API tools faq
paste
Advertisement
Guest User

Untitled

a guest
Aug 19th, 2019
82
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 17.89 KB | None | 0 0
raw download clone embed print report
  1. <?php
  2.  
  3. use WSS\Alfresco\AlfrescoUtils;
  4.  
  5. /**
  6. * @author wichmann
  7. *
  8. */
  9. class WSS_Site {
  10.  
  11. const SITEMAP_XML_FILENAME = "sitemap.xml";
  12.  
  13. /**
  14. * The Units that shall not get a sitemap in any case (they may be set to "online" ...)
  15. * @var array
  16. */
  17. private static $UNWANTED_UNITS = array('comuser','test','training','user');
  18.  
  19.  
  20. private $unit_key_filter;
  21. private $output_type = "XML"; // HTML|XML default = XML
  22. private $used_downloads;
  23.  
  24. /**
  25. *
  26. * instanzierbar mit einem pfad z.b. www.schott.com/tubing/ dann wird nur eine sitemap
  27. * für alle seiten erstellt die unter tubing liegen
  28. * @param string $unit_key_filter just the unit key, no path
  29. */
  30. public function __construct($unit_key_filter= null) {
  31. $this->unit_key_filter = $unit_key_filter;
  32. }
  33.  
  34. public function create_sitemap_index() {
  35. $doc = new DOMDocument('1.0', 'UTF-8');
  36. $root = $doc->createElementNS("http://www.sitemaps.org/schemas/sitemap/0.9",'sitemapindex');
  37. $doc->appendChild($root);
  38. $units = WSS_Unit_Table::instance()->findAll(null, true, WSS_Config::isUs() );
  39. foreach ($units as $unit){
  40. $unitPath = $unit->createUnitPath();
  41. if ($unitPath == '')
  42. $unitPath = WSS_Path_Model::createUnitLanguagePath($unit->unit_key, $unit->unit_key);
  43. $sitemapPath = $unitPath . '/' . self::SITEMAP_XML_FILENAME;
  44. $sitemapAbsolutePath = $_SERVER['DOCUMENT_ROOT'] . $sitemapPath;
  45. if($this->isUnitRelevant($unit->unit_key)){
  46.  
  47. $this->unit_key_filter = $unit->unit_key;
  48. // old sitemap file may not exist yet
  49. if (file_exists($sitemapAbsolutePath)) {
  50. $oldSitemap = file_get_contents($sitemapAbsolutePath);
  51. } else {
  52. $oldSitemap = '';
  53. }
  54. $newSitemap = $this->create_sitemap();
  55. if (strcmp($oldSitemap, $newSitemap)!=0) {
  56. file_put_contents($sitemapAbsolutePath, $newSitemap);
  57. }
  58.  
  59. $root->appendChild($this->addSitemapIndexElement($doc, WSS_Url::create($sitemapPath)->render(true), WSS_Date::unix_timestamp_to_w3c(filemtime($sitemapAbsolutePath))));
  60.  
  61. } else {
  62.  
  63. if(file_exists($sitemapAbsolutePath))
  64. unlink($sitemapAbsolutePath);
  65.  
  66. }
  67. }
  68.  
  69. clearstatcache();
  70. foreach (WSS_Config::instance()->get('sitemap.index-additions.uri') as $sitemapUri) {
  71. if (WSS_Url::exists($sitemapUri) || WSS_Url::isRedirect($sitemapUri)) {
  72. $root->appendChild($this->addSitemapIndexElement($doc, $sitemapUri));
  73. }
  74. }
  75.  
  76. $doc->formatOutput = true;
  77. $indexSitemap = $doc->saveXML();
  78. file_put_contents($_SERVER['DOCUMENT_ROOT'] . '/' . self::SITEMAP_XML_FILENAME, $indexSitemap);
  79.  
  80. return $indexSitemap;
  81. }
  82.  
