Vnews support

2020-10-03 03:41:11 +05:00
parent cf4e6267b6
commit af2965f7ba
4 changed files with 162 additions and 0 deletions
--- a/app/Services/Scrapers/VnewsScraper.php
+++ b/app/Services/Scrapers/VnewsScraper.php
@@ -0,0 +1,58 @@
+<?php
+
+namespace App\Services\Scrapers;
+
+use Goutte\Client;
+use Illuminate\Support\Carbon;
+
+class VnewsScraper
+{
+    protected $client;
+
+    protected $title;
+    protected $content;
+    protected $author = "unknown";
+    protected $topics;
+
+    public function __construct()
+    {
+        $this->client = new Client;
+    }
+
+    public function extract($url, $date)
+    {
+
+        $crawler = $this->client->request('GET', $url);
+
+        $crawler->filter('.news-content > p')->each(function ($node) {
+            $this->content[] = $node->text();
+        });
+
+        if ($crawler->filter('.author')->count() > 0) {
+            $this->author = $crawler->filter('.author')->first()->text();
+        }
+
+        $crawler->filter('.news-tags a')->each(function ($node) {
+            $this->topics[] = [
+                "name" => $node->text(),
+                "slug" => str_replace("#", "", str_replace("https://www.vnews.mv/", "", $node->attr('href')))
+            ];
+        });
+
+
+        //Remove all the alphabets from string
+        //preg_replace("/[a-zA-Z]/", "",$string);
+        return [
+            'source'    => 'VNews',
+            'title'      => $crawler->filter('h1')->first()->text(),
+            'og_title'   => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
+            'image'      => $crawler->filter(".news-image img")->first()->attr('src'),
+            'content'    => $this->content,
+            'url'        => $url,
+            'date'       => Carbon::parse($date)->format("Y-m-d H:i:s"),
+            'guid'       => str_replace("https://www.vnews.mv/", "", $url),
+            'author'     => $this->author,
+            'topics'       =>  $this->topics
+        ];
+    }
+}