refactored llms-full for published articles

src/App/src/Factory/LlmsFullGeneratorFactory.php

@@ -5,15 +5,22 @@
 namespace Light\App\Factory;
 
 use Light\App\Service\LlmsFullGenerator;
+use Light\Blog\Repository\PostRepository;
 use Psr\Container\ContainerInterface;
 
+use function assert;
+
 class LlmsFullGeneratorFactory
 {
     public function __invoke(ContainerInterface $container): LlmsFullGenerator
     {
+        $postRepository = $container->get(PostRepository::class);
+        assert($postRepository instanceof PostRepository);
+
         $config = $container->get('config');
 
         return new LlmsFullGenerator(
+            $postRepository,
             $config['llms']['sourceDir'],
             $config['llms']['outputFile'],
             $config['application']['url'] ?? '',

src/App/src/Service/LlmsFullGenerator.php

@@ -4,6 +4,8 @@
 
 namespace Light\App\Service;
 
+use Light\Blog\Entity\Post;
+use Light\Blog\Repository\PostRepository;
 use RuntimeException;
 
 use function array_map;
@@ -15,27 +17,23 @@
 use function glob;
 use function implode;
 use function is_file;
-use function ltrim;
+use function preg_replace;
 use function sort;
 use function sprintf;
 use function str_replace;
 use function trim;
 
 /**
- * Concatenates every article under the markdown source directory into a single
- * plain-text file for LLM consumption, in the same format as llms.txt/llms-full.txt
- * conventions (index.md first, then every other article sorted by relative path).
- *
- * When a pages directory is configured, the flat `*.md` files in it - the markdown
- * versions of the static pages - are appended after the articles, labelled with a
- * `md-pages/` prefix so their origin stays readable in the output.
+ * Generate llms-full.txt from index, published articles, pages
+ * FAQ is removed
  */
 readonly class LlmsFullGenerator
 {
     private const SEPARATOR  = '<!-- ============================================================ -->';
     private const PAGES_ROOT = 'md-pages';
 
     public function __construct(
+        private PostRepository $postRepository,
         private string $sourceDir,
         private string $outputFile,
         private string $baseUrl,
@@ -73,19 +71,18 @@ public function write(): int
      */
     private function collectEntries(): array
     {
-        $categoryFiles = glob($this->sourceDir . '/*/*.md') ?: [];
-        $categoryFiles = array_map(
-            fn (string $path): string => ltrim(str_replace($this->sourceDir, '', $path), '/'),
-            $categoryFiles
+        $articleLabels = array_map(
+            fn (Post $post): string => sprintf('%s/%s.md', $post->getCategory()->getSlug(), $post->getSlug()),
+            $this->postRepository->getPublishedPosts()
         );
-        sort($categoryFiles);
+        sort($articleLabels);
 
         $labels = is_file($this->sourceDir . '/index.md') ? ['index.md'] : [];
-        $labels = array_merge($labels, $categoryFiles);
+        $labels = array_merge($labels, $articleLabels);
 
         $entries = array_map(
             fn (string $label): array => [
-                'file'  => $this->sourceDir . '/' . $label,
+                'file'  => $this->resolveArticleFile($label),
                 'label' => $label,
             ],
             $labels
@@ -94,6 +91,17 @@ private function collectEntries(): array
         return array_merge($entries, $this->collectPageEntries());
     }
 
+    private function resolveArticleFile(string $label): string
+    {
+        $file = $this->sourceDir . '/' . $label;
+
+        if (! is_file($file)) {
+            throw new RuntimeException("Missing markdown file for published post: {$label}");
+        }
+
+        return $file;
+    }
+
     /**
      * The flat `*.md` files in the pages directory, appended after the articles.
      *
@@ -125,6 +133,7 @@ private function buildSection(string $file, string $relativePath): string
         }
 
         $contents = str_replace('{{base_url}}', $this->baseUrl, $contents);
+        $contents = $this->stripFaqSection($contents);
 
         return sprintf(
             "%s\n<!-- %s -->\n%s\n\n%s",
@@ -134,4 +143,14 @@ private function buildSection(string $file, string $relativePath): string
             trim($contents)
         );
     }
+
+    /**
+     * Removes the FAQ section, leaves the rest of the content
+     */
+    private function stripFaqSection(string $contents): string
+    {
+        $contents = preg_replace('/^## FAQ\s*$.*?(?=^## |\z)/ms', '', $contents) ?? $contents;
+
+        return preg_replace('/\n{3,}/', "\n\n", $contents) ?? $contents;
+    }
 }

test/Unit/App/Factory/LlmsFullGeneratorFactoryTest.php

@@ -6,6 +6,7 @@
 
 use Light\App\Factory\LlmsFullGeneratorFactory;
 use Light\App\Service\LlmsFullGenerator;
+use Light\Blog\Repository\PostRepository;
 use LightTest\Unit\UnitTest;
 use PHPUnit\Framework\Attributes\DataProvider;
 use PHPUnit\Framework\MockObject\Exception;
@@ -94,7 +95,12 @@ public static function missingUrlProvider(): array
     private function createContainer(array $config): ContainerInterface
     {
         $container = $this->createStub(ContainerInterface::class);
-        $container->method('get')->willReturn($config);
+        $container
+            ->method('get')
+            ->willReturnCallback(fn (string $id): mixed => match ($id) {
+                'config' => $config,
+                default  => $this->createStub(PostRepository::class),
+            });
 
         return $container;
     }