From 1e1f84aedee62f301df3f6d0a1b9ce26b167b473 Mon Sep 17 00:00:00 2001
From: Mohamed Jinas <j@live.mv>
Date: Tue, 11 Aug 2020 04:03:59 +0500
Subject: [PATCH] Create SunScraper.php

---
 app/Services/Scrapers/SunScraper.php | 57 ++++++++++++++++++++++++++++
 1 file changed, 57 insertions(+)
 create mode 100644 app/Services/Scrapers/SunScraper.php

diff --git a/app/Services/Scrapers/SunScraper.php b/app/Services/Scrapers/SunScraper.php
new file mode 100644
index 0000000..541dfb1
--- /dev/null
+++ b/app/Services/Scrapers/SunScraper.php
@@ -0,0 +1,57 @@
+<?php
+
+namespace App\Services\Scrapers;
+
+use Goutte\Client;
+
+class SunScraper
+{
+    protected $client;
+
+    protected $title;
+    protected $content;
+    protected $topics = [];
+    protected $author;
+
+    public function __construct()
+    {
+        $this->client = new Client;
+    }
+
+    public function extract($url)
+    {
+
+        $crawler = $this->client->request('GET', $url);
+
+        $crawler->filter('h1')->each(function ($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/", "", $node->first()->attr('href'))
+            ];
+        });
+
+        //Remove all the alphabets from string
+        //preg_replace("/[a-zA-Z]/", "",$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'     => $crawler->filter(".author .name")->first()->text(),
+            'topics'       => collect($this->topics)->unique()->values()->toArray()
+        ];
+    }
+}