Devil Killer Is Here MiNi Shell
MiNi SheLL

Current Path : /home/vmanager/www/console/controllers/
Linux 9dbcd5f6333d 5.15.0-102-generic #112-Ubuntu SMP Tue Mar 5 16:50:32 UTC 2024 x86_64
Current File : /home/vmanager/www/console/controllers/ImportNewCompaniesFromPktController.php
<?php

namespace console\controllers;

use \Yii;
use \yii\console\Controller;
use \yii\helpers\Console;
use yii\console\ExitCode;
use common\models\CompaniesToImport;

class ImportNewCompaniesFromPktController extends Controller
{
    use \console\components\traits\ImportNewCompaniesFunctions;
    
    public function actionGetSearchResults()
    {
        $modelCategory = $this->getUnprocessedCategory();
        if(empty($modelCategory)) {
            $this->stdout('Brak kategorii do przetworzenia.');
            return ExitCode::OK;
        }
        
        $phrase = str_replace(' ', '-', mb_strtolower($modelCategory->phrase));
        
        $url = 'https://www.pkt.pl/szukaj/'.$phrase.'/'.($modelCategory->last_processed_page + 1);
        
        $curlResult = $this->makeCurlExec($url, []);        
        if($curlResult === null) {
            $this->stderr('Błąd podczas ładowania danych HTML.');
            return ExitCode::IOERR;
        }
        
        $matches = [];        
        if(
            (preg_match('|<script type="application/ld\+json">(.+)</script>|is', $curlResult, $matches) !== 1) 
            || (mb_strpos($curlResult, '<div class="box-fall-back-messages"> <i class="icon--info"></i> <p>Wyniki wyszukiwania zawierają niektóre z podanych słów.</p>') !== false)
        ) {
            $this->stdout('Brak wyników.');
            $modelCategory->or_completed = 1;
            $modelCategory->save();  
            return ExitCode::OK;
        }      
        
        $items = json_decode($matches[1], true);

        $orExistAnyCompany = false;
        foreach($items as $item) {
            if($item['@type'] !== 'LocalBusiness') {
                continue;
            }
            $orExistAnyCompany = true;
            $id = mb_substr($item['@id'], mb_strrpos($item['@id'], '-') + 1);
            
            $model = CompaniesToImport::find()->where(['_id' => 'pkt_'.$id])->one();
            if($model === null) {
                $model = new CompaniesToImport();
                $model->_id = 'pkt_'.$id;
                $model->source_name = $item['name'];
                $model->ids_categories = [$modelCategory->_id];
                $model->searched_by_phrases = [$modelCategory->phrase];
                $model->or_visited = 0;
                $model->url = $item['@id'];
                $model->save();
                
            } elseif(!in_array($modelCategory->_id, $model->ids_categories)) {
                $idsCategories = $model->ids_categories;
                $searchedByPhrases = $model->searched_by_phrases;
                $idsCategories[] = $modelCategory->_id;
                $searchedByPhrases[] = $modelCategory->phrase;
                
                $model->ids_categories = $idsCategories;
                $model->searched_by_phrases = $searchedByPhrases;
                $model->save();
            }
        }

        if(!$orExistAnyCompany) {
            $modelCategory->or_completed = 1;
        } else {
            $modelCategory->last_processed_page += 1;
        }        
        $modelCategory->save();        
        
        return ExitCode::OK;
    }
    
    public function actionGetCompanyData()
    {
        $modelCompany = $this->getUnprocessedCompany();
        if(empty($modelCompany)) {
            $this->stdout('Brak firm do przetworzenia.');
            return ExitCode::OK;
        }
        
        $this->getDataFromSource($modelCompany);        
        $this->getDataFromBIR($modelCompany);
        
        $flag = $modelCompany->save();
        if(!$flag) {
            var_dump($modelCompany->toArray());
            var_dump($modelCompany->getErrors());
            return ExitCode::DATAERR;
        }
        
        return ExitCode::OK;
    }
    
    protected function getDataFromSource($modelCompany)
    {
        //@todo proteza
//        $modelCompany->url = 'https://www.pkt.pl/firma/tormet-sp-z-o-o-sp-k-1568024';
//        $modelCompany->url = 'https://www.pkt.pl/firma/raf-tech-rafal-hoffmann-120223262';
        
