Psm Support

2020-09-22 23:09:24 +05:00
parent 71284d8b01
commit 78f7493649
4 changed files with 164 additions and 0 deletions
--- a/app/Services/Scrapers/PsmScraper.php
+++ b/app/Services/Scrapers/PsmScraper.php
@@ -0,0 +1,59 @@
+<?php
+
+namespace App\Services\Scrapers;
+
+use Goutte\Client;
+
+class PsmScraper
+{
+    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);
+
+        $crawler->filter('h1')->each(function ($node) {
+            $this->title = $node->text();
+        });
+
+        $crawler->filter('.content > p')->each(function ($node) {
+            $this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text());
+        });
+
+
+        if($crawler->filter('div[class*="flex items-center  border-grey-light border-solid"] p a')->count() == 1)
+        {
+            $this->author = $crawler->filter('div[class*="flex items-center  border-grey-light border-solid"] p a')->first()->text();
+        }
+        //Remove all the alphabets from string
+        //preg_replace("/[a-zA-Z]/", "",$string);
+        return [
+            'source'    => 'PSM',
+            'title'      => $this->title,
+            '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'       => $date,
+            'guid'       => str_replace("https://www.psmnews.mv/", "", $url),
+            'author'     => $this->author,
+            'topics'       => [
+                [
+                    "name" => "ވަކި މަޢުލޫއެއް ނޭންގެ",
+                    "slug" => "no-specific-topic"
+                ]
+            ]
+        ];
+    }
+}