示例#1
0
 public function crawl()
 {
     $logger = Logger::getInstance();
     $config = Config::getInstance();
     $instance_dao = DAOFactory::getDAO('FacebookInstanceDAO');
     $owner_instance_dao = DAOFactory::getDAO('OwnerInstanceDAO');
     $owner_dao = DAOFactory::getDAO('OwnerDAO');
     $plugin_option_dao = DAOFactory::GetDAO('PluginOptionDAO');
     $options = $plugin_option_dao->getOptionsHash('facebook', true);
     //get cached
     $max_crawl_time = isset($options['max_crawl_time']) ? $options['max_crawl_time']->option_value : 20;
     //convert to seconds
     $max_crawl_time = $max_crawl_time * 60;
     $current_owner = $owner_dao->getByEmail(Session::getLoggedInUser());
     //crawl Facebook user profiles and pages
     $profiles = $instance_dao->getActiveInstancesStalestFirstForOwnerByNetworkNoAuthError($current_owner, 'facebook');
     $pages = $instance_dao->getActiveInstancesStalestFirstForOwnerByNetworkNoAuthError($current_owner, 'facebook page');
     $instances = array_merge($profiles, $pages);
     foreach ($instances as $instance) {
         $logger->setUsername(ucwords($instance->network) . ' | ' . $instance->network_username);
         $logger->logUserSuccess("Starting to collect data for " . $instance->network_username . "'s " . ucwords($instance->network), __METHOD__ . ',' . __LINE__);
         $tokens = $owner_instance_dao->getOAuthTokens($instance->id);
         $access_token = $tokens['oauth_access_token'];
         $instance_dao->updateLastRun($instance->id);
         $facebook_crawler = new FacebookCrawler($instance, $access_token, $max_crawl_time);
         $dashboard_module_cacher = new DashboardModuleCacher($instance);
         try {
             $facebook_crawler->fetchPostsAndReplies();
         } catch (APIOAuthException $e) {
             $logger->logUserError(get_class($e) . ": " . $e->getMessage(), __METHOD__ . ',' . __LINE__);
             //Don't send reauth email if it's app-level API rate limting
             //https://developers.facebook.com/docs/reference/ads-api/api-rate-limiting/#applimit
             if (strpos($e->getMessage(), 'Application request limit reached') === false && strpos($e->getMessage(), 'Please retry your request later.') === false) {
                 //The access token is invalid, save in owner_instances table
                 $owner_instance_dao->setAuthErrorByTokens($instance->id, $access_token, '', $e->getMessage());
                 //Send email alert
                 //Get owner by auth tokens first, then send to that person
                 $owner_email_to_notify = $owner_instance_dao->getOwnerEmailByInstanceTokens($instance->id, $access_token, '');
                 $email_attempt = $this->sendInvalidOAuthEmailAlert($owner_email_to_notify, $instance->network_username);
                 if ($email_attempt) {
                     $logger->logUserInfo('Sent reauth email to ' . $owner_email_to_notify, __METHOD__ . ',' . __LINE__);
                 } else {
                     $logger->logInfo('Didn\'t send reauth email to ' . $owner_email_to_notify, __METHOD__ . ',' . __LINE__);
                 }
             } else {
                 $logger->logInfo('Facebook API returned an error: ' . $e->getMessage() . ' Do nothing now and try again later', __METHOD__ . ',' . __LINE__);
             }
         } catch (Exception $e) {
             $logger->logUserError(get_class($e) . ": " . $e->getMessage(), __METHOD__ . ',' . __LINE__);
         }
         $dashboard_module_cacher->cacheDashboardModules();
         $instance_dao->save($facebook_crawler->instance, 0, $logger);
         Reporter::reportVersion($instance);
         $logger->logUserSuccess("Finished collecting data for " . $instance->network_username . "'s " . ucwords($instance->network), __METHOD__ . ',' . __LINE__);
     }
 }
 public function crawl()
 {
     $logger = Logger::getInstance();
     $config = Config::getInstance();
     $instance_dao = DAOFactory::getDAO('InstanceDAO');
     $owner_instance_dao = DAOFactory::getDAO('OwnerInstanceDAO');
     $owner_dao = DAOFactory::getDAO('OwnerDAO');
     $plugin_option_dao = DAOFactory::GetDAO('PluginOptionDAO');
     $options = $plugin_option_dao->getOptionsHash('facebook', true);
     //get cached
     $max_crawl_time = isset($options['max_crawl_time']) ? $options['max_crawl_time']->option_value : 20;
     //convert to seconds
     $max_crawl_time = $max_crawl_time * 60;
     $current_owner = $owner_dao->getByEmail(Session::getLoggedInUser());
     //crawl Facebook user profiles and pages
     $profiles = $instance_dao->getActiveInstancesStalestFirstForOwnerByNetworkNoAuthError($current_owner, 'facebook');
     $pages = $instance_dao->getActiveInstancesStalestFirstForOwnerByNetworkNoAuthError($current_owner, 'facebook page');
