Psm Support

app/Console/Commands/ScrapePsmCommand.php

@@ -0,0 +1,73 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\Source;
+use Illuminate\Console\Command;
+use App\Services\PsmService;
+use App\Topic;
+use Illuminate\Support\Carbon;
+
+class ScrapePsmCommand extends Command
+{
+    /**
+     * The name and signature of the console command.
+     *
+     * @var string
+     */
+    protected $signature = 'scrape:psm';
+
+    /**
+     * The console command description.
+     *
+     * @var string
+     */
+    protected $description = 'Scrape PSM News';
+
+    /**
+     * 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', 'psm')->first();
+
+        $articles = (new PsmService)->scrape();
+
+        foreach ($articles as $article) {
+
+            // Attach the relationship between source and article and return the curren article instance
+            $articleModel = $source->articles()->updateOrCreate(["guid" => $article["guid"]],
+                [
+                "title" => $article["title"],
+                "url" => $article["url"],
+                "author" => $article["author"],
+                "featured_image" => $article["image"],
+                "body" => $article["content"],
+                "published_date" => Carbon::parse($article["date"])->format("Y-m-d H:i:s"),
+                "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

@@ -51,6 +51,9 @@ class Kernel extends ConsoleKernel
 
         $schedule->command('scrape:dhen')->everyFiveMinutes()
                  ->pingOnSuccess(env('APP_URL')."/api/ping/dhen");
+        
+        $schedule->command('scrape:psm')->everyFiveMinutes()
+                 ->pingOnSuccess(env('APP_URL')."/api/ping/psm");
     }
 
     /**

app/Services/PsmService.php

@@ -0,0 +1,29 @@
+<?php
+
+namespace App\Services;
+
+use App\Services\Scrapers\PsmScraper;
+use Illuminate\Support\Str;
+
+class PsmService extends Client
+{
+    /**
+     * Scrap all the rss articles from mihaaru
+     *
+     * @return array
+     */
+    public function scrape(): array
+    {
+        $articles = $this->get("https://www.psmnews.mv/feed")["entry"];
+
+        $articlesitems = [];
+        //Looping through the articles and scraping and while scraping it creates a new instance of the scraper.
+        foreach ($articles as $article) {
+            $link = $article['id'];
+            $date = $article['updated'];
+            $articlesitems[] = (new PsmScraper)->extract($link, $date);
+        }
+
+        return $articlesitems;
+    }
+}

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"
+                ]
+            ]
+        ];
+    }
+}