sveden-parser/app/app.php

<?php
namespace App;

use App\Library\ContingentManager;
use App\Library\DatabaseConfig;
use App\Library\Logger;
use GuzzleHttp\Exception\ClientException;
use GuzzleHttp\Exception\ConnectException;
use GuzzleHttp\Exception\RequestException;
use GuzzleHttp\Exception\ServerException;
use App\Library\ContingentParser;
use App\Library\Database;
use GuzzleHttp\Client;

$pathLogErrorHtml = 'error-html.log';
$pathLogErrorHttp = 'error-http-curl.log';

$dbOpendata = new Database(new DatabaseConfig('opendata'));
$dbNiimko = new Database(new DatabaseConfig('niimko'));

$sites = ContingentManager::getInstance()->getSites($dbNiimko);
$specializations = ContingentManager::getInstance()->getSpecializations($dbNiimko);
$orgs = ContingentManager::getInstance()->getOrgs($dbOpendata);
// print_r($sites);
// print_r($specializations);
// print_r($org);

// $errorSites = [];
// $filename = 'error-html.log';
// $array = file($filename);
// for ($i = 0; $i < count($array); $i++) {
//     $arr = explode(' ', $array[$i]);
//     if (!in_array($arr[2], $orgs)) {
//         $errorSites[] = $arr[2];
//     }
// }
// $filename = 'error-http.log';
// $array = file($filename);
// for ($i = 0; $i < count($array); $i++) {
//     $arr = explode(' ', $array[$i]);
//     if (!in_array($arr[2], $orgs)) {
//         $errorSites[] = $arr[2];
//     }
// }

$start = 999;

for ($i = $start; $i < count($sites); $i++) {
    // Нет URL сайта вуза
    if (empty($sites[$i]['site'])) {
        $message = $sites[$i]['org_id'] . ' ' . $sites[$i]['site'];
        Logger::log($pathLogErrorHttp, $message);
        continue;
    }
    // Уже в базе
    if (in_array($sites[$i]['org_id'], $orgs)) {
        continue;
    }
    // С ошибками разметки игнорируем
    // if (in_array($sites[$i]['org_id'], $errorSites)) {
    //     continue;
    // }
    try {
        $client = new Client([
            'allow_directs'   => true,
            'track_redirects' => true,
            'connect_timeout' => 300.0,
            'verify' => false,
            // 'http_errors' => false,
            'headers' => [
                'User-Agent' => 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 YaBrowser/24.6.0.0 Safari/537.36',
                'Content-Type' => 'text/html;charset=utf-8'
            ]
        ]);
        
        $url = $sites[$i]['site'];
        $url = ContingentManager::getInstance()->buildURL($url);
        print(($i+1).". Current url: $url\n");

        $response = $client->get($url);
        echo $response->getStatusCode() .PHP_EOL;

        $html = $response->getBody()->getContents();
        if (empty($html)) {
            $message = $sites[$i]['org_id'] . ' ' . $sites[$i]['site'];
            Logger::log($pathLogErrorHtml, $message);
            continue;
        }
        $parser = new ContingentParser($html, '//tr[@itemprop="eduChislen"]//');
        $contingent = $parser->getDataTable();

        // Добавляем поле spec_id по spec_code
        ContingentManager::getInstance()->addSpecId($contingent, $specializations);

        // Добавляем поле org_id
        ContingentManager::getInstance()->addOrgId($contingent, $sites[$i]['org_id']);
        print_r($contingent);
        if (empty($contingent)) {
            $message = $sites[$i]['org_id'] . ' ' . $sites[$i]['site'];
            Logger::log($pathLogErrorHtml, $message);
        } else {
            $set = ContingentManager::getInstance()->checkContingent($contingent);
            if ($set) {
                // Заносим в базу
                ContingentManager::getInstance()->insertContingent($dbOpendata, $contingent);
            } else {
                $message = $sites[$i]['org_id'] . ' ' . $sites[$i]['site'];
                Logger::log($pathLogErrorHtml, $message);
            }
            unset($contingent);
        }
    } catch (ClientException
        | RequestException
        | ConnectException
        | ServerException $e
    ) {
        $message = implode(' ', $sites[$i]) . "\t" . $e->getCode() . "\t" . $e->getMessage();
        Logger::log($pathLogErrorHttp, $message);
    } 
}