<?php

namespace App\Services\Scrapers;

use Goutte\Client;
use Illuminate\Support\Carbon;

class VoiceScraper
{
    protected $client;

    protected $author;
    protected $content;
    protected $topics = [];

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

    public function extract($url)
    {

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

        $title = $crawler->filter('.content h1')->first()->text();


        $image = $crawler->filter('.image img')->first()->attr('src');

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

        $crawler->filter('.related-tags-holder a')->each(function ($node) {

            $this->topics[] = [
                "name" => $node->text(),
                "slug" => str_replace("/", "", $node->attr('href'))
            ];
        });

        if ($crawler->filter(".authorname a")->count() == 1) {
            $this->author = str_replace("- ", "", $crawler->filter('.authorname a')->first()->text());
        }


        //Remove all the alphabets from string
        //preg_replace("/[a-zA-Z]/", "",$string);
        return [
            'source'    => 'Voice',
            'title'      => $title,
            'og_title'   => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
            'image'      => $image,
            'content'    => $this->content,
            'url'        => $url,
            'date'       => Carbon::parse(str_replace("- ", "", $crawler->filter('.authordatecomment .date')->first()->text()))->format("Y-m-d H:i:s"),
            'guid'       => str_replace("https://voice.mv/", "", $url),
            'author'     => $this->author,
            'topics'       => $this->topics
        ];
    }
}