     $instances = array_merge($profiles, $pages);
     foreach ($instances as $instance) {
         $logger->setUsername(ucwords($instance->network) . ' | ' . $instance->network_username);
         $logger->logUserSuccess("Starting to collect data for " . $instance->network_username . "'s " . ucwords($instance->network), __METHOD__ . ',' . __LINE__);
         $tokens = $owner_instance_dao->getOAuthTokens($instance->id);
         $access_token = $tokens['oauth_access_token'];
         $instance_dao->updateLastRun($instance->id);
         $facebook_crawler = new FacebookCrawler($instance, $access_token, $max_crawl_time);
         $dashboard_module_cacher = new DashboardModuleCacher($instance);
         try {
             $facebook_crawler->fetchPostsAndReplies();
         } catch (APIOAuthException $e) {
             //The access token is invalid, save in owner_instances table
             $owner_instance_dao->setAuthError($current_owner->id, $instance->id, $e->getMessage());
             //Send email alert
             $this->sendInvalidOAuthEmailAlert($current_owner->email, $instance->network_username);
             $logger->logUserError('EXCEPTION: ' . $e->getMessage(), __METHOD__ . ',' . __LINE__);
         } catch (Exception $e) {
             $logger->logUserError('EXCEPTION: ' . $e->getMessage(), __METHOD__ . ',' . __LINE__);
         }
         $dashboard_module_cacher->cacheDashboardModules();
         $instance_dao->save($facebook_crawler->instance, 0, $logger);
         Reporter::reportVersion($instance);
         $logger->logUserSuccess("Finished collecting data for " . $instance->network_username . "'s " . ucwords($instance->network), __METHOD__ . ',' . __LINE__);
     }
 }
示例#3
0
 public function crawl()
 {
     $logger = Logger::getInstance();
     $config = Config::getInstance();
     $instance_dao = DAOFactory::getDAO('InstanceDAO');
     $owner_instance_dao = DAOFactory::getDAO('OwnerInstanceDAO');
     $owner_dao = DAOFactory::getDAO('OwnerDAO');
     $plugin_option_dao = DAOFactory::GetDAO('PluginOptionDAO');
     $options = $plugin_option_dao->getOptionsHash('facebook', true);
     //get cached
     $max_crawl_time = isset($options['max_crawl_time']) ? $options['max_crawl_time']->option_value : 20;
     //convert to seconds
     $max_crawl_time = $max_crawl_time * 60;
     $current_owner = $owner_dao->getByEmail(Session::getLoggedInUser());
     //crawl Facebook user profiles and pages
     $profiles = $instance_dao->getAllActiveInstancesStalestFirstByNetwork('facebook');
     $pages = $instance_dao->getAllActiveInstancesStalestFirstByNetwork('facebook page');
     $instances = array_merge($profiles, $pages);
     foreach ($instances as $instance) {
         if (!$owner_instance_dao->doesOwnerHaveAccessToInstance($current_owner, $instance)) {
             // Owner doesn't have access to this instance; let's not crawl it.
             continue;
         }
         $logger->setUsername(ucwords($instance->network) . ' | ' . $instance->network_username);
         $logger->logUserSuccess("Starting to collect data for " . $instance->network_username . "'s " . ucwords($instance->network), __METHOD__ . ',' . __LINE__);
         $tokens = $owner_instance_dao->getOAuthTokens($instance->id);
         $access_token = $tokens['oauth_access_token'];
         $instance_dao->updateLastRun($instance->id);
         $crawler = new FacebookCrawler($instance, $access_token, $max_crawl_time);
         try {
             $crawler->fetchPostsAndReplies();
         } catch (Exception $e) {
             $logger->logUserError('EXCEPTION: ' . $e->getMessage(), __METHOD__ . ',' . __LINE__);
         }
         $instance_dao->save($crawler->instance, 0, $logger);
         $logger->logUserSuccess("Finished collecting data for " . $instance->network_username . "'s " . ucwords($instance->network), __METHOD__ . ',' . __LINE__);
     }
 }
示例#4
0
 public function testPostReplyPaging()
 {
     $fbc = new FacebookCrawler($this->instance, 'fauxaccesstoken');
     $fbc->fetchPostsAndReplies('133954286636768', true);
     $post_dao = new PostMySQLDAO();
     $post = $post_dao->getPost('144568048938151', 'facebook page');
     $this->assertEqual($post->reply_count_cache, 70);
 }
示例#5
0
 public function testPaginatedPostLikes()
 {
     $fbc = new FacebookCrawler($this->profile5_instance, 'fauxaccesstoken', 10);
     $fbc->fetchPostsAndReplies('501771984', 'facebook');
     $post_dao = new PostMySQLDAO();
     $post = $post_dao->getPost('10151734003261985', 'facebook');
     $this->assertEqual($post->favlike_count_cache, 27);
 }
 public function testFetchPostsAndRepliesForPage()
 {
     $fbc = new FacebookCrawler($this->page1_instance, 'abc', 10);
     $fbc->fetchPostsAndReplies();
     $post_dao = new PostMySQLDAO();