  83. private function addSitemapIndexElement(DOMDocument $doc, $loc, $lastmod = null) {
  84. $sitemap = $doc->createElement('sitemap');
  85. $sitemap->appendChild($doc->createElement('loc',$loc));
  86. if (!empty($lastmod)) {
  87. $sitemap->appendChild($doc->createElement('lastmod',$lastmod));
  88. }
  89. return $sitemap;
  90. }
  91.  
  92. public function create_sitemap( $output_type = 'XML' ) {
  93.  
  94. if( $output_type == "HTML" )
  95. {
  96. $this->output_type = "HTML";
  97. }
  98.  
  99. if( $this->output_type == "HTML" )
  100. {
  101. // HTML OUTPUT
  102. $sitemap = "<html>\n";
  103. $sitemap .= "<head>\n";
  104. $sitemap .= "<title>Sitemap</title>\n";
  105. $sitemap .= "</head>\n";
  106. $sitemap .= "<body>\n";
  107. $sitemap .= "<ul>";
  108. $sitemap .= $this->create_sitemap_entries();
  109. $sitemap .= "</ul>";
  110. $sitemap .= "</body>\n";
  111. $sitemap .= "</html>\n";
  112. }else{
  113. // XML OUTPUT
  114. $sitemap = "<?xml version='1.0' encoding='UTF-8'?>\n";
  115. $sitemap .= "<urlset xmlns='http://www.sitemaps.org/schemas/sitemap/0.9'>\n";
  116. $sitemap .= $this->create_sitemap_entries();
  117. $sitemap .= "</urlset>\n";
  118. }
  119. return $sitemap;
  120. }
  121.  
  122. public function createVideoSitemap() {
  123. $doc = new DOMDocument('1.0', 'UTF-8');
  124. $urlSet = $doc->createElementNS('http://www.sitemaps.org/schemas/sitemap/0.9', 'urlset');
  125. $doc->appendChild( $urlSet );
  126. // We add videos only once to the sitemap.
  127. $collectedVideos = array();
  128. // First collect corporate videos
  129. foreach (WSS_Unit_Language_Table::instance()->findAll(array(WSS_Unit_Row::TYPE_CORPORATE), true, WSS_Config::isUs())
  130. as $unit) {
  131. $collectedVideos = $this->createVideoSitemapForPath($unit->createUnitLanguagePath(), $doc, $collectedVideos);
  132. }
  133. // Then collect other videos
  134. $this->createVideoSitemapForPath('%', $doc, $collectedVideos);
  135. $doc->formatOutput = true;
  136. $content = $doc->saveXML();
  137. $sitemapPath = $_SERVER['DOCUMENT_ROOT'] . '/video_sitemap.xml';
  138. file_put_contents($sitemapPath, $content);
  139. }
  140.  
  141. public function createTubingSelectorSitemap($saveXmlFile = true) {
  142. $products = [];
  143. if (WSS_Config::isSchott() && WSS_Config::isCom()) {
  144. $doc = new DOMDocument('1.0', 'UTF-8');
  145. $doc->appendChild($doc->createElementNS('http://www.sitemaps.org/schemas/sitemap/0.9', 'urlset'));
  146.  
  147. $masterUrl = WSS_Config::getSchemeAndHost().'/common/tubing/selector-rest-master-settings.php?language=english&region=all';
  148. $rawMasterContent = file_get_contents($masterUrl);
  149. if($rawMasterContent){
  150. // Increase timeout for fetching selector meta data:
  151. // As of 11.06. 2018 necessary on preview servers. This might be because of additionally set dl categories.
