Decouple source and fetching

app/Console/Commands/FetchArticleCommand.php

@@ -0,0 +1,27 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\Models\Article;
+use App\Services\Article\ArticleFetcher;
+use Illuminate\Console\Command;
+
+class FetchArticleCommand extends Command
+{
+    protected $signature = 'article:fetch {url}';
+
+    protected $description = 'Fetch article from url';
+
+    public function handle(): int
+    {
+        $article = Article::createQuietly([
+            'url' => $this->argument('url'),
+        ]);
+
+        $res = ArticleFetcher::fetchArticleData($article);
+
+        dump($res);
+
+        return self::SUCCESS;
+    }
+}

app/Console/Commands/PublishToLemmyCommand.php

@@ -21,7 +21,7 @@ public function handle(): int
         $this->info('Publishing article: ' . $article->url);
 
         try {
-            LemmyPublisher::fromConfig()->publish($article, ArticleFetcher::fetchArticle($article));
+            LemmyPublisher::fromConfig()->publish($article, ArticleFetcher::fetchArticleData($article));
         } catch (Exception) {
             return self::FAILURE;
         }

app/Contracts/ArticleParserInterface.php

@@ -0,0 +1,21 @@
+<?php
+
+namespace App\Contracts;
+
+interface ArticleParserInterface
+{
+    /**
+     * Check if this parser can handle the given URL
+     */
+    public function canParse(string $url): bool;
+
+    /**
+     * Extract article data from HTML
+     */
+    public function extractData(string $html): array;
+
+    /**
+     * Get the source name for this parser
+     */
+    public function getSourceName(): string;
+}

app/Contracts/HomepageParserInterface.php

@@ -0,0 +1,26 @@
+<?php
+
+namespace App\Contracts;
+
+interface HomepageParserInterface
+{
+    /**
+     * Check if this parser can handle the given homepage URL
+     */
+    public function canParse(string $url): bool;
+
+    /**
+     * Extract article URLs from homepage HTML
+     */
+    public function extractArticleUrls(string $html): array;
+
+    /**
+     * Get the homepage URL for this source
+     */
+    public function getHomepageUrl(): string;
+
+    /**
+     * Get the source name for this parser
+     */
+    public function getSourceName(): string;
+}

app/Listeners/PublishArticle.php

@@ -18,6 +18,6 @@ public function handle(ArticleReadyToPublish $event): void
 
         logger('Publishing article: ' . $article->id . ' : ' . $article->url);
 
-        LemmyPublisher::fromConfig()->publish($article, ArticleFetcher::fetchArticle($article));
+        LemmyPublisher::fromConfig()->publish($article, ArticleFetcher::fetchArticleData($article));
     }
 }

app/Services/Article/ArticleFetcher.php

@@ -3,87 +3,49 @@
 namespace App\Services\Article;
 
 use App\Models\Article;
+use App\Services\Http\HttpFetcher;
+use App\Services\Factories\ArticleParserFactory;
+use App\Services\Factories\HomepageParserFactory;
 use Exception;
 use Illuminate\Support\Collection;
-use Illuminate\Support\Facades\Http;
 
 class ArticleFetcher
 {
     public static function getNewArticles(): Collection
     {
-        return self::fetchArticles()
+        try {
+            $allArticles = collect();
+
+            foreach (HomepageParserFactory::getAllParsers() as $parser) {
+                $html = HttpFetcher::fetchHtml($parser->getHomepageUrl());
+                $urls = $parser->extractArticleUrls($html);
+
+                $articles = collect($urls)
                     ->map(fn (string $url) => self::saveArticle($url));
+
+                $allArticles = $allArticles->merge($articles);
             }
 
-    public static function fetchArticle(Article $article): array
+            return $allArticles->filter();
+        } catch (Exception $e) {
+            logger()->error("Failed to get new articles", ['error' => $e->getMessage()]);
+            return new Collection([]);
+        }
+    }
+
+    public static function fetchArticleData(Article $article): array
     {
         try {
-            $response = Http::get($article->url);
-
-            if (!$response->successful()) {
-                logger()->error('Failed to fetch article', [
-                    'url' => $article->url,
-                    'status' => $response->status()
-                ]);
-                return [];
-            }
-
-            $html = $response->body();
-
-            return ArticleDataExtractor::extractData($html);
+            $html = HttpFetcher::fetchHtml($article->url);
+            $parser = ArticleParserFactory::getParser($article->url);
 
