Roman_applicant-site/app/Services/PageScrapper.php

<?php

namespace App\Services;

use DiDom\Document;

class PageScrapper
{
    private string $url;
    private string $contentMarker;
    private Document $document;

    public function __construct($url, $contentMarker)
    {
        $this->url = $url;
        $this->contentMarker = $contentMarker;
        $this->document = new Document($this->url, true);
    }

    public function getFullHTML()
    {
        return $this->document;
    }

    public function printHTML()
    {
        $rez = $this->document;
        $content = $rez->first($this->contentMarker)->html();
        return $content;
    }

    public function normalizePath()
    {
        $rez = $this->document;
        $content = $rez->first($this->contentMarker);
        $links = $content->find('a[href]');
        $srclinks = $content->find('img[src]');

        $html0 = $content->html();

        preg_match_all('/<a href="(.*)">/isU', $html0, $arr);
        foreach ($arr[1] as $el) {
            $html0 = str_replace($el, urldecode($el), $html0);
        }
        preg_match_all('/<img src="(.*)">/isU', $html0, $arr2);
        foreach ($arr2[1] as $el) {
            $html0 = str_replace($el, urldecode($el), $html0);
        }

        foreach ($links as $k => $link) {
            $href = $link->attr('href');
            if (!str_contains($link->attr('href'), "https://")) {
                $unchanged = $link->attr('href');
                $changed = $link->href = 'https://mkgtu.ru' . $href;
                $html0 = str_replace(urldecode($unchanged), $changed, $html0);
            }
        }


        foreach ($srclinks as $k => $srclink) {
            $src = $srclink->attr('src');
            if (!str_contains($srclink->attr('src'), "https://")) {
                $unchanged = $srclink->attr('src');
                $changed = $srclink->src = 'https://mkgtu.ru' . $src;
                $html0 = str_replace(urldecode($unchanged), $changed, $html0);
            }
        }

        return $html0;
    }
}