Support for Avas mv

2020-08-20 15:43:47 +05:00
parent 110223d5be
commit f4d411f81e
5 changed files with 236 additions and 0 deletions
--- a/app/Services/Scrapers/AvasScraper.php
+++ b/app/Services/Scrapers/AvasScraper.php
@@ -0,0 +1,71 @@
+<?php
+namespace App\Services\Scrapers;
+
+use Goutte\Client;
+use Illuminate\Support\Carbon;
+
+class AvasScraper
+{
+    protected $client;
+
+    protected $content;
+    protected $author;
+    protected $topics = [];
+    
+    public function __construct()
+    {
+        $this->client = new Client;
+    }
+
+    public function extract($url)
+    {
+
+        $crawler = $this->client->request('GET', $url);
+
+        $title = $crawler->filter('h1')->first()->text();
+
+
+
+        $image = $crawler->filter('figure img')->first()->attr('src');
+
+        $crawler->filter('.post_content p')->each(function ($node) {
+            $this->content[] = preg_replace("/[a-zA-Z]/","",$node->text());
+        });
+
+        $crawler->filter('div[class*="border-t border-grey-light border-dotted mt-7 py-3"] a')->each(function ($node) {
+           
+            //Removing the show more tags button
+            if($node->text() == "+")
+            {
+                return;
+            }
+            $this->topics[] = [
+                "name" => $node->text(),
+                "slug" => str_replace("/", "", $node->attr('href'))
+            ];
+        });
+
+      
+
+        if($crawler->filter('div[class*="font-waheed text-grey ml-3 pl-3 text-lg border-l border-grey border-dotted"] a')->count() == 1)
+        {
+            $this->author = $crawler->filter('div[class*="font-waheed text-grey ml-3 pl-3 text-lg border-l border-grey border-dotted"] a')->first()->text();
+        }
+
+
+        //Remove all the alphabets from string
+        //preg_replace("/[a-zA-Z]/", "",$string);
+       return [
+            'source'    => 'Avas',
+            'title'      => $title,
+            'og_title'   => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
+            'image'      => $image,
+            'content'    => $this->content,
+            'url'        => $url,
+            'date'       =>  Carbon::parse($crawler->filter('timeago')->first()->attr('datetime'))->format("Y-m-d H:i:s"),
+            'guid'       => str_replace("https://avas.mv/","",$url),
+            'author'     => $this->author,
+            'topics'       => $this->topics
+        ];
+    }
+}