private function extract_profiles()
{
$county = KsCounty::where('status', '=', '1')->orderBy('county_id')->first();
if ($county) {
if (App::runningInConsole()) {
fwrite(STDOUT, "=========== Extract Profiles : " . $county->name . PHP_EOL);
}
$curr_county_id = $county->county_id;
echo $curr_county_id . "\n";
$html = Storage::get('sexoffenders/_html/' . str_slug($this->state->state_name) . '/' . $curr_county_id . '.html');
$link_crawler = new Crawler($html);
try {
$table = $link_crawler->filter('#ctl00_ContentPlaceHolder1_OffenderSearchListView_divGeocodeTrue')->nextAll();
$links = $table->filter('a')->extract(array('href'));
foreach ($links as $link) {
$url = str_replace("./", $this->url_base, $link);
$hash = md5($url);
$profile = KsProfile::firstOrNew(array('hash' => $hash));
$profile->url = $url;
$profile->hash = $hash;
$profile->county_id = $curr_county_id;
$profile->save();
unset($profile);
}
$status = 3;
} catch (\InvalidArgumentException $e) {
$status = 4;
}
$county = KsCounty::firstOrNew(['county_id' => $curr_county_id]);
$county->status = $status;
$county->save();
$this->state->records_expected = KsProfile::count();
$this->state->save();
//
unset($html);
unset($link_crawler);
unset($county);
//
$this->extract_profiles();
} else {
if (App::runningInConsole()) {
fwrite(STDOUT, "=========== Extract DONE" . PHP_EOL);
}
$this->process_profiles();
}
}