        $curlResult = $this->makeCurlExec($modelCompany->url, []);
        if($curlResult === null) {
            $this->stderr('Błąd podczas ładowania danych HTML.');
            return ExitCode::IOERR;
        }
        
        $matches = [];
        $orMatch = preg_match('|<script type="application/ld\+json">(.+)</script>|is', $curlResult, $matches);        
        if($orMatch !== 1) {
            $this->stdout('Brak wyników.');
            return ExitCode::OK;
        }
        
        $items = json_decode($matches[1], true);
        $item = null;
        foreach($items as $value) {
            if($value['@type'] === 'LocalBusiness') {
                $item = $value;
                break;
            }
        }        
            
        if(!empty($item['telephone'])) {
            $modelCompany->source_phone = preg_replace('|[^0-9]+|i', '', $item['telephone']);
        }
        if(!empty($item['url'])) {
            $modelCompany->source_www = trim($item['url']);
        }
        if(!empty($item['image']) && $item['image'] !== 'https://www.pkt.pl/images/ikona-firmy-pkt.png') {
            $modelCompany->image_url = trim($item['image']);
        }

        $modelCompany->source_postcode = trim($item['address']['postalCode']);
        $modelCompany->source_locality = trim($item['address']['addressLocality']);
        $modelCompany->source_address = trim($item['address']['streetAddress']);
        $modelCompany->source_region = trim(str_replace('woj.', '', $item['address']['addressRegion']));

        $this->loadDOM($curlResult);

        $elements = $this->xpath->query("//html/body/div[contains(@class, \"box-detail container cf\")]/div[contains(@class, \"company-row\")]");
        $companyNodeContainer = $elements[0];
        list($modelCompany->lng, $modelCompany->lat) = array_map('doubleval', explode(',', $companyNodeContainer->getAttribute('data-geo')));            

        $emailNode = $this->xpath->query("div[contains(@class, \"company-map--info\")]/div/div[contains(@class, \"cf\")]/div[contains(@class, \"call--email\")]/a/span", $companyNodeContainer);
        if(!empty($emailNode[0]) && !empty($emailNode[0]->getAttribute('title'))) {
            $modelCompany->source_email = trim($emailNode[0]->getAttribute('title'));
        }

        $additionalDataNodes = $this->xpath->query("div/div[contains(@class, \"aggregated-data\")]/div[contains(@class, \"aggregated-data__content\")]/div[contains(@class, \"aggregated-data__phone\")]/div", $this->domDocument->getElementById('main')); 
        $registrationDataNode = null;
        foreach($additionalDataNodes as $node) {
            for($i = 1; $i <= 2; $i++) {
                if(!empty($node->childNodes[$i]) && $node->childNodes[$i]->nodeValue === 'Dane rejestrowe:') {
                    $registrationDataNode = $node->childNodes[$i]->nextSibling->nextSibling;
                    break;
                }
            }
        }

        $nationalCourtRegisterNumberNodes = $this->xpath->query("div/div[contains(@class, \"full-company-data-box__krs\")]/span", $registrationDataNode);
        if($nationalCourtRegisterNumberNodes->length > 0) {
            $modelCompany->source_national_court_register_number = trim($nationalCourtRegisterNumberNodes[0]->nodeValue);
        }

        $taxIdNodes = $this->xpath->query("div/div[contains(@class, \"full-company-data-box__nip\")]/span", $registrationDataNode);
        if($taxIdNodes->length > 0) {
            $modelCompany->source_tax_id = trim($taxIdNodes[0]->nodeValue);
        }

        $businessRegistrationNumberNodes = $this->xpath->query("div/div[contains(@class, \"full-company-data-box__regon\")]/span", $registrationDataNode);
        if($businessRegistrationNumberNodes->length > 0) {
            $modelCompany->source_business_registration_number = trim($businessRegistrationNumberNodes[0]->nodeValue);
        }            