  152. $ctx = stream_context_create(array('http'=>
  153. array(
  154. 'timeout' => 180, // Seconds
  155. ),
  156. 'https'=>
  157. array(
  158. 'timeout' => 180, // Seconds
  159. ),
  160. ));
  161. foreach (json_decode($rawMasterContent)->masterSettings->language_list as $lang) {
  162. $catalogueUrl = WSS_Config::getSchemeAndHost().'/common/tubing/selector-rest-catalogue.php?language='.$lang. '&region=all';
  163. $rawCatalogueContent = file_get_contents($catalogueUrl, false, $ctx);
  164. if($rawCatalogueContent){
  165. foreach (json_decode($rawCatalogueContent) as $product) {/* @see Tubing_Selector_CatalogueJson */
  166. $path = '/tubing/english/product_selector/#!/region--all/lang--' . $lang . '/product--' . $product->productId . '/';
  167. $doc->getElementsByTagName('urlset')->item(0)->appendChild($this->createEntry($doc, $path, WSS_Date::unix_timestamp_to_w3c($product->modificationDate)));
  168. $products[] = ['loc' => $path, 'lastmod' => $product->modificationDate];
  169. }
  170. } else {
  171. WSS_Logger::warn('XML site: Tubing product meta data call failed: '.$catalogueUrl);
  172. }
  173. $path = '/tubing/english/product_selector/#!/region--all/lang--'.$lang.'/filter--list/';
  174. $doc->getElementsByTagName('urlset')->item(0)->appendChild($this->createEntry($doc, $path));
  175. $products[] = ['loc' => $path];
  176. }
  177. } else {
  178. WSS_Logger::warn('XML site: Tubing product master data call failed: '.$masterUrl);
  179. }
  180. $doc->formatOutput = true;
  181. if ($saveXmlFile && count($products) > 0) {
  182. file_put_contents($_SERVER['DOCUMENT_ROOT'] . '/tubing/english/product_selector/sitemap.xml', $doc->saveXML());
  183. }
  184. }
  185. return $products;
  186. }
  187.  
  188. /**
  189. *
  190. * @param string $path
  191. * @param DOMDocument $doc
  192. * @param array $collectedVideos
  193. * @return array collected videos with added new onews
  194. */
  195. private function createVideoSitemapForPath($path, DOMDocument $doc, array $collectedVideos) {
  196.  
  197. $is_us = WSS_Config::isUs();
  198. $seitenTable = WSS_Pagepart_Table::instance();
  199. foreach ($seitenTable->findByBlock('media_videolibrary.slf', $path.'%') as $seiten) {
  200. $status = $seiten->findPage();
  201. if ($status->is_online($is_us)) {
  202. $path = $status->create_path();
  203. $url = $this->createEntry($doc, $path, WSS_Date::mysql_timestamp_to_w3c($status->cdate));
  204. // might be null if filtered
  205. if ($url != null) {
  206. foreach ($seitenTable->findByBlock('media_videolibrary_item.slf', $seiten->URLDIR,$seiten->URLFILE) as $itemSeiten) {
  207. $seitenTitle = $seitenTable->findByPathAndBaustein($path, $itemSeiten->getChildBaustein('Title',$itemSeiten->selfID));
  208. $title = null;
  209. $thumbnail_loc = null;
  210. $videoAsset = WSS_Asset_Null::newInstance();
  211. $Teaser = '';
  212. if ($seitenTitle!=null) {
  213. $title = $seitenTitle->fetchContent();
  214. }
  215. $seitencontent_loc = $seitenTable->findByPathAndBaustein($path, $itemSeiten->getChildBaustein('Video',$itemSeiten->selfID));
  216. if ($seitencontent_loc!=null) {
  217. $videoAsset= WSS_Asset_ManagerFactory::create()->find($seitencontent_loc->alfresco_id);
  218. }
  219. $seitenthumbnail_loc = $seitenTable->findByPathAndBaustein($path, $itemSeiten->getChildBaustein('Image',$itemSeiten->selfID));
  220. if ($seitenthumbnail_loc!=null) {
