public static function boot()
{
parent::boot();
static::creating(function ($workerunit) {
// dd($workerunit);
// Inherit documentType, domain and format
if (empty($workerunit->documentType)) {
$j = Job::where('_id', $workerunit->job_id)->first();
$workerunit->documentType = $j->documentType;
}
// transform answer into annotation vector to prepare for the CrowdTruth metrics
$workerunit->annotationVector = $workerunit->createAnnotationVector();
// Activity if not exists
if (empty($workerunit->activity_id)) {
try {
$activity = new Activity();
$activity->label = "Workerunit is saved.";
$activity->softwareAgent_id = $workerunit->softwareAgent_id;
$activity->save();
$workerunit->activity_id = $activity->_id;
Log::debug("Saving workerunit {$workerunit->_id} with activity {$workerunit->activity_id}.");
} catch (Exception $e) {
if ($activity) {
$activity->forceDelete();
}
//if($workerunit) $workerunit->forceDelete();
throw new Exception('Error saving activity for workerunit.');
}
}
});
}
public function fire($job, $data)
{
// Metrics
set_time_limit(600);
$workerunit = unserialize($data['workerunit']);
$spammer = Job::where('_id', $workerunit->job_id)->whereIn('metrics.spammers.list', [$workerunit->crowdAgent_id])->exists();
$workerunit->spam = $spammer;
$workerunit->save();
\Log::debug("Updated Workerunit {$workerunit->_id}.");
$job->delete();
}
/**
* @param $data = array(unitids)
*/
public function fire($job, $data)
{
foreach ($data as $id) {
set_time_limit(600);
$unit = Entity::id($id)->first();
$batches = count(Batch::whereIn('parents', [$unit->_id])->get()->toArray());
$workerunit = array('count' => 0, 'spam' => 0, 'nonSpam' => 0);
$workerlist = $workersspam = $workersnonspam = $joblist = array();
foreach (Workerunit::where('unit_id', $unit->_id)->get() as $a) {
$joblist[] = $a->job_id;
$workerlist[] = $a->crowdAgent_id;
if ($a->spam) {
$workerunit['spam']++;
$workersspam[] = $a->crowdAgent_id;
} else {
$workerunit['nonSpam']++;
$workersnonspam[] = $a->crowdAgent_id;
}
}
$workerunit['count'] = $workerunit['spam'] + $workerunit['nonSpam'];
$workers['count'] = count(array_unique($workerlist));
$workers['spam'] = count(array_unique($workersspam));
$workers['nonSpam'] = count(array_unique($workersnonspam));
$workers['potentialSpam'] = count(array_intersect($workersspam, $workersnonspam));
$platformField = array();
$platformField['cf'] = count(Entity::where('unit_id', $unit->_id)->where('softwareAgent_id', 'cf')->get()->toArray());
$platformField['amt'] = count(Entity::where('unit_id', $unit->_id)->where('softwareAgent_id', 'amt')->get()->toArray());
//filtered
$filteredField = array();
$filteredField['job_ids'] = array_flatten(Job::where('metrics.filteredUnits.list', 'all', array($unit['_id']))->get(['_id'])->toArray());
$filteredField['count'] = count($filteredField['job_ids']);
$derivatives = Entity::whereIn('parents', array($unit->_id))->lists('_id');
$children["count"] = count($derivatives);
$children["list"] = $derivatives;
$unit->cache = ["jobs" => 1, "workers" => $workers, "softwareAgent" => $platformField, "workerunits" => $workerunit, "filtered" => $filteredField, "batches" => $batches, "children" => $children];
$avg_clarity = Entity::where('metrics.units.withoutSpam.' . $unit->_id, 'exists', 'true')->avg('metrics.units.withoutSpam.' . $unit->id . '.avg.max_relation_Cos');
if (!isset($avg_clarity)) {
$avg_clarity = 0;
}
$unit->avg_clarity = $avg_clarity;
$unit->update();
\Log::debug("Updated unit {$unit->_id}.");
}
$job->delete();
// the Queue job...
