prefix $_SERVER['HTTP_HOST'] with 'http://' so that parse_url to help parse_url identify the host name component; we could use another protocol (like 
 69             // 'ftp://' as the prefix and the code would work the same. 
 70             if (parse_url($page->AbsoluteLink(), PHP_URL_HOST) != parse_url('http://' . $_SERVER['HTTP_HOST'], PHP_URL_HOST)) continue;
 71             if ($page instanceof ErrorPage) continue;
 72             if ($page->Priority < 0) continue;
 73             if (!$page->canView()) continue;
 74             
 75             // The one field that isn't easy to deal with in the template is
 76             // Change frequency, so we set that here.
 77             $created = new SS_Datetime();
 78             $created->value = $page->Created;
 79 
 80             $now = new SS_Datetime();
 81             $now->value = date('Y-m-d H:i:s');
 82 
 83             $versions = $page->Version;
 84             $timediff = $now->format('U') - $created->format('U');
 85 
 86             // Check how many revisions have been made over the lifetime of the
 87             // Page for a rough estimate of it's changing frequency.
 88 
 89             $period = $timediff / ($versions + 1);
 90 
 91             if($period > 60*60*24*365) { // > 1 year
 92                 $page->ChangeFreq='yearly';
 93             } elseif($period > 60*60*24*30) { // > ~1 month
 94                 $page->ChangeFreq='monthly';
 95             } elseif($period > 60*60*24*7) { // > 1 week
 96                 $page->ChangeFreq='weekly';
 97             } elseif($period > 60*60*24) { // > 1 day
 98                 $page->ChangeFreq='daily';
 99             } elseif($period > 60*60) { // > 1 hour
100                 $page->ChangeFreq='hourly';
101             } else { // < 1 hour
102                 $page->ChangeFreq='always';
103             }
104 
105             $newPages->push($page);
106         }
107         return $newPages;
108     }
109     
110     /**
111      * Notifies Google about changes to your sitemap.
112      * Triggered automatically on every publish/unpublish of a page.
113      * This behaviour is disabled by default, enable with:
114      * GoogleSitemap::enable_google_notificaton();
115      * 
116      * If the site is in "dev-mode", no ping will be sent regardless wether
117      * the Google notification is enabled.
118      * 
119      * @return string Response text
120      */
121     static function ping() {
122         if(!self::$enabled) return false;
123         
124         //Don't ping if the site has disabled it, or if the site is in dev mode
125         if(!GoogleSitemap::$google_notification_enabled || Director::isDev())
126             return;
127             
128         $location = urlencode(Director::absoluteBaseURL() . '/sitemap.xml');
129         
130         $response = HTTP::sendRequest("www.google.com", "/webmasters/sitemaps/ping",
131             "sitemap=" . $location);
132             
133         return $response;
134     }
135     
136     /**
137      * Enable pings to google.com whenever sitemap changes.
138      */
139     public static function enable_google_notification() {
140         self::$google_notification_enabled = true;
141     }
142     
143     /**
144      * Disables pings to google when the sitemap changes.
145      */
146     public static function disable_google_notification() {
147         self::$google_notification_enabled = false;
148     }
149     
150     function index($request) {
151         if(self::$enabled) {
152             $cache_path = self::local_sitemap_path();
153             if ($request->requestVar('force')) {
154                 self::clear_cache();
155             }
156             
157             if (!Director::is_cli()) {
158                 if (@file_exists($cache_path) && (@filemtime($cache_path) + self::$cache_expire > time())) {
159                     Controller::curr()->getResponse()->addHeader("Content-Type", "text/xml");
160                     readfile($cache_path);                  
161                     return;
162                 }
163             }
164         
165             increase_time_limit_to(1000);
166             increase_memory_limit_to('1000M');
167             $markStart = time();
168             $html = '<?xml version="1.0" encoding="UTF-8"?><urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">';           
169             file_put_contents($cache_path, $html);
170             
171             $start = 0;
172             $totCount = 0;
173             $f = true;
174             while ($f) {
175                 $pages = $this->Items($start);
176                 if (!$pages) {
177                     $f = false;
178                 } else {
179                     $html = '';
180                     $pCount = $pages->Count();
181                     foreach($pages as $page) {
182                         $html .= $page->renderWith('SitemapURL');
183                         $page->destroy();
184                         $totCount++;
185                     }
186                     file_put_contents($cache_path, $html, FILE_APPEND);
187                     $start += $pCount;
188                     unset($pages);
189                 }
190             }
191             file_put_contents($cache_path, '</urlset>', FILE_APPEND);
192         
193             if (Director::is_cli()) {
194                 $markEnd = time();
195                 echo 
196                     'start: ' . date('Y-m-d H:i:s', $markStart) . "\n"
197                     . '  end: ' . date('Y-m-d H:i:s', $markEnd) . "\n"
198                     . 'duration: ' . ($markEnd - $markStart) . "\n"
199                     . Director::absoluteURL(Director::makeRelative($cache_path)) ."\n"
200                 ;
201             }
202             else {
203                 Controller::curr()->getResponse()->addHeader("Content-Type", "text/xml");
204                 readfile($cache_path);
205             }
206             
207             // But we want to still render.
208             //return array();
209         } else {
210             return new SS_HTTPResponse('Not allowed', 405);
211         }
212     }
213     
214     public static function enable() {
215         self::$enabled = true;
216     }
217     
218     public static function disable() {
219         self::$enabled = false;
220     }
221 }
222 
223