From c5a926446a5f98345158d1820334a027a19a6828 Mon Sep 17 00:00:00 2001
From: myrmidex <myrmidex@myrmidex.net>
Date: Sat, 28 Jun 2025 12:08:34 +0200
Subject: [PATCH] Fetch articles from news site

---
 bin/check-and-post                            | 21 +++++
 src/Domain/Articles/Article.php               | 19 +++--
 src/Domain/Articles/ArticleFetcher.php        | 79 +++++++++++++++++++
 .../Articles/Jobs/PostArticleToLemmy.php      | 36 +++++++++
 4 files changed, 148 insertions(+), 7 deletions(-)
 create mode 100644 bin/check-and-post
 create mode 100644 src/Domain/Articles/ArticleFetcher.php
 create mode 100644 src/Domain/Articles/Jobs/PostArticleToLemmy.php

diff --git a/bin/check-and-post b/bin/check-and-post
new file mode 100644
index 0000000..20726a7
--- /dev/null
+++ b/bin/check-and-post
@@ -0,0 +1,21 @@
+#!/usr/bin/env php
+<?php
+
+require __DIR__ . '/../bootstrap/autoload.php';
+
+use Feddev\LemmyArticlePoster\Domain\Articles\ArticleFetcher;
+
+$articles = ArticleFetcher::getNewArticles();
+
+$newArticles = $articles->filter(function($article) {
+    return $article->wasRecentlyCreated;
+});
+
+echo "Found " . $articles->count() . " articles (" . $newArticles->count() . " new)\n";
+
+$newArticles->each(function($article) {
+    echo "New article: " . $article->url . "\n";
+    // $article->publish();
+});
+
+echo "\n";
\ No newline at end of file
diff --git a/src/Domain/Articles/Article.php b/src/Domain/Articles/Article.php
index 8f2b3e3..dc5d300 100644
--- a/src/Domain/Articles/Article.php
+++ b/src/Domain/Articles/Article.php
@@ -1,7 +1,8 @@
 <?php
 
-namespace Domain\Articles;
+namespace Feddev\LemmyArticlePoster\Domain\Articles;
 
+use Feddev\LemmyArticlePoster\Domain\Articles\Jobs\PostArticleToLemmy;
 use Illuminate\Database\Eloquent\Model;
 use Illuminate\Database\Eloquent\Relations\HasMany;
 
@@ -28,12 +29,16 @@ public function posts(): HasMany
         return $this->hasMany(ArticlePosted::class);
     }
 
-    public function publish(string $community = 'default'): void
+    public function publish(?string $community = null): void
     {
-        ArticlePosted::create([
-            'article_id' => $this->id,
-            'community' => $community,
-            'posted_at' => now(),
-        ]);
+        $config = require __DIR__ . '/../../../bootstrap/config.php';
+        $community ??= $config['lemmy']['community'];
+
+        if ($this->posts()->where('community', $community)->exists()) {
+            return;
+        }
+
+        $job = new PostArticleToLemmy($this, $community);
+        $job->handle();
     }
 }
diff --git a/src/Domain/Articles/ArticleFetcher.php b/src/Domain/Articles/ArticleFetcher.php
new file mode 100644
index 0000000..cd38b37
--- /dev/null
+++ b/src/Domain/Articles/ArticleFetcher.php
@@ -0,0 +1,79 @@
+<?php
+
+namespace Feddev\LemmyArticlePoster\Domain\Articles;
+
+use Illuminate\Support\Collection;
+use Illuminate\Support\Carbon;
+use GuzzleHttp\Client;
+
+class ArticleFetcher
+{
+    public static function getNewArticles(): Collection
+    {
+        return self::fetchArticles()->map(fn ($article) => self::saveArticle($article));
+    }
+
+    private static function fetchArticles(): \Illuminate\Support\Collection
+    {
+        try {
+            $client = new Client();
+            $response = $client->get('https://www.vrt.be/vrtnws/en/');
+            $html = $response->getBody()->getContents();
+            
+            // Extract article links using regex
+            preg_match_all('/href="(\/vrtnws\/en\/\d{4}\/\d{2}\/\d{2}\/[^"]+)"/', $html, $matches);
+            
+            $articles = collect($matches[1] ?? [])
+                ->unique()
+                ->take(10) // Limit to 10 articles
+                ->map(function ($path) use ($client) {
+                    $fullUrl = 'https://www.vrt.be' . $path;
+                    
+                    try {
+                        // Fetch the article page to get title
+                        $articleResponse = $client->get($fullUrl);
+                        $articleHtml = $articleResponse->getBody()->getContents();
+                        
+                        // Extract title from meta tag or h1
+                        preg_match('/<meta property="og:title" content="([^"]+)"/', $articleHtml, $titleMatch);
+                        $title = $titleMatch[1] ?? null;
+                        
+                        // Extract publish date
+                        preg_match('/<meta property="article:published_time" content="([^"]+)"/', $articleHtml, $dateMatch);
+                        $publishedAt = $dateMatch[1] ?? null;
+                        
+                        return [
+                            'url' => $fullUrl,
+                            'title' => $title,
+                            'published_at' => $publishedAt,
+                        ];
+                    } catch (\Exception $e) {
+                        logger('article_fetcher')->error("Failed to fetch article: {$fullUrl}", ['error' => $e->getMessage()]);
+                        return [
+                            'url' => $fullUrl,
+                            'title' => null,
+                            'published_at' => null,
+                        ];
+                    }
+                })
+                ->filter(fn($article) => !empty($article['url']));
+                
+            logger('article_fetcher')->info("Fetched " . $articles->count() . " articles from VRT");
+            return $articles;
+            
+        } catch (\Exception $e) {
+            logger('article_fetcher')->error("Failed to fetch VRT homepage", ['error' => $e->getMessage()]);
+            return collect([]);
+        }
+    }
+
+    protected static function saveArticle(array $data): Article
+    {
+        return Article::firstOrCreate(
+            ['url' => $data['url']],
+            [
+                'created_at' => Carbon::now(),
+            ]
+        );
+    }
+}
\ No newline at end of file
diff --git a/src/Domain/Articles/Jobs/PostArticleToLemmy.php b/src/Domain/Articles/Jobs/PostArticleToLemmy.php
new file mode 100644
index 0000000..71a2423
--- /dev/null
+++ b/src/Domain/Articles/Jobs/PostArticleToLemmy.php
@@ -0,0 +1,36 @@
+<?php
+
+namespace Feddev\LemmyArticlePoster\Domain\Articles\Jobs;
+
+use Feddev\LemmyArticlePoster\Domain\Articles\Article;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Support\Carbon;
+
+class PostArticleToLemmy implements ShouldQueue
+{
+    public int $articleId;
+    public string $community;
+
+    public function __construct(Article $article, string $community = 'default')
+    {
+        $this->articleId = $article->id;
+        $this->community = $community;
+    }
+
+    public function handle(): void
+    {
+        $article = Article::findOrFail($this->articleId);
+
+        // TODO: Replace with real Lemmy API call
+        $fakeResponse = [
+            'status' => 'success',
+            'post_url' => 'https://lemmy.world/post/12345',
+        ];
+
+        $article->posts()->create([
+            'community' => $this->community,
+            'posted_at' => Carbon::now(),
+            'response' => $fakeResponse,
+        ]);
+    }
+}