public function crawl() { $config = Config::getInstance(); $logger = Logger::getInstance(); $instance_dao = DAOFactory::getDAO('TwitterInstanceDAO'); $owner_instance_dao = DAOFactory::getDAO('OwnerInstanceDAO'); $owner_dao = DAOFactory::getDAO('OwnerDAO'); $instance_hashtag_dao = DAOFactory::getDAO('InstanceHashtagDAO'); // get oauth values $plugin_option_dao = DAOFactory::GetDAO('PluginOptionDAO'); $options = $plugin_option_dao->getOptionsHash('twitter', true); $current_owner = $owner_dao->getByEmail(Session::getLoggedInUser()); $instances = $instance_dao->getActiveInstancesStalestFirstForOwnerByNetworkNoAuthError($current_owner, 'twitter'); foreach ($instances as $instance) { $logger->setUsername($instance->network_username); $logger->logUserSuccess("Starting to collect data for " . $instance->network_username . " on Twitter.", __METHOD__ . ',' . __LINE__); $tokens = $owner_instance_dao->getOAuthTokens($instance->id); $num_twitter_errors = isset($options['num_twitter_errors']) ? $options['num_twitter_errors']->option_value : null; $dashboard_module_cacher = new DashboardModuleCacher($instance); try { if (isset($tokens['oauth_access_token']) && $tokens['oauth_access_token'] != '' && isset($tokens['oauth_access_token_secret']) && $tokens['oauth_access_token_secret'] != '') { $archive_limit = isset($options['archive_limit']->option_value) ? $options['archive_limit']->option_value : 3200; $api = new CrawlerTwitterAPIAccessorOAuth($tokens['oauth_access_token'], $tokens['oauth_access_token_secret'], $options['oauth_consumer_key']->option_value, $options['oauth_consumer_secret']->option_value, $archive_limit, $num_twitter_errors); $twitter_crawler = new TwitterCrawler($instance, $api); $instance_dao->updateLastRun($instance->id); $twitter_crawler->fetchInstanceUserTweets(); $twitter_crawler->fetchInstanceUserMentions(); $twitter_crawler->fetchInstanceUserFriends(); $twitter_crawler->fetchInstanceUserFollowers(); $twitter_crawler->fetchInstanceUserGroups(); $twitter_crawler->fetchRetweetsOfInstanceUser(); $twitter_crawler->fetchInstanceUserFavorites(); $twitter_crawler->updateStaleGroupMemberships(); $twitter_crawler->fetchStrayRepliedToTweets(); $twitter_crawler->fetchUserFriendsByIDs(); $twitter_crawler->fetchUnloadedFriendDetails(); $twitter_crawler->fetchUnloadedFollowerDetails(); $twitter_crawler->cleanUpFollows(); $twitter_crawler->updateFriendsProfiles(); //Retrieve search results for saved keyword/hashtags $instances_hashtags = $instance_hashtag_dao->getByInstance($instance->id); foreach ($instances_hashtags as $instance_hashtag) { $twitter_crawler->fetchInstanceHashtagTweets($instance_hashtag); } } else { throw new Exception('Missing Twitter OAuth tokens.'); } } catch (Exception $e) { $logger->logUserError(get_class($e) . " while crawling " . $instance->network_username . " on Twitter: " . $e->getMessage(), __METHOD__ . ',' . __LINE__); } $dashboard_module_cacher->cacheDashboardModules(); // Save instance if (isset($twitter_crawler->user)) { $instance_dao->save($instance, $twitter_crawler->user->post_count, $logger); } Reporter::reportVersion($instance); $logger->logUserSuccess("Finished collecting data for " . $instance->network_username . " on Twitter.", __METHOD__ . ',' . __LINE__); } }
public function testFetchInstanceUserGroupsBudget() { self::setUpInstanceUserAnilDash(); // set up crawl limit budget $crawl_limit = array('fetchInstanceUserGroups' => array('count' => 2, 'remaining' => 0)); $this->api->setCallerLimits($crawl_limit); $twitter_crawler = new TwitterCrawler($this->instance, $this->api); $twitter_crawler->fetchInstanceUserGroups(); $group_dao = DAOFactory::getDAO('GroupDAO'); $this->assertFalse($group_dao->isGroupInStorage($group = '1234566', 'twitter'), 'group does not exist'); $group_member_dao = DAOFactory::getDAO('GroupMemberDAO'); $this->assertFalse($group_member_dao->isGroupMemberInStorage($user = '******', $group = '1234566', 'twitter'), 'group member does not exists'); }
