public function testGetWebPageDetails() { $details = URLExpander::getWebPageDetails('http://google.com'); print_r($details); $details = URLExpander::getWebPageDetails('http://smarterware.org'); print_r($details); $details = URLExpander::getWebPageDetails("http://www.slate.com/blogs/future_tense/2013/08/29/klout_deletes" . "_tweet_martin_luther_king_would_have_had_an_awesome_klout_score.html?utm_content=buffer9b03f&utm_source=bu" . "ffer&utm_medium=twitter&utm_campaign=Buffer"); print_r($details); $details = URLExpander::getWebPageDetails('https://twitter.com/LamarrWilson/status/372582861548703745/photo/1'); print_r($details); $details = URLExpander::getWebPageDetails('http://justdelete.me/'); print_r($details); $details = URLExpander::getWebPageDetails('https://stellar.io/x/manage/membership'); print_r($details); $details = URLExpander::getWebPageDetails('https://www.wepay.com/donations/free-barrett-brown'); print_r($details); }
public function testGetWebPageDetails() { $details = URLExpander::getWebPageDetails('http://google.com'); print_r($details); $details = URLExpander::getWebPageDetails('http://smarterware.org'); print_r($details); $details = URLExpander::getWebPageDetails("http://www.slate.com/blogs/future_tense/2013/08/29/klout_deletes" . "_tweet_martin_luther_king_would_have_had_an_awesome_klout_score.html?utm_content=buffer9b03f&utm_source=bu" . "ffer&utm_medium=twitter&utm_campaign=Buffer"); print_r($details); $details = URLExpander::getWebPageDetails('https://twitter.com/LamarrWilson/status/372582861548703745/photo/1'); print_r($details); $details = URLExpander::getWebPageDetails('http://justdelete.me/'); print_r($details); $details = URLExpander::getWebPageDetails('https://stellar.io/x/manage/membership'); print_r($details); $details = URLExpander::getWebPageDetails('https://www.wepay.com/donations/free-barrett-brown'); print_r($details); $details = URLExpander::getWebPageDetails('https://www.facebook.com/photo.php?fbid=386789799914&set=' . 'at.386779984914.169691.68330209914.606837591&type=1&relevant_count=1'); print_r($details); $details = URLExpander::getWebPageDetails('https://www.facebook.com/photo.php?fbid=10152071367417592&' . 'set=a.10150271482787592.335919.606837591&type=1&relevant_count=1'); print_r($details); }
/** * Save expanded version of all unexpanded URLs to data store, as well as intermediary short links. */ public function expandOriginalURLs($flickr_api_key = null) { $links_to_expand = $this->link_dao->getLinksToExpand($this->link_limit); $this->logger->logUserInfo(count($links_to_expand) . " links to expand. Please wait. Working...", __METHOD__ . ',' . __LINE__); $total_expanded = 0; $total_errors = 0; $has_expanded_flickr_link = false; foreach ($links_to_expand as $index => $link) { if (Utils::validateURL($link->url)) { $endless_loop_prevention_counter = 0; $this->logger->logInfo("Expanding " . ($total_expanded + 1) . " of " . count($links_to_expand) . " (" . $link->url . ")", __METHOD__ . ',' . __LINE__); //make sure shortened short links--like t.co--get fully expanded $fully_expanded = false; $short_link = $link->url; while (!$fully_expanded) { //begin Flickr thumbnail processing if (isset($flickr_api_key) && substr($short_link, 0, strlen('http://flic.kr/')) == 'http://flic.kr/') { self::expandFlickrThumbnail($flickr_api_key, $short_link, $link->url); $has_expanded_flickr_link = true; $fully_expanded = true; } //end Flickr thumbnail processing $expanded_url = URLExpander::expandURL($short_link, $link->url, $index, count($links_to_expand), $this->link_dao, $this->logger); if ($expanded_url == $short_link || $expanded_url == '' || $endless_loop_prevention_counter > self::EXPANSION_CAP) { $fully_expanded = true; } else { try { $this->short_link_dao->insert($link->id, $short_link); } catch (DataExceedsColumnWidthException $e) { $this->logger->logError($short_link . " short link record exceeds column width, cannot save", __METHOD__ . ',' . __LINE__); $fully_expanded = true; } } if (strlen($expanded_url) < 256) { $short_link = $expanded_url; } else { $fully_expanded = true; } $endless_loop_prevention_counter++; } if (!$has_expanded_flickr_link) { if ($expanded_url != '') { $image_src = URLProcessor::getImageSource($expanded_url); $url_details = URLExpander::getWebPageDetails($expanded_url); try { $this->link_dao->saveExpandedUrl($link->url, $expanded_url, $url_details['title'], $image_src, $url_details['description']); $total_expanded = $total_expanded + 1; } catch (DataExceedsColumnWidthException $e) { $this->logger->logError($link->url . " record exceeds column width, cannot save", __METHOD__ . ',' . __LINE__); $this->link_dao->saveExpansionError($link->url, "URL exceeds column width"); $total_errors = $total_errors + 1; } } else { $this->logger->logError($link->url . " not a valid URL - relocates to nowhere", __METHOD__ . ',' . __LINE__); $this->link_dao->saveExpansionError($link->url, "Invalid URL - relocates to nowhere"); $total_errors = $total_errors + 1; } } } else { $this->logger->logError($link->url . " not a valid URL", __METHOD__ . ',' . __LINE__); $this->link_dao->saveExpansionError($link->url, "Invalid URL"); $total_errors = $total_errors + 1; } $has_expanded_flickr_link = false; } $this->logger->logUserSuccess($total_expanded . " URLs successfully expanded (" . $total_errors . " errors).", __METHOD__ . ',' . __LINE__); }