+            return $parser->extractData($html);
         } catch (Exception $e) {
-            logger()->error('Exception while fetching article', [
+            logger()->error('Exception while fetching article data', [
                 'url' => $article->url,
                 'error' => $e->getMessage()
             ]);
-            return $article;
-        }
-    }
-
-
-    private static function fetchArticles(): Collection
-    {
-        try {
-            $response = Http::get('https://www.vrt.be/vrtnws/en/');
-            $html = $response->body();
-
-            // Extract article links using regex
-            preg_match_all('/href="(\/vrtnws\/en\/\d{4}\/\d{2}\/\d{2}\/[^"]+)"/', $html, $matches);
-
-            $urls = collect($matches[1] ?? [])
-                ->unique()
-                ->map(fn ($path) => 'https://www.vrt.be' . $path)
-                ->toArray();
-
-            $responses = Http::pool(function ($pool) use ($urls) {
-                foreach ($urls as $url) {
-                    $pool->get($url);
-                }
-            });
-
-            return collect($responses)
-                ->map(function ($response, $index) use ($urls) {
-                    if (!isset($urls[$index])) {
-                        return null;
-                    }
-
-                    $url = $urls[$index];
-
-                    try {
-                        if ($response->successful()) {
-                            return $url;
-                        } else {
-                            return null;
-                        }
-                    } catch (Exception) {
-                        return null;
-                    }
-                })
-                ->filter(fn($article) => !empty($article));
-        } catch (Exception $e) {
-            logger()->error("Failed to fetch VRT homepage", ['error' => $e->getMessage()]);
-            return new Collection([]);
+            return [];
         }
     }
 

app/Services/Article/ValidationService.php

@@ -10,11 +10,29 @@ public static function validate(Article $article): Article
     {
         logger('Checking keywords for article: ' . $article->id);
 
+        $articleData = ArticleFetcher::fetchArticleData($article);
+        $validationResult = self::validateByKeywords($articleData['full_article']);
+
         $article->update([
-            'is_valid' => true,
+            'is_valid' => $validationResult,
             'validated_at' => now(),
         ]);
 
         return $article->refresh();
     }
+
+    private static function validateByKeywords(string $full_article): bool
+    {
+        $keywords = [
+            'N-VA', 'Bart De Wever', 'Frank Vandenbroucke',
+        ];
+
+        foreach ($keywords as $keyword) {
+            if (stripos($full_article, $keyword) !== false) {
+                return true;
+            }
+        }
+
+        return false;
+    }
 }

app/Services/Factories/ArticleParserFactory.php

@@ -0,0 +1,42 @@
+<?php
+
+namespace App\Services\Factories;
+
+use App\Contracts\ArticleParserInterface;
+use App\Services\Parsers\VrtArticleParser;
+use Exception;
+
+class ArticleParserFactory
+{
+    private static array $parsers = [
+        VrtArticleParser::class,
+    ];
+
+    public static function getParser(string $url): ArticleParserInterface
+    {
+        foreach (self::$parsers as $parserClass) {
+            $parser = new $parserClass();
+            
+            if ($parser->canParse($url)) {
+                return $parser;
+            }
+        }
+
+        throw new Exception("No parser found for URL: {$url}");
+    }
+
+    public static function getSupportedSources(): array
+    {
+        return array_map(function($parserClass) {
+            $parser = new $parserClass();
+            return $parser->getSourceName();
+        }, self::$parsers);
+    }
+
+    public static function registerParser(string $parserClass): void
+    {
+        if (!in_array($parserClass, self::$parsers)) {
+            self::$parsers[] = $parserClass;
+        }
+    }
+}

app/Services/Factories/HomepageParserFactory.php

@@ -0,0 +1,47 @@
+<?php
+
+namespace App\Services\Factories;
+
+use App\Contracts\HomepageParserInterface;
+use App\Services\Parsers\VrtHomepageParserAdapter;
+use Exception;
+
+class HomepageParserFactory
+{
+    private static array $parsers = [
+        VrtHomepageParserAdapter::class,
+    ];
+
+    public static function getParser(string $url): HomepageParserInterface
+    {
+        foreach (self::$parsers as $parserClass) {
+            $parser = new $parserClass();
+            
+            if ($parser->canParse($url)) {
+                return $parser;
+            }
+        }
+
+        throw new Exception("No homepage parser found for URL: {$url}");
+    }
+
+    public static function getAllParsers(): array
+    {
+        return array_map(fn($parserClass) => new $parserClass(), self::$parsers);
+    }
+
+    public static function getSupportedSources(): array
+    {
+        return array_map(function($parserClass) {
+            $parser = new $parserClass();
+            return $parser->getSourceName();
+        }, self::$parsers);
+    }
+
+    public static function registerParser(string $parserClass): void
+    {
+        if (!in_array($parserClass, self::$parsers)) {
+            self::$parsers[] = $parserClass;
+        }
+    }
+}