  221. $thumbnail_loc = $seitenthumbnail_loc->IncludeFile;
  222. }
  223. $seitenTeaser = $seitenTable->findByPathAndBaustein($path, $itemSeiten->getChildBaustein('Teaser',$itemSeiten->selfID));
  224. if ($seitenTeaser!=null) {
  225. $Teaser = $seitenTeaser->fetchContent();
  226. }
  227. if ($title!=''
  228. && $thumbnail_loc!=''
  229. && !$videoAsset->isEmpty()
  230. ) {
  231. if (isset($collectedVideos[$videoAsset->getId()])
  232. || array_search($title, $collectedVideos)!==FALSE) {
  233. // do not add video twice
  234. // We also do not add videos with same title because Google does
  235. // not like this, although this may cause loss of some really different videos
  236. continue;
  237. } else {
  238. $collectedVideos[$videoAsset->getId()] = $title;
  239. }
  240. $video = $url->appendChild($doc->createElement('video:video'));
  241. $titleElement = $video->appendChild($doc->createElement('video:title'));
  242. $titleElement->appendChild($doc->createCDATASection($title));
  243. if ($Teaser=='') {
  244. $description = $title;
  245. } else {
  246. $description = $Teaser;
  247. }
  248. $descriptionElement = $video->appendChild($doc->createElement('video:description'));
  249. $descriptionElement->appendChild($doc->createCDATASection($description));
  250. $video->appendChild($doc->createElement('video:thumbnail_loc',WSS_Url::create($thumbnail_loc)->render(true)));
  251. $video->appendChild($doc->createElement('video:content_loc',$videoAsset->getDocument()));
  252. if (!empty($videoAsset->getDuration())) {
  253. $video->appendChild($doc->createElement('video:duration',$videoAsset->getDuration()));
  254. }
  255. }
  256. }
  257. // Only add video page if it contains new videos
  258. if ($url->getElementsByTagName('video')) {
  259. $doc->getElementsByTagName('urlset')->item(0)->appendChild($url);
  260. }
  261. }
  262. }
  263. }
  264.  
  265. return $collectedVideos;
  266. }
  267.  
  268. private function create_sitemap_entries() {
  269. $sitemap = '';
  270. $sitemap .= $this->create_sitemap_page_entries();
  271. $sitemap .= $this->create_sitemap_downloads_entries();
  272. $unit = WSS_Unit_Table::fetchFromCache($this->unit_key_filter);
  273. if ($unit->is_corporate())
  274. $sitemap .= $this->create_sitemap_news_entries();
  275. if($unit->is_corporate() || $unit->is_so())
  276. $sitemap .= $this->create_sitemap_jobs_entries($unit);
  277. //TODO still required after Alf migration?
  278. if($this->unit_key_filter == 'advanced_optics') {
  279. $sitemap .= $this->create_abbe_downloads_entries();
  280. } else if($this->unit_key_filter == 'tubing' || $this->output_type == "HTML") {
  281. $sitemap .= $this->createTubingSelectorEntries();
  282. }
  283.  
  284. return $sitemap;
  285. }
  286. /**
  287. *
  288. * @param DOMDocument $doc
  289. * @param string $path
  290. * @param string $lastmod
  291. * @return DOMElement|NULL
  292. */
  293. private function createEntry( DOMDocument $doc, $path, $lastmod = null) {
  294. if ($this->isPathRelevant($path)) {
  295. $loc = $this->getLocLink($path);
  296. $url = $doc->createElement('url');
  297. $url->appendChild($doc->createElement('loc',$loc));
  298. if ( $lastmod != NULL ) {
  299. $url->appendChild($doc->createElement('lastmod',$lastmod));
  300. }
  301. return $url;
  302. } else {
  303. return null;
  304. }
  305. }
  306.  
