<?php

namespace App\Services\Scrapers;

use Goutte\Client;

class SunScraper
{
    protected $client;

    protected $title;
    protected $content;
    protected $topics = [];
    protected $author = "unknown";

    public function __construct()
    {
        $this->client = new Client;
    }

    public function extract($url)
    {

        $crawler = $this->client->request('GET', $url);

        $crawler->filter('.component-article-title h1')->each(function ($node) {
            $this->title = $node->text();
        });

        $crawler->filter('.component-article-content p')->each(function ($node) {
            $this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text());
        });

        $crawler->filter('.component-article-tag li a')->each(function ($node) {

            $this->topics[] = [
                "name" => $node->first()->text(),
                "slug" => str_replace("https://sun.mv/", "", $node->first()->attr('href'))
            ];
        });

        if($crawler->filter(".author .name")->count() == 1)
        {
            $this->author = $crawler->filter(".author .name")->first()->text();
        }
        //Remove all the alphabets from string
        //preg_replace("/[a-zA-Z]/", "",$string);
        return [
            'source'    => 'Sun',
            'title'      => $this->title,
            'og_title'   => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
            'image'      => $crawler->filter(".component-article-featured img")->first()->attr('src'),
            'content'    => $this->content,
            'url'        => $url,
            'date'       => $crawler->filter(".author .time")->first()->text(),
            'guid'       => str_replace("https://sun.mv/", "", $url),
            'author'     => $this->author,
            'topics'       => collect($this->topics)->unique()->values()->toArray()
        ];
    }
}