Dhiyares intergration

2021-02-05 04:18:29 +05:00
parent 8ea594723f
commit e2bfb8de57
5 changed files with 211 additions and 0 deletions
--- a/app/Services/Scrapers/DhiyaresScraper.php
+++ b/app/Services/Scrapers/DhiyaresScraper.php
@@ -0,0 +1,66 @@
+<?php
+
+namespace App\Services\Scrapers;
+
+use Goutte\Client;
+use Illuminate\Support\Carbon;
+
+class DhiyaresScraper
+{
+    protected $client;
+
+    protected $title;
+    protected $content;
+    protected $topics = [];
+    protected $author = "unknown";
+
+    public function __construct()
+    {
+        $this->client = new Client;
+    }
+
+    public function extract($url, $date)
+    {
+
+        $crawler = $this->client->request('GET', $url);
+
+        $crawler->filter('.content p')->each(function ($node) {
+            $this->content[] = $node->text();
+        });
+
+        $crawler->filter('a[class*="bg-blue-50 border border-blue-50 font-normal text-blue-400 hover:text-white hover:bg-base p-1 px-4 mx-2 mb-4 rounded-full font-mv-bold text-base"]')->each(function ($node) {
+           if(!preg_match('/[^A-Za-z0-9-]/', basename($node->attr('href'))))
+           {
+            $this->topics[] = [
+                "name" => $node->text(),
+                "slug" => basename($node->attr('href'))
+            ];
+           }
+            
+        });
+
+
+        if ($crawler->filter('div[class*="text-base font-mv-bold"] a[class*="py-4"]')->count() == 1) {
+            $this->author = $crawler->filter('div[class*="text-base font-mv-bold"] a[class*="py-4"]')->first()->text();
+        }
+        //Remove all the alphabets from string
+        //preg_replace("/[a-zA-Z]/", "",$string);
+        return [
+            'source'    => 'Dhiyares',
+            'title'      => $crawler->filter('h1')->first()->text(),
+            'og_title'   => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
+            'image'      => $crawler->filter("figure img")->first()->attr('src'),
+            'content'    => $this->content,
+            'url'        => $url,
+            'date'       => Carbon::parse($date)->format("Y-m-d H:i:s"),
+            'guid'       =>basename($url),
+            'author'     => $this->author,
+            'topics'       => $this->topics ? : [
+                [
+                    "name" => "ވަކި މަޢުލޫއެއް ނޭންގެ",
+                    "slug" => "no-specific-topic"
+                ]
+            ]
+        ];
+    }
+}