  307. public function isUnitRelevant( $unitKey ) {
  308. if ($unitKey==WSS_Unit_Row::CORPORATE_UNIT_PSEUDO_KEY) {
  309. // handling corporate download links
  310. $unitKey = 'english';
  311. }
  312. if(in_array($unitKey, static::$UNWANTED_UNITS)){
  313. return false;
  314. }
  315. $unit = WSS_Unit_Table::fetchFromCache($unitKey);
  316. if ( $unit == null || ! $unit->is_online()) {
  317. return false;
  318. }
  319. return true;
  320. }
  321.  
  322. public function isPathRelevant( $path ) {
  323. // Check for excluded pathes
  324. $offlinePathes = WSS_Config::instance()->get('visibility.search-engine-restricted');
  325. if ($offlinePathes != '' && preg_match($offlinePathes, $path) == 1) {
  326. return false;
  327. }
  328. $pathParts = parse_url($path);
  329. $unitKey = WSS_Path_Model::newInstance($pathParts['path'])->getUnitDirectory();
  330. return $this->isUnitRelevant($unitKey);
  331. }
  332.  
  333. private function getLocLink($path, $isAbsolutePath = false) {
  334. $loc = ($isAbsolutePath) ? $path : WSS_Url::create($path)->render(true);
  335. $loc = WSS_String::specialChars($loc);
  336. return $loc;
  337. }
  338. /**
  339. *
  340. * @param string $path: may contain query string
  341. * @param string $lastmod: W3C format like "yyyy-MM-dd'T'HH:mm:ss'+01:00'"
  342. * @param string $unitKey: given for Alfresco downloads. If set then this param is relevant for filtering only.
  343. * @param boolean $isAbsolutePath
  344. * @return string
  345. */
  346. public function create_sitemap_entry( $path, $lastmod, $unitKey = null, $isAbsolutePath = false )
  347. {
  348. if ($unitKey==null ) {
  349. if (!$this->isPathRelevant($path)) {
  350. return '';
  351. }
  352. } else {
  353. if (!$this->isUnitRelevant($unitKey)) {
  354. return '';
  355. }
  356. }
  357.  
  358. $entry = '';
  359. $loc = $this->getLocLink($path, $isAbsolutePath);
  360. if( $this->output_type == "HTML")
  361. {
  362. // HTML OUTPUT
  363. $entry = "<li><a href=\"" . $loc . "\">" . $loc . "</a></li>\n";
  364. }else{
  365. // XML OUTPUT
  366. $entry = "<url>\n";
  367. $entry .= "<loc>".$loc."</loc>\n";
  368. if ( $lastmod != NULL ) {
  369. $entry .= "<lastmod>".$lastmod."</lastmod>\n";
  370. }
  371. $entry .= "</url>\n";
  372. }
  373.  
  374. return $entry;
  375. }
  376.  
  377. private function create_sitemap_downloads_entries() {
  378. $pages = "";
  379. if (!isset($this->used_downloads)) {
  380.  
  381. $this->used_downloads = [];
  382. $units = WSS_Unit_Table::instance()->findAll(null, true, WSS_Config::isUs() );
  383. foreach ($units as $unit){
  384. //TODO not all dls are added due to Alf limit of 1000 search results
  385. foreach (WSS_Asset_ManagerFactory::create()->createQuery()
  386. ->setRegionBlacklist()
  387. ->setPath(WSS_Asset_XPath::newInstance($unit->unit_key)->toOnlineDownloadFolderRecursive())
  388. //->setPath(WSS_Asset_XPath::newInstanceDefault()->toOnlineDownloadFolderRecursiveAllUnits())
  389. ->getResultList() as $download) { /* @var $download WSS_Asset_Base */
  390. $path = $download->downloadPath();
  391. $this->used_downloads[$path] = $download;
  392. }
  393. }
  394.  
  395. }
  396. foreach ( $this->used_downloads as $download_path => $value ) {
  397. $pages .= $this->create_sitemap_entry( $download_path,
  398. date("c", strtotime($value->getModified())),
  399. AlfrescoUtils::unitKeyFromQNamePath($value->getQnamePath()) );
  400. }
  401. return $pages;
  402. }
  403.  