}
public function updateStats2()
{
// take all the jobs for that worker
// TODO: change to Job::
if ($crowdAgentJobs = Job::where('metrics.workers.withFilter.' . $this->_id, 'exists', true)->get(['_id'])) {
//if there is at least one job with that worker
if (count($crowdAgentJobs->toArray()) > 0) {
$spam = $nonspam = $totalNoOfWorkerunits = 0;
foreach ($this->workerunits as $a) {
$totalNoOfWorkerunits++;
if ($a->spam) {
$spam++;
} else {
$nonspam++;
}
$types[] = $a->type;
$jobids[] = $a->job_id;
$unitids[] = $a->unit_id;
}
// $this->WorkerunitStats = array('count'=>$total['count'], 'spam'=>$spam, 'nonspam'=>$nonspam);
$distinctWorkerunitTypes = array_unique($types);
// These actually are the Workerunit types
$workerParticipatedIn = count(array_unique($unitids));
dd($workerParticipatedIn);
$cache["workerunits"] = ["count" => $totalNoOfWorkerunits, "spam" => $spam, "nonspam" => $nonspam];
// take all distinct batches
$distinctBatchIds = Entity::whereIn('_id', array_flatten($crowdAgentJobs->toArray()))->distinct('batch_id')->get(['_id']);
/*
foreach($distinctBatchIds as $distinctBatchId) {
$batchParents = array_flatten(Entity::where('_id', '=', $distinctBatchId[0])->lists('parents'));
//print_r($batchParents[0]);
$batchParentsType = Entity::where('_id', '=', $batchParents[0])->distinct('documentType')->get(['documentType']);
//print_r(array_flatten($batchParentsType->toArray())[0]);
if(isset($cache["mediaTypes"]["types"][array_flatten($batchParentsType->toArray())[0]])) {
$cache["mediaTypes"]["types"][array_flatten($batchParentsType->toArray())[0]] = $cache["mediaTypes"][array_flatten($batchParentsType->toArray())[0]] + 1;
}
else {
$cache["mediaTypes"]["types"] = [];
$cache["mediaTypes"]["types"][array_flatten($batchParentsType->toArray())[0]] = 1;
}
}
$cache["mediaTypes"]["distinct"] = sizeof(array_keys($cache["mediaTypes"]["types"]));
*/
if (count($distinctWorkerunitTypes) > 0) {
$cache["jobTypes"] = ["distinct" => count($distinctWorkerunitTypes), "count" => count(array_unique($jobids)), "types" => []];
foreach ($distinctWorkerunitTypes as $distinctJobType) {
$distinctJobTypeCount = Job::whereIn('_id', array_flatten($crowdAgentJobs->toArray()))->type($distinctJobType)->count();
$distinctJobTemplateTypes = Job::whereIn('_id', array_flatten($crowdAgentJobs->toArray()))->type($distinctJobType)->distinct('template')->get()->toArray();
$countJobTemplateTypes = Job::whereIn('_id', array_flatten($crowdAgentJobs->toArray()))->type($distinctJobType)->count();
//$cache["jobTypes"]["types"][$distinctJobType[0]] = [];
$cache["jobTypes"]["types"][$distinctJobType]['distinct'] = count($distinctJobTemplateTypes);
$cache["jobTypes"]["types"][$distinctJobType]['count'] = count($countJobTemplateTypes);
$cache["jobTypes"]["types"][$distinctJobType]["templates"] = [];
foreach ($distinctJobTemplateTypes as $distinctJobTemplateType) {
$distinctJobTemplateAndCount = Job::whereIn('_id', array_flatten($crowdAgentJobs->toArray()))->where('template', $distinctJobTemplateType)->count();
$cache["jobTypes"]["types"][$distinctJobType]["templates"][$distinctJobTemplateType[0]] = $distinctJobTemplateAndCount;
}
}
}
$jobsAsSpammer = Job::whereIn('_id', array_flatten($crowdAgentJobs->toArray()))->whereIn('metrics.spammers.list', [$this->_id])->lists('platformJobId');
$cache["spammer"]["count"] = count($jobsAsSpammer);
$cache["spammer"]["jobs"] = array_flatten($jobsAsSpammer);
$this->cache = $cache;
$this->save();
} else {
$this->save();
}
}
}
