105 lines
3.3 KiB
PHP
105 lines
3.3 KiB
PHP
<?php
|
||
namespace ContingentParser;
|
||
|
||
use ContingentParser\Database\DatabaseFacade;
|
||
use ContingentParser\Http\HttpClientFacade;
|
||
use ContingentParser\Http\UrlBuilder;
|
||
use ContingentParser\Logger\HtmlLogger;
|
||
use ContingentParser\Parser\ContingentFacade;
|
||
|
||
class Facade
|
||
{
|
||
private DatabaseFacade $databaseFacade;
|
||
private HttpClientFacade $httpClientFacade;
|
||
private ContingentFacade $contingentFacade;
|
||
private UrlBuilder $urlBuilder;
|
||
private HtmlLogger $htmlLogger;
|
||
/**
|
||
* Конструктор
|
||
*/
|
||
public function __construct()
|
||
{
|
||
$this->databaseFacade = new DatabaseFacade();
|
||
$this->httpClientFacade = new HttpClientFacade();
|
||
$this->urlBuilder = new UrlBuilder();
|
||
$this->contingentFacade = new ContingentFacade();
|
||
$this->htmlLogger = new HtmlLogger('log/html.log');
|
||
}
|
||
/**
|
||
* Получить массив сайтов
|
||
* @param array $params
|
||
* Массив сайтов, у которых нужны обновиленные URL
|
||
* @return array
|
||
*/
|
||
public function getSites(array $params = []) : array
|
||
{
|
||
if (empty($params)) {
|
||
return $this->databaseFacade->getSitesFromNiimko();
|
||
} else {
|
||
return $this->databaseFacade->getSitesFromMiccedu($params);
|
||
}
|
||
}
|
||
/**
|
||
* Cобирает из микроразметки данные таблицы
|
||
* "Информация о численности обучающихся" в разделе "Образование"
|
||
* @param array $site
|
||
* Сайт содержащий id организации и URL
|
||
* @return void
|
||
*/
|
||
public function collectDataFromContingent(array $site) : void
|
||
{
|
||
list('org_id' => $orgId, 'site' => $url) = $site;
|
||
// Нет URL сайта вуза
|
||
if (empty($site)) {
|
||
// $httpLogger->log($orgId);
|
||
return;
|
||
}
|
||
// Уже в базе
|
||
if (in_array($orgId, $this->databaseFacade->universities())) {
|
||
return;
|
||
}
|
||
$url = $this->urlBuilder->build($url);
|
||
print(implode(' ', $site). "\033[0m" . PHP_EOL);
|
||
$html = $this->httpClientFacade->processEducationContingentSites(
|
||
$url,
|
||
$site
|
||
);
|
||
|
||
$contingent = $this->contingentFacade->getContingent(
|
||
$html,
|
||
$this->databaseFacade->specialties(),
|
||
$orgId
|
||
);
|
||
|
||
if (empty($contingent)) {
|
||
print("\033[91mNo result\033[0m\n");
|
||
$this->htmlLogger->log("$orgId $url");
|
||
} else {
|
||
if ($this->contingentFacade->isValidContingent($contingent)) {
|
||
// Заносим в базу
|
||
print_r($contingent);
|
||
// $this->databaseFacade->insertContingent($contingent);
|
||
} else {
|
||
$this->htmlLogger->log("$orgId $url");
|
||
print("\033[91mNo result\033[0m\n");
|
||
}
|
||
}
|
||
|
||
unset($contingent);
|
||
print(PHP_EOL);
|
||
}
|
||
|
||
public function getExclusionSites(string $path) : array
|
||
{
|
||
$logs = file($path);
|
||
$result = [];
|
||
foreach ($logs as $log) {
|
||
$data = explode(' ', $log);
|
||
$result[] = [
|
||
'org_id' => $data[2],
|
||
'site' => $data[3] ? $data[3] : ''
|
||
];
|
||
}
|
||
return $result;
|
||
}
|
||
} |