  404. private function create_sitemap_jobs_entries(WSS_Unit_Row $unit) {
  405. $pages = '';
  406. foreach($unit->findLanguages() as $language){
  407. $jobsPagePath = WSS_Unit_Path::getByUnitLanguagePath(WSS_Unit_Path::JOBS, $unit->getUnitPath().'/'.$language);
  408. if(WSS_Page_Table::instance()->isOnlinePath($jobsPagePath)){
  409. $locale = WSS_Language::getLocale($language);
  410. if($locale == null)
  411. $locale = 'en_US';
  412. foreach(WSS_Bootstrap::getService('WSSSuccessFactorsService')->createSitemapJobs($locale, $jobsPagePath) as $job){
  413. $pages .= $this->create_sitemap_entry(WSS_Config::getSchemeAndHost().$job->link, date('c', $job->timestamp), null, true);
  414. }
  415. }
  416. }
  417. return $pages;
  418. }
  419.  
  420. private function create_sitemap_news_entries() {
  421. $pages = '';
  422. $selection = WSS_News_Selection::create();
  423. $selection->setIsOnlineFiltered(WSS_News_Table::MAIN_UNIT_KEY);
  424. $selection->setShowArchivedNews(true);
  425. $newsList = $selection->fetch();
  426. foreach ( $newsList as $wssNews ) {
  427. $link = $wssNews->getItemLink();
  428. $pages .= $this->create_sitemap_entry($link,
  429. $wssNews->get_newsdate() );
  430. }
  431. return $pages;
  432. }
  433.  
  434. private function create_sitemap_page_entries() {
  435. $pages = "";
  436. $table = new WSS_Page_Table();
  437. $isUS = WSS_Config::isUs();
  438. $path = '/'.$this->unit_key_filter;
  439. $pages .= $this->addWSS_Page_Row($table->findVisitorContent($path, $isUS));
  440. if ( $isUS ) {
  441. $pages .= $this->addWSS_Page_Row($table->findOnlineFiltered($path));
  442. }
  443. return $pages;
  444. }
  445.  
  446. private function addWSS_Page_Row($list) {
  447. $pages = "";
  448. foreach ( $list as $status_Settings ) {
  449. if ( $status_Settings->isRelevantForCrawler() ) {
  450. $pages .= $this->create_sitemap_page_entry( $status_Settings );
  451. }
  452. }
  453. return $pages;
  454. }
  455.  
  456. private function create_sitemap_page_entry( WSS_Page_Row $status_Settings ) {
  457. return $this->create_sitemap_entry( $status_Settings->create_path(),
  458. WSS_Date::mysql_timestamp_to_w3c($status_Settings->cdate) );
  459. }
  460.  
  461. private function create_abbe_downloads_entries()
  462. {
  463. // require does not work for unit tests
  464. require_once($_SERVER['DOCUMENT_ROOT'].'/advanced_optics/abbe_pdf/datasheets.inc');
  465. $pages = "";
  466.  
  467. $items = abbe_collect_downloads();
  468. foreach ( $items as $path )
  469. {
  470. $absolute_path = $_SERVER['DOCUMENT_ROOT'].$path;
  471. $lastmod = filemtime($absolute_path);
  472. $pages .= $this->create_sitemap_entry($path,
  473. WSS_Date::unix_timestamp_to_w3c($lastmod));
  474. }
  475. return $pages;
  476. }
  477.  
  478. private function createTubingSelectorEntries()
  479. {
  480. $pages = '';
  481. foreach($this->createTubingSelectorSitemap(false) as $product){
  482. $pages .= $this->create_sitemap_entry($product['loc'], empty($product['lastmod']) ? '' : WSS_Date::unix_timestamp_to_w3c($product['lastmod']));
  483. }
  484. return $pages;
  485. }
  486.  
  487. }
  488.  
  489.  
  490. ?>
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!