The Press intergration

app/Console/Commands/ScrapeThePressCommand.php

@@ -0,0 +1,72 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\Source;
+use Illuminate\Console\Command;
+use App\Services\ThePressService;
+use App\Topic;
+
+class ScrapeThePressCommand extends Command
+{
+    /**
+     * The name and signature of the console command.
+     *
+     * @var string
+     */
+    protected $signature = 'scrape:thepress';
+
+    /**
+     * The console command description.
+     *
+     * @var string
+     */
+    protected $description = 'Scrape The press';
+
+    /**
+     * Create a new command instance.
+     *
+     * @return void
+     */
+    public function __construct()
+    {
+        parent::__construct();
+    }
+
+    /**
+     * Execute the console command.
+     *
+     * @return int
+     */
+    public function handle()
+    {
+        $source = Source::where('slug', 'thepress')->first();
+
+        $articles = (new ThePressService)->scrape();
+
+        foreach ($articles as $article) {
+
+            // Attach the relationship between source and article and return the curren article instance
+            $articleModel = $source->articles()->firstOrCreate(["guid" => $article["guid"]],
+                [
+                "title" => $article["title"],
+                "url" => $article["url"],
+                "author" => $article["author"],
+                "featured_image" => $article["image"],
+                "body" => $article["content"],
+                "published_date" => $article["date"],
+                "meta" => [
+                    "title" => $article["og_title"]
+                ]
+
+            ]);
+
+            collect($article["topics"])->each(function($topic) use ($articleModel) {
+                $topicModel = Topic::firstOrCreate(["slug" => $topic["slug"]],["name" => $topic["name"]]);
+
+                $topicModel->articles()->syncWithoutDetaching($articleModel);
+            });
+
+        }
+    }
+}

app/Console/Kernel.php

@@ -28,6 +28,7 @@ class Kernel extends ConsoleKernel
         $schedule->command('scrape:mihaaru')->everyFiveMinutes();
         $schedule->command('scrape:sun')->everyFiveMinutes();
         $schedule->command('scrape:thiladhun')->everyFiveMinutes();
+        $schedule->command('scrape:thepress')->everyFiveMinutes();
     }
 
     /**

app/Services/Scrapers/ThePressScraper.php

@@ -0,0 +1,62 @@
+<?php
+namespace App\Services\Scrapers;
+
+use Goutte\Client;
+
+class ThePressScraper
+{
+    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('.article-image img')->first()->attr('src');
+
+        $crawler->filter('article p')->each(function ($node) {
+            $this->content[] = preg_replace("/[a-zA-Z]/","",$node->text());
+        });
+
+        $crawler->filter('.article-tags a')->each(function ($node) {
+           
+            $this->topics[] = [
+                "name" => $node->text(),
+                "slug" => str_replace("https://thepress.mv/", "", $node->attr('href'))
+            ];
+        });
+
+        if($crawler->filter(".author-details strong")->count() == 1)
+        {
+            $this->author = $crawler->filter('.author-details strong')->first()->text();
+        }
+
+        //Remove all the alphabets from string
+        //preg_replace("/[a-zA-Z]/", "",$string);
+       return [
+            'source'    => 'The Press',
+            'title'      => $title,
+            'og_title'   => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
+            'image'      => $image,
+            'content'    => $this->content,
+            'url'        => $url,
+            'date'       => $crawler->filter('.article-header .datetime')->first()->text(),
+            'guid'       => str_replace("https://thepress.mv/","",$url),
+            'author'     => $this->author,
+            'topics'       => $this->topics
+        ];
+    }
+}

app/Services/ThePressService.php

@@ -0,0 +1,30 @@
+<?php
+
+namespace App\Services;
+
+use App\Services\Scrapers\ThePressScraper;
+use Illuminate\Support\Str;
+
+class ThePressService extends Client
+{
+    /**
+     * Scrap all the rss articles from Press
+     *
+     * @return array
+     */
+    public function scrape(): array
+    {
+        //Return only the rss that contains "news" keyboard in its url
+        $articles = $this->get("https://thepress.mv/rss")["channel"]["item"];
+
+
+        $articlesitems = [];
+        //Looping through the articles and scraping and while scraping it creates a new instance of the scraper.
+        foreach ($articles as $article) {
+            $link = $article['link'];
+            $articlesitems[] = (new ThePressScraper)->extract($link);
+        }
+
+        return $articlesitems;
+    }
+}