/**
* @param InputInterface $input
* @param OutputInterface $output
* @return void|int
*/
protected function execute(InputInterface $input, OutputInterface $output)
{
list($from, $to) = $this->getDateRangeToAttribute($input);
$this->visitorGeolocator = $this->createGeolocator($output, $input);
$this->percentStep = $this->getPercentStep($input);
$this->amountOfVisits = $this->dao->countVisitsWithDatesLimit($from, $to);
$output->writeln(sprintf('Re-attribution for date range: %s to %s. %d visits to process with provider "%s".', $from, $to, $this->amountOfVisits, $this->visitorGeolocator->getProvider()->getId()));
$this->timer = new Timer();
$this->processSpecifiedLogsInChunks($output, $from, $to, $input->getOption(self::SEGMENT_LIMIT_OPTION));
$output->writeln("Completed. <comment>" . $this->timer->__toString() . "</comment>");
return 0;
}
/**
* Purges old data from the following tables:
* - log_visit
* - log_link_visit_action
* - log_conversion
* - log_conversion_item
* - log_action
*
* @param int $deleteLogsOlderThan The number of days after which log entires are considered old.
* Visits and related data whose age is greater than this number
* will be purged.
*/
public function purgeData($deleteLogsOlderThan)
{
$dateUpperLimit = Date::factory("today")->subDay($deleteLogsOlderThan);
$this->logDeleter->deleteVisitsFor($start = null, $dateUpperLimit->getDatetime());
$logTables = self::getDeleteTableLogTables();
// delete unused actions from the log_action table (but only if we can lock tables)
if (Db::isLockPrivilegeGranted()) {
$this->rawLogDao->deleteUnusedLogActions();
} else {
$logMessage = get_class($this) . ": LOCK TABLES privilege not granted; skipping unused actions purge";
Log::warning($logMessage);
}
// optimize table overhead after deletion
Db::optimizeTables($logTables);
}
/**
* Deletes visits within the specified date range and belonging to the specified site (if any). Visits are
* deleted in chunks, so only `$iterationStep` visits are deleted at a time.
*
* @param string|null $startDatetime A datetime string. Visits that occur at this time or after are deleted. If not supplied,
* visits from the beginning of time are deleted.
* @param string|null $endDatetime A datetime string. Visits that occur before this time are deleted. If not supplied,
* visits from the end of time are deleted.
* @param int|null $idSite The site to delete visits from.
* @param int $iterationStep The number of visits to delete at a single time.
* @param callable $afterChunkDeleted Callback executed after every chunk of visits are deleted.
* @return int The number of visits deleted.
*/
public function deleteVisitsFor($startDatetime, $endDatetime, $idSite = null, $iterationStep = 1000, $afterChunkDeleted = null)
{
$fields = array('idvisit');
$conditions = array();
if (!empty($startDatetime)) {
$conditions[] = array('visit_last_action_time', '>=', $startDatetime);
}
if (!empty($endDatetime)) {
$conditions[] = array('visit_last_action_time', '<', $endDatetime);
}
if (!empty($idSite)) {
$conditions[] = array('idsite', '=', $idSite);
}
$logsDeleted = 0;
$logPurger = $this;
$this->rawLogDao->forAllLogs('log_visit', $fields, $conditions, $iterationStep, function ($logs) use($logPurger, &$logsDeleted, $afterChunkDeleted) {
$ids = array_map(function ($row) {
return reset($row);
}, $logs);
$logsDeleted += $logPurger->deleteVisits($ids);
if (!empty($afterChunkDeleted)) {
$afterChunkDeleted($logsDeleted);
}
});
return $logsDeleted;
}
protected function processSpecifiedLogsInChunks(OutputInterface $output, $from, $to, $segmentLimit)
{
$visitFieldsToSelect = array_merge(array('idvisit', 'location_ip'), array_keys(VisitorGeolocator::$logVisitFieldsToUpdate));
$lastId = 0;
do {
$logs = $this->dao->getVisitsWithDatesLimit($from, $to, $visitFieldsToSelect, $lastId, $segmentLimit);
if (!empty($logs)) {
$lastId = $logs[count($logs) - 1]['idvisit'];
$this->reattributeVisitLogs($output, $logs);
}
} while (count($logs) == $segmentLimit);
}
protected function insertActionsToKeep($maxIds, $olderThan = true, $insertIntoTempIterationStep = 100000)
{
parent::insertActionsToKeep($maxIds, $olderThan, 2);
// we use 2 to force iterations during tests
// allow code to be executed after data is inserted. for concurrency testing purposes.
if ($olderThan && $this->insertActionsOlderThanCallback) {
$callback = $this->insertActionsOlderThanCallback;
$callback();
} else {
if ($this->insertActionsNewerThanCallback) {
$callback = $this->insertActionsNewerThanCallback;
$callback();
}
}
}
/**
* Re-geolocate visits within a date range for a specified site (if any).
*
* @param string $from A datetime string to treat as the lower bound. Visits newer than this date are processed.
* @param string $to A datetime string to treat as the upper bound. Visits older than this date are processed.
* @param int|null $idSite If supplied, only visits for this site are re-attributed.
* @param int $iterationStep The number of visits to re-attribute at the same time.
* @param callable|null $onLogProcessed If supplied, this callback is called after every row is processed.
* The processed visit and the updated values are passed to the callback.