app/Services/Http/HttpFetcher.php

@@ -0,0 +1,77 @@
+<?php
+
+namespace App\Services\Http;
+
+use Illuminate\Support\Facades\Http;
+use Exception;
+
+class HttpFetcher
+{
+    public static function fetchHtml(string $url): string
+    {
+        try {
+            $response = Http::get($url);
+            
+            if (!$response->successful()) {
+                throw new Exception("Failed to fetch URL: {$url} - Status: {$response->status()}");
+            }
+            
+            return $response->body();
+        } catch (Exception $e) {
+            logger()->error('HTTP fetch failed', [
+                'url' => $url,
+                'error' => $e->getMessage()
+            ]);
+            throw $e;
+        }
+    }
+
+    public static function fetchMultipleUrls(array $urls): array
+    {
+        try {
+            $responses = Http::pool(function ($pool) use ($urls) {
+                foreach ($urls as $url) {
+                    $pool->get($url);
+                }
+            });
+
+            return collect($responses)
+                ->map(function ($response, $index) use ($urls) {
+                    if (!isset($urls[$index])) {
+                        return null;
+                    }
+                    
+                    $url = $urls[$index];
+                    
+                    try {
+                        if ($response->successful()) {
+                            return [
+                                'url' => $url,
+                                'html' => $response->body(),
+                                'success' => true
+                            ];
+                        } else {
+                            return [
+                                'url' => $url,
+                                'html' => null,
+                                'success' => false,
+                                'status' => $response->status()
+                            ];
+                        }
+                    } catch (Exception) {
+                        return [
+                            'url' => $url,
+                            'html' => null,
+                            'success' => false,
+                            'error' => 'Exception occurred'
+                        ];
+                    }
+                })
+                ->filter(fn($result) => $result !== null)
+                ->toArray();
+        } catch (Exception $e) {
+            logger()->error('Multiple URL fetch failed', ['error' => $e->getMessage()]);
+            return [];
+        }
+    }
+}

app/Services/Parsers/VrtArticlePageParser.php

@@ -1,8 +1,8 @@
 <?php
 
-namespace App\Services\Article;
+namespace App\Services\Parsers;
 
-class ArticleDataExtractor
+class VrtArticlePageParser
 {
     public static function extractTitle(string $html): ?string
     {
@@ -39,15 +39,6 @@ public static function extractDescription(string $html): ?string
         return null;
     }
     
-    public static function extractData(string $html): array
-    {
-        return [
-            'title' => self::extractTitle($html),
-            'description' => self::extractDescription($html),
-            'full_article' => self::extractFullArticle($html),
-        ];
-    }
-    
     public static function extractFullArticle(string $html): ?string
     {
         // Remove scripts, styles, and other non-content elements
@@ -72,4 +63,13 @@ public static function extractFullArticle(string $html): ?string
         
         return null;
     }
+    
+    public static function extractData(string $html): array
+    {
+        return [
+            'title' => self::extractTitle($html),
+            'description' => self::extractDescription($html),
+            'full_article' => self::extractFullArticle($html),
+        ];
+    }
 }

app/Services/Parsers/VrtArticleParser.php

@@ -0,0 +1,23 @@
+<?php
+
+namespace App\Services\Parsers;
+
+use App\Contracts\ArticleParserInterface;
+
+class VrtArticleParser implements ArticleParserInterface
+{
+    public function canParse(string $url): bool
+    {
+        return str_contains($url, 'vrt.be');
+    }
+
+    public function extractData(string $html): array
+    {
+        return VrtArticlePageParser::extractData($html);
+    }
+
+    public function getSourceName(): string
+    {
+        return 'VRT News';
+    }
+}

app/Services/Parsers/VrtHomepageParser.php

@@ -0,0 +1,19 @@
+<?php
+
+namespace App\Services\Parsers;
+
+class VrtHomepageParser
+{
+    public static function extractArticleUrls(string $html): array
+    {
+        // Extract article links using regex
+        preg_match_all('/href="(\/vrtnws\/en\/\d{4}\/\d{2}\/\d{2}\/[^"]+)"/', $html, $matches);
+
+        $urls = collect($matches[1] ?? [])
+            ->unique()
+            ->map(fn ($path) => 'https://www.vrt.be' . $path)
+            ->toArray();
+
+        return $urls;
+    }
+}

app/Services/Parsers/VrtHomepageParserAdapter.php

@@ -0,0 +1,28 @@
+<?php
+
+namespace App\Services\Parsers;
+
+use App\Contracts\HomepageParserInterface;
+
+class VrtHomepageParserAdapter implements HomepageParserInterface
+{
+    public function canParse(string $url): bool
+    {
+        return str_contains($url, 'vrt.be');
+    }
+
+    public function extractArticleUrls(string $html): array
+    {
+        return VrtHomepageParser::extractArticleUrls($html);
+    }
+
+    public function getHomepageUrl(): string
+    {
+        return 'https://www.vrt.be/vrtnws/en/';
+    }
+
+    public function getSourceName(): string
+    {
+        return 'VRT News';
+    }
+}