public function crawl() { $config = Config::getInstance(); $logger = Logger::getInstance(); $instance_dao = DAOFactory::getDAO('TwitterInstanceDAO'); $owner_instance_dao = DAOFactory::getDAO('OwnerInstanceDAO'); $owner_dao = DAOFactory::getDAO('OwnerDAO'); // get oauth values $plugin_option_dao = DAOFactory::GetDAO('PluginOptionDAO'); $options = $plugin_option_dao->getOptionsHash('twitter', true); $current_owner = $owner_dao->getByEmail(Session::getLoggedInUser()); $instances = $instance_dao->getActiveInstancesStalestFirstForOwnerByNetworkNoAuthError($current_owner, 'twitter'); foreach ($instances as $instance) { $logger->setUsername($instance->network_username); $logger->logUserSuccess("Starting to collect data for " . $instance->network_username . " on Twitter.", __METHOD__ . ',' . __LINE__); $tokens = $owner_instance_dao->getOAuthTokens($instance->id); $noauth = true; $num_twitter_errors = isset($options['num_twitter_errors']) ? $options['num_twitter_errors']->option_value : null; $max_api_calls_per_crawl = isset($options['max_api_calls_per_crawl']) ? $options['max_api_calls_per_crawl']->option_value : 350; if (isset($tokens['oauth_access_token']) && $tokens['oauth_access_token'] != '' && isset($tokens['oauth_access_token_secret']) && $tokens['oauth_access_token_secret'] != '') { $noauth = false; } $api_calls_to_leave_unmade_per_minute = isset($options['api_calls_to_leave_unmade_per_minute']) ? $options['api_calls_to_leave_unmade_per_minute']->option_value : 2.0; if ($noauth) { $api = new CrawlerTwitterAPIAccessorOAuth('NOAUTH', 'NOAUTH', $options['oauth_consumer_key']->option_value, $options['oauth_consumer_secret']->option_value, $api_calls_to_leave_unmade_per_minute, $options['archive_limit']->option_value, $num_twitter_errors, $max_api_calls_per_crawl); } else { $api = new CrawlerTwitterAPIAccessorOAuth($tokens['oauth_access_token'], $tokens['oauth_access_token_secret'], $options['oauth_consumer_key']->option_value, $options['oauth_consumer_secret']->option_value, $api_calls_to_leave_unmade_per_minute, $options['archive_limit']->option_value, $num_twitter_errors, $max_api_calls_per_crawl); } $twitter_crawler = new TwitterCrawler($instance, $api); $dashboard_module_cacher = new DashboardModuleCacher($instance); $api->init(); // budget our twitter calls $call_limits = $this->budgetCrawlLimits($api->available_api_calls_for_crawler, $noauth); $api->setCallerLimits($call_limits); if ($api->available_api_calls_for_crawler > 0) { $instance_dao->updateLastRun($instance->id); // No auth for public Twitter users $twitter_crawler->fetchInstanceUserTweets(); if (!$noauth) { // Auth req'd, for calling user only $twitter_crawler->fetchInstanceUserMentions(); $twitter_crawler->fetchInstanceUserFriends(); $twitter_crawler->fetchInstanceFavorites(); $twitter_crawler->fetchInstanceUserFollowers(); $twitter_crawler->fetchInstanceUserGroups(); $twitter_crawler->fetchRetweetsOfInstanceUser(); $twitter_crawler->cleanUpMissedFavsUnFavs(); $twitter_crawler->updateStaleGroupMemberships(); } $twitter_crawler->fetchStrayRepliedToTweets(); $twitter_crawler->fetchUnloadedFollowerDetails(); $twitter_crawler->cleanUpFollows(); $twitter_crawler->fetchFriendTweetsAndFriends(); if ($noauth) { // No auth req'd $twitter_crawler->fetchSearchResults($instance->network_username); } $dashboard_module_cacher->cacheDashboardModules(); // Save instance if (isset($twitter_crawler->user)) { $instance_dao->save($instance, $twitter_crawler->user->post_count, $logger); } $logger->logUserSuccess("Finished collecting data for " . $instance->network_username . " on Twitter.", __METHOD__ . ',' . __LINE__); } } }
public function testFetchInstanceUserGroups() { $this->debug(__METHOD__); self::setUpInstanceUserAnilDash(); $twitter_crawler = new TwitterCrawler($this->instance, $this->api); $twitter_crawler->api->to->setDataPathFolder('testoftwittercrawler/anildash/'); $twitter_crawler->fetchInstanceUserGroups(); $group_dao = DAOFactory::getDAO('GroupDAO'); $this->assertTrue($group_dao->isGroupInStorage($group = '79854285', 'twitter')); $group_member_dao = DAOFactory::getDAO('GroupMemberDAO'); $this->assertTrue($group_member_dao->isGroupMemberInStorage($user = '******', $group = '79854285', 'twitter')); $this->assertFalse($group_member_dao->isGroupMemberInStorage($user = '******', $group = '79854285', 'twitter')); $count_history_dao = DAOFactory::getDAO('CountHistoryDAO'); $history = $count_history_dao->getHistory($user = '******', 'twitter', 'DAYS', null, 'group_memberships'); $this->assertEqual(count($history['history']), 1); }