*/
public function reattributeVisitLogs($from, $to, $idSite = null, $iterationStep = 1000, $onLogProcessed = null)
{
$visitFieldsToSelect = array_merge(array('idvisit', 'location_ip'), array_keys(VisitorGeolocator::$logVisitFieldsToUpdate));
$conditions = array(array('visit_last_action_time', '>=', $from), array('visit_last_action_time', '<', $to));
if (!empty($idSite)) {
$conditions[] = array('idsite', '=', $idSite);
}
$self = $this;
$this->dao->forAllLogs('log_visit', $visitFieldsToSelect, $conditions, $iterationStep, function ($logs) use($self, $onLogProcessed) {
foreach ($logs as $row) {
$updatedValues = $self->attributeExistingVisit($row);
if (!empty($onLogProcessed)) {
$onLogProcessed($row, $updatedValues);
}
}
});
}
/**
* Geolcates an existing visit and then updates it if it's current attributes are different than
* what was geolocated. Also updates all conversions of a visit.
*
* **This method should NOT be used from within the tracker.**
*
* @param array $visit The visit information. Must contain an `"idvisit"` element and `"location_ip"` element.
* @param bool $useClassCache
* @return array|null The visit properties that were updated in the DB mapped to the updated values. If null,
* required information was missing from `$visit`.
*/
public function attributeExistingVisit($visit, $useClassCache = true)
{
if (empty($visit['idvisit'])) {
$this->logger->debug('Empty idvisit field. Skipping re-attribution..');
return null;
}
$idVisit = $visit['idvisit'];
if (empty($visit['location_ip'])) {
$this->logger->debug('Empty location_ip field for idvisit = %s. Skipping re-attribution.', array('idvisit' => $idVisit));
return null;
}
$ip = IPUtils::binaryToStringIP($visit['location_ip']);
$location = $this->getLocation(array('ip' => $ip), $useClassCache);
$valuesToUpdate = $this->getVisitFieldsToUpdate($visit, $location);
if (!empty($valuesToUpdate)) {
$this->logger->debug('Updating visit with idvisit = {idVisit} (IP = {ip}). Changes: {changes}', array('idVisit' => $idVisit, 'ip' => $ip, 'changes' => $valuesToUpdate));
$this->dao->updateVisits($valuesToUpdate, $idVisit);
$this->dao->updateConversions($valuesToUpdate, $idVisit);
} else {
$this->logger->debug('Nothing to update for idvisit = %s (IP = {ip}). Existing location info is same as geolocated.', array('idVisit' => $idVisit, 'ip' => $ip));
}
return $valuesToUpdate;
}
/**
* Detects whether a site had visits since midnight in the websites timezone
*
* @return bool
*/
private function hadWebsiteTrafficSinceMidnightInTimezone($idSite)
{
$timezone = Site::getTimezoneFor($idSite);
$nowInTimezone = Date::factory('now', $timezone);
$midnightInTimezone = $nowInTimezone->setTime('00:00:00');
$secondsSinceMidnight = $nowInTimezone->getTimestamp() - $midnightInTimezone->getTimestamp();
$secondsSinceLastArchive = $this->getSecondsSinceLastArchive();
if ($secondsSinceLastArchive < $secondsSinceMidnight) {
$secondsSinceMidnight = $secondsSinceLastArchive;
}
$from = Date::now()->subSeconds($secondsSinceMidnight)->getDatetime();
$to = Date::now()->addHour(1)->getDatetime();
$dao = new RawLogDao();
$hasVisits = $dao->hasSiteVisitsBetweenTimeframe($from, $to, $idSite);
if ($hasVisits) {
$this->logger->info("{$idSite} has visits between {$from} and {$to}");
} else {
$this->logger->info("{$idSite} has no visits between {$from} and {$to}");
}
return $hasVisits;
}
/**
* @dataProvider getVisitsInTimeFrameData
*/
public function test_hasSiteVisitsInTimeframe_shouldDetectWhetherThereAreVisitsInCertainTimeframe($from, $to, $idSite, $expectedHasVisits)
{
Fixture::getTracker($this->idSite, '2015-01-25 05:35:27')->doTrackPageView('/test');
$hasVisits = $this->dao->hasSiteVisitsBetweenTimeframe($from, $to, $idSite);
$this->assertSame($expectedHasVisits, $hasVisits);
}
/**
* Detects whether a site had visits since midnight in the websites timezone
*
* @param $idSite
* @return bool
*/
private function hadWebsiteTrafficSinceMidnightInTimezone($idSite)
{
$timezone = Site::getTimezoneFor($idSite);
$nowInTimezone = Date::factory('now', $timezone);
$midnightInTimezone = $nowInTimezone->setTime('00:00:00');
$secondsSinceMidnight = $nowInTimezone->getTimestamp() - $midnightInTimezone->getTimestamp();
$secondsSinceLastArchive = $this->getSecondsSinceLastArchive();
if ($secondsSinceLastArchive < $secondsSinceMidnight) {
$secondsBackToLookForVisits = $secondsSinceLastArchive;
$sinceInfo = "(since the last successful archiving)";
} else {
$secondsBackToLookForVisits = $secondsSinceMidnight;
$sinceInfo = "(since midnight)";
}
$from = Date::now()->subSeconds($secondsBackToLookForVisits)->getDatetime();
$to = Date::now()->addHour(1)->getDatetime();
$dao = new RawLogDao();
$hasVisits = $dao->hasSiteVisitsBetweenTimeframe($from, $to, $idSite);
if ($hasVisits) {
$this->logger->info("- tracking data found for website id {$idSite} since {$from} UTC {$sinceInfo}");
} else {
$this->logger->info("- no new tracking data for website id {$idSite} since {$from} UTC {$sinceInfo}");
}
return $hasVisits;
}
