public function testAddRmOldFavMaintSearch()
{
$this->logger->logInfo("in testAddRmOldFavMaintSearch", __METHOD__ . ',' . __LINE__);
//set plugin cfg values
$namespace = OptionDAO::PLUGIN_OPTIONS . '-1';
$builder2 = FixtureBuilder::build('options', array('namespace' => $namespace, 'option_name' => 'favs_older_pages', 'option_value' => 1));
$builder3 = FixtureBuilder::build('options', array('namespace' => $namespace, 'option_name' => 'favs_cleanup_pages', 'option_value' => 3));
$instance_dao = DAOFactory::getDAO('TwitterInstanceDAO');
self::setUpInstanceUserAmygdala();
$this->api->available_api_calls_for_crawler = 3;
$this->api->to->setDataPath('webapp/plugins/twitter/tests/testdata/favs_tests/favs_stage3/');
$twitter_crawler = new TwitterCrawler($this->instance, $this->api);
$twitter_crawler->fetchInstanceUserInfo();
$retval = $twitter_crawler->cleanUpMissedFavsUnFavs();
$this->assertEqual($retval, true);
$this->assertEqual($this->instance->last_unfav_page_checked, 3);
// Save instance
if (isset($twitter_crawler->user)) {
$instance_dao->save($this->instance, $twitter_crawler->user->post_count, $this->logger);
}
$this->instance = $instance_dao->getByUsernameOnNetwork("amygdala", "twitter");
// check fav count
$this->assertEqual($this->instance->owner_favs_in_system, 40);
$this->logger->logInfo("in testAddRmOldFavMaintSearch, second traversal", __METHOD__ . ',' . __LINE__);
// now add an additional older fav , remove one, and traverse again
$this->api->available_api_calls_for_crawler = 3;
$this->instance->last_unfav_page_checked = 2;
$this->api->to->setDataPath('webapp/plugins/twitter/tests/testdata/favs_tests/favs_stage6/');
$twitter_crawler = new TwitterCrawler($this->instance, $this->api);
$twitter_crawler->fetchInstanceUserInfo();
$retval = $twitter_crawler->cleanUpMissedFavsUnFavs();
$this->assertEqual($retval, true);
// Save instance
if (isset($twitter_crawler->user)) {
$instance_dao->save($this->instance, $twitter_crawler->user->post_count, $this->logger);
}
$this->instance = $instance_dao->getByUsernameOnNetwork("amygdala", "twitter");
// check fav count- should have removed 2 and added 21...
// update: due to issue with TwitterAPI, not currently removing un-favs from database
// $this->assertEqual($this->instance->owner_favs_in_system, 59);
$this->assertEqual($this->instance->owner_favs_in_system, 61);
$builder2 = null;
$builder3 = null;
}
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 = $instance_dao->getAllActiveInstancesStalestFirstByNetwork('twitter');
foreach ($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 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);
}
$crawler = new TwitterCrawler($instance, $api);
$api->init();
if ($api->available_api_calls_for_crawler > 0) {
$instance_dao->updateLastRun($instance->id);
// No auth req'd
//$crawler->fetchInstanceUserInfo();
// No auth for public Twitter users
$crawler->fetchInstanceUserTweets();
if (!$noauth) {
// Auth req'd, for calling user only
$crawler->fetchInstanceUserMentions();
$crawler->fetchInstanceUserFriends();
$crawler->fetchInstanceFavorites();
$crawler->fetchInstanceUserFollowers();
$crawler->fetchRetweetsOfInstanceUser();
$crawler->cleanUpMissedFavsUnFavs();
}
$crawler->fetchStrayRepliedToTweets();
$crawler->fetchUnloadedFollowerDetails();
$crawler->fetchFriendTweetsAndFriends();
if ($noauth) {
// No auth req'd
$crawler->fetchSearchResults($instance->network_username);
}
$crawler->cleanUpFollows();
// Save instance
if (isset($crawler->user)) {
$instance_dao->save($instance, $crawler->user->post_count, $logger);
}
$logger->logUserSuccess("Finished collecting data for " . $instance->network_username . " on Twitter.", __METHOD__ . ',' . __LINE__);
}
}
}