From 7d4815fcfafe4fdbd8e54cc8ce5fa2469bbfc2f9 Mon Sep 17 00:00:00 2001
From: Mohamed Jinas <j@live.mv>
Date: Sat, 15 Aug 2020 02:24:57 +0500
Subject: [PATCH] Sangu scraper boiler plate

---
 app/Services/Feeds/SanguFeed.php       | 23 +++++++++++
 app/Services/Scrapers/SanguScraper.php | 57 ++++++++++++++++++++++++++
 2 files changed, 80 insertions(+)
 create mode 100644 app/Services/Feeds/SanguFeed.php
 create mode 100644 app/Services/Scrapers/SanguScraper.php

diff --git a/app/Services/Feeds/SanguFeed.php b/app/Services/Feeds/SanguFeed.php
new file mode 100644
index 0000000..effbc44
--- /dev/null
+++ b/app/Services/Feeds/SanguFeed.php
@@ -0,0 +1,23 @@
+<?php
+namespace App\Services\Feeds;
+
+use Goutte\Client;
+
+class SanguFeed
+{
+    protected $client;
+
+    public function __construct()
+    {
+        $this->client = new Client();        
+    }
+    public function get()
+    {
+      
+        $crawler = $this->client->request('GET', "http://sangu.mv/");
+
+        return $crawler->filter('div[class*="large-5 medium-4 large-pull-7 medium-pull-8 small-12  columns"] .news-list .list-title a')
+                            ->extract(['_text', '_attr' => 'href']);             
+
+    }
+}
\ No newline at end of file
diff --git a/app/Services/Scrapers/SanguScraper.php b/app/Services/Scrapers/SanguScraper.php
new file mode 100644
index 0000000..2250323
--- /dev/null
+++ b/app/Services/Scrapers/SanguScraper.php
@@ -0,0 +1,57 @@
+<?php
+
+namespace App\Services\Scrapers;
+
+use Goutte\Client;
+use Illuminate\Support\Carbon;
+
+class SanguScraper
+{
+    protected $client;
+
+
+    protected $content;
+    protected $topics = [];
+
+    public function __construct()
+    {
+        $this->client = new Client;
+    }
+
+    public function extract($url)
+    {
+
+        $crawler = $this->client->request('GET', $url);
+
+        $title = $crawler->filter('.brk-title2')->first()->text();
+        $image = $crawler->filter('.main-img img')->first()->attr('src');
+        $author = $crawler->filter('.author span')->first()->text();
+
+        $crawler->filter('article p')->each(function ($node) {
+            $this->content[] = preg_replace("/[a-zA-Z]/","",$node->text());
+        });
+
+        $crawler->filter('.article-tags')->each(function ($node) {
+           
+            $this->topics[] = [
+                "name" => $node->text(),
+                "slug" => str_replace("https://mihaaru.com/", "", $node->attr('href'))
+            ];
+        });
+
+        //Remove all the alphabets from string
+        //preg_replace("/[a-zA-Z]/", "",$string);
+       return [
+            'source'    => 'Sangu',
+            'title'      => $title,
+            'og_title'   => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
+            'image'      => $image,
+            'content'    => $this->content,
+            'url'        => $url,
+            'date'       => "",
+            'guid'       => str_replace("https://mihaaru.com/news/","",$url),
+            'author'     => $author,
+            'topics'       => $this->topics
+        ];
+    }
+}