     $post = $post_dao->getPost('10152714332426356', 'facebook page');
     $this->assertEqual($post->post_text, 'Looks like a very serene place to work:');
     $this->assertFalse($post->is_protected);
     $this->assertEqual($post->reply_count_cache, 4);
     //test link with image
     $this->assertEqual(sizeof($post->links), 1);
     $this->assertEqual($post->links[0]->url, 'http://lifehac.kr/Jg0mR3E');
     $this->assertEqual($post->links[0]->expanded_url, '');
     $this->assertEqual($post->links[0]->image_src, 'https://fbexternal-a.akamaihd.net/safe_image.php?d=AQDJyDKV3JDoU0-y&w=130&h=130&url=http%3A%2F%2' . 'Fi.kinja-img.com%2Fgawker-media%2Fimage%2Fupload%2Fs--G2ekOhHZ--%2Fgobnypf78nigkkl7cwpl.png&cfs=1');
     $this->assertEqual($post->links[0]->description, 'Today’s featured workspace looks immaculate, with its pristine white walls, simple white desk, and ' . 'wood accents. Those skylights help too.');
     //assert user network is set to Facebook, not Facebook Page
     $ud = new UserMySQLDAO();
     $user = $ud->getUserByName('Gregory Robert Dumas', 'facebook');
     $this->assertEqual($user->full_name, 'Gregory Robert Dumas');
     $this->assertEqual($user->network, 'facebook');
     $this->assertTrue($user->is_protected);
     $user = $ud->getUserByName('Matthew Fleisher', 'facebook page');
     $this->assertEqual($user, null);
     $fav_dao = new FavoritePostMySQLDAO();
     $favs = $fav_dao->getUsersWhoFavedPost('10152714332426356', 'facebook page');
     $this->assertEqual($favs[0]['user_name'], 'Peter-Sarah Crofton');
     $this->assertEqual($favs[0]['user_id'], '25913266');
     // Test Facebook paging by confirming post on second "page" was captured
     $post = $post_dao->getPost('10152712264921356', 'facebook page');
     $this->assertNotNull($post);
     $this->assertEqual($post->author_user_id, '7568536355');
 }
示例#7
0
 public function crawl()
 {
     $logger = Logger::getInstance();
     $config = Config::getInstance();
     $instance_dao = DAOFactory::getDAO('InstanceDAO');
     $owner_instance_dao = DAOFactory::getDAO('OwnerInstanceDAO');
     $owner_dao = DAOFactory::getDAO('OwnerDAO');
     $plugin_option_dao = DAOFactory::GetDAO('PluginOptionDAO');
     $options = $plugin_option_dao->getOptionsHash('facebook', true);
     //get cached
     $current_owner = $owner_dao->getByEmail(Session::getLoggedInUser());
     //crawl Facebook user profiles
     $instances = $instance_dao->getAllActiveInstancesStalestFirstByNetwork('facebook');
     foreach ($instances as $instance) {
         if (!$owner_instance_dao->doesOwnerHaveAccess($current_owner, $instance)) {
             // Owner doesn't have access to this instance; let's not crawl it.
             continue;
         }
         $logger->setUsername($instance->network_username);
         $logger->logUserSuccess("Starting to collect data for " . $instance->network_username . " on Facebook.", __METHOD__ . ',' . __LINE__);
         $tokens = $owner_instance_dao->getOAuthTokens($instance->id);
         $access_token = $tokens['oauth_access_token'];
         $instance_dao->updateLastRun($instance->id);
         $crawler = new FacebookCrawler($instance, $access_token);
         try {
             $crawler->fetchInstanceUserInfo();
             $crawler->fetchPostsAndReplies($instance->network_user_id, false);
         } catch (Exception $e) {
             $logger->logUserError('PROFILE EXCEPTION: ' . $e->getMessage(), __METHOD__ . ',' . __LINE__);
         }
         $instance_dao->save($crawler->instance, 0, $logger);
         $logger->logUserSuccess("Finished collecting data for " . $instance->network_username . " on Facebook.", __METHOD__ . ',' . __LINE__);
     }
     //crawl Facebook pages
     $instances = $instance_dao->getAllActiveInstancesStalestFirstByNetwork('facebook page');
     foreach ($instances as $instance) {
         $logger->setUsername($instance->network_username);
         $logger->logUserSuccess("Starting to collect data for " . $instance->network_username . "'s Facebook Page.", __METHOD__ . ',' . __LINE__);
         $tokens = $owner_instance_dao->getOAuthTokens($instance->id);
         $access_token = $tokens['oauth_access_token'];
         $instance_dao->updateLastRun($instance->id);
         $crawler = new FacebookCrawler($instance, $access_token);
         try {
             $crawler->fetchPostsAndReplies($instance->network_user_id, true);
         } catch (Exception $e) {
             $logger->logUserError('PAGE EXCEPTION: ' . $e->getMessage(), __METHOD__ . ',' . __LINE__);
         }
         $instance_dao->save($crawler->instance, 0, $logger);
         $logger->logUserSuccess("Finished collecting data for " . $instance->network_username . "'s Facebook Page.", __METHOD__ . ',' . __LINE__);
     }
 }