        $facebookDataNodes = $this->xpath->query("div/div[contains(@class, \"aggregated-data\")]/div[contains(@class, \"aggregated-data__content\")]/div[contains(@class, \"aggregated-data__social-link\")]/div[2]/div/div[contains(@class, \"social--facebook\")]/a", $this->domDocument->getElementById('main')); 
        if($facebookDataNodes->length > 0) {
            $modelCompany->source_fb = trim($facebookDataNodes[0]->getAttribute('href'));
        } 

        $businessSectors = [];

        $businessSectorDataNodes = $this->xpath->query("div/div[contains(@class, \"aggregated-data\")]/div[contains(@class, \"aggregated-data__content\")]/div[contains(@class, \"aggregated-data__category\")][last()]/div/div/a", $this->domDocument->getElementById('main')); 
        if($businessSectorDataNodes->length > 0) {
            $businessSectors[] = trim($businessSectorDataNodes[0]->nodeValue);
        }
        $businessSectorDataNodes = $this->xpath->query("div/div[contains(@class, \"aggregated-data\")]/div[contains(@class, \"aggregated-data__content\")]/div[contains(@class, \"aggregated-data__category\")][last()]/div/div/p/a", $this->domDocument->getElementById('main')); 
        if($businessSectorDataNodes->length > 0) {
            $businessSectors[] = trim($businessSectorDataNodes[0]->nodeValue);
        }

        if(!empty($businessSectors)) {
            $modelCompany->source_business_sectors = $businessSectors;
        }

        $modelCompany->or_visited = 1;
        
    }
    
    protected function getDataFromBIR($modelCompany)
    {         
        $sessionKey = $this->BIRLogin();
        
        $responseShort = $this->BIRGetShortData($modelCompany, $sessionKey);
        if($responseShort === false) {
            return;
        }
        
        if(!empty($responseShort['dane']['Regon'])) {
            $modelCompany->bir_business_registration_number = trim($responseShort['dane']['Regon']);
        }
        
        list($responseFullData, $parsedFullData) = $this->BIRGetFullData($modelCompany, $responseShort['dane']['Typ'], $responseShort['dane']['Regon'], $sessionKey);

        if(!empty($parsedFullData['phone'])) {
            $modelCompany->bir_phone = preg_replace('|[^0-9]+|i', '', $parsedFullData['phone']);
        }
        if(!empty($parsedFullData['email'])) {
            $modelCompany->bir_email = trim($parsedFullData['email']);
        }
        if(!empty($parsedFullData['www'])) {
            $modelCompany->bir_www = trim($parsedFullData['www']);
        }
        
        $modelCompany->bir_locality = $parsedFullData['locality'];
        $modelCompany->bir_street = $parsedFullData['street'];
        $modelCompany->bir_building_number = $parsedFullData['building_number'];
        $modelCompany->bir_apartment_number = $parsedFullData['apartment_number'];
        $modelCompany->bir_postcode = $parsedFullData['postcode'];
        $modelCompany->bir_region = $parsedFullData['region'];
    }  
    
    
    
    
    
    //        $doc = new \DOMDocument();
//        libxml_use_internal_errors(true);
//        $this->domDocument->loadHTML($curlResult);
//        libxml_clear_errors();
//
//        $xpath = new \DOMXpath($doc);
//        $elements = $this->xpath->query("/html/body/div/div/main/div[contains(@class, \"company\")]");
//        if(empty($elements)) {
//            $this->stdout('Brak wyników.');
//            return ExitCode::OK;
//        }
//        
//        $companies = [];
//        
//        foreach($elements as $element) {
//            $company = [];
//            $company['id'] = (int)$element->getAttribute('data-company-id');
//            list($company['lng'], $company['lat']) = explode(",", $element->getAttribute('data-geo'));
//            $emailNode = $this->xpath->query("div[contains(@class, \"cf\")]/div[contains(@class, \"call--email\")]/a/span", $element);
//            if(!empty($emailNode[0]) && !empty($emailNode[0]->getAttribute('title'))) {
//                $company['email'] = $emailNode[0]->getAttribute('title');
//            }
//            var_dump($company);
//            exit;
//        }
//        var_dump($elements);
//        $document = new HtmlDocument($curlResult);
//        $element = $document->getElementById('main');
//        var_dump($element->find("div.company-row")); exit;
}
Creat By MiNi SheLL
Email: jattceo@gmail.com