protected function standardize()
{
$CI =& get_instance();
$this->store_id = $CI->store_id;
$CI->load->model('crawl_data_m', 'Crawl');
$this->merchants = ensure_array($this->merchants);
$this->api_type = ensure_array($this->api_type);
$this->flagDates24 = $this->date_from === $this->date_to;
// get the last crawl data
$this->last_crawl = $CI->Crawl->last_crawl();
$this->crawl_range = $CI->Crawl->last_crawl_range();
if ($this->time_frame) {
// This report's dates are the last x many days
$tF = getTimeFrame($this->time_frame);
$this->date_from = $tF['date_from'];
$this->date_to = $tF['date_to'];
}
switch ($this->report_type) {
case 'violationoverview':
$this->time_frame = 24;
break;
case 'pricingoverview':
$this->time_frame = 24;
break;
case 'pricingviolation':
break;
case 'whois':
break;
}
$this->set_filename();
}
protected function _run_cron_job($api, $from = 'no')
{
//if ( ! in_array($from, $this->fromAr)) {
// echo "You are not allowed to run this cron";
// exit;
//}
// Count the active proxies
// Send a warning if 5 or less are active
$threshold = 10;
$proxies = getProxyIPS(TRUE, $threshold + 1);
if (count($proxies) <= $threshold) {
$this->email_alert('Low on proxies', 'The active proxy count has reached ' . $threshold . ' or less as of ' . date('Y-m-d H:i:s'));
}
$apis = ensure_array($api);
$storesData = $this->Store->get_stores(TRUE);
$currentDate = date('Y-m-d H:i:s');
$this->Crowl_model->stats = array();
// Cycle Through API crons
for ($j = 0, $n = count($apis); $j < $n; $j++) {
$this->Crowl_model->stats[$apis[$j]] = array('data_found' => 0, 'price_found' => 0);
$this->Crowl_model->load_sticky_api($apis[$j]);
$data = array('api_type' => $apis[$j], 'key' => generate_rand(32), 'run_from' => $from, 'start_datetime' => $currentDate);
try {
// Update cron log information
$lastLog = $this->getMaxCronLog($data['api_type']);
$lastRun = empty($lastLog) ? '0000-00-00 00:00:00' : $lastLog->datetime;
if (!empty($lastLog)) {
if ($lastLog->datetime === "0000-00-00 00:00:00") {
$update = array('datetime' => $currentDate, 'end_datetime' => $currentDate);
$this->db->where('datetime', '0000-00-00 00:00:00')->where('api_type', $data['api_type'])->update('cron_log', $update);
}
}
if ($lastRun == '0000-00-00 00:00:00' || $currentDate >= $lastRun) {
// No entry in cron log table
$insert_id = $this->create_cron_log($data);
$this->resetViolationCount();
// Crawl the products for each store
for ($i = 0, $z = count($storesData); $i < $z; $i++) {
$this->index($storesData[$i]->id, $data['api_type'], $insert_id);
}
$data_edit['end_datetime'] = $data_edit['datetime'] = date("Y-m-d H:i:s");
$this->update_cron_log($insert_id, $data_edit);
}
} catch (Exception $e) {
email_alertToTeam('Run Cron Job ' . ucwords($apis[$j]), $e->getMessage());
}
// Now that the crawl has finished,
// compare these crawl numbers with the last
// and send out notifications if needed.
$currentLog = $this->getMaxCronLog($data['api_type']);
$lastCount = $lastLog ? $lastLog->google_count : 0;
$currentCount = $currentLog->google_count;
$percentThresh = 25;
// send a notification if completed this percent of the last crawl
$smallThresh = 10;
// send a notification if complete less than this many products
$msg = '';
$percentDrop = 100 - $percentThresh;
if ($lastCount > 0 and $currentCount * 100 / $lastCount <= $percentThresh) {
$msg = ' crawl completed ' . $percentDrop . '% less products than its previous crawl.';
} elseif ($currentCount < $smallThresh) {
$msg = ' crawl completed less than ' . $smallThresh . ' products.';
}
if (!empty($msg)) {
$msg_txt = 'Crawl Warning: ' . $data['api_type'] . $msg;
$msg_html = 'Crawl Warning: ' . '<b>' . $data['api_type'] . '</b>' . $msg;
log_message('error', $msg_txt);
send_email($this->config->item('alerts'), 'TrackStreet Errors', 'TrackStreet Crawl Warning', $msg_html, $msg_txt);
}
}
$this->_checkNotFoundRatio();
// todo: should we not return to the main script here?
exit;
}
