Initial Commit

2026-08-29 00:17:41 +10:00
parent 37ff8e0a88
commit b2f2d8b104
13 changed files with 378 additions and 35 deletions
@@ -3,6 +3,7 @@
 namespace App\Console\Commands;

 use App\Models\Post;
+use App\Services\PostMediaCacher;
 use Illuminate\Console\Attributes\Description;
 use Illuminate\Console\Attributes\Signature;

@@ -21,28 +22,45 @@ class ScrapeFeed extends RunsPlaywrightScript
            return;
        }

+        $mediaCacher = app(PostMediaCacher::class);
        $posts = json_decode(file_get_contents($path), true);

        foreach ($posts as $post) {
-            Post::updateOrCreate(
-                ['id' => $post['id']],
-                [
-                    'code' => $post['code'],
-                    'username' => $post['username'],
-                    'full_name' => $post['full_name'],
-                    'caption' => $post['caption'],
-                    'accessibility_caption' => $post['accessibility_caption'],
-                    'is_video' => $post['is_video'],
-                    'is_carousel' => $post['is_carousel'],
-                    'like_count' => $post['like_count'],
-                    'comment_count' => $post['comment_count'],
-                    'taken_at' => $post['taken_at'],
-                    'image_url' => $post['image_url'],
-                    'video_url' => $post['video_url'],
-                    'carousel_images' => $post['carousel_images'],
-                    'profile_pic_url' => $post['profile_pic_url'],
-                ]
-            );
+            $attributes = [
+                'code' => $post['code'],
+                'username' => $post['username'],
+                'full_name' => $post['full_name'],
+                'caption' => $post['caption'],
+                'accessibility_caption' => $post['accessibility_caption'],
+                'is_video' => $post['is_video'],
+                'is_carousel' => $post['is_carousel'],
+                'like_count' => $post['like_count'],
+                'comment_count' => $post['comment_count'],
+                'taken_at' => $post['taken_at'],
+                'image_url' => $post['image_url'],
+                'video_url' => $post['video_url'],
+                'carousel_images' => $post['carousel_images'],
+                'profile_pic_url' => $post['profile_pic_url'],
+            ];
+
+            $existing = Post::find($post['id']);
+
+            // Once a post's media has been cached locally, its url columns
+            // no longer point at Instagram — don't let a re-scrape clobber
+            // them with a freshly (and just as temporarily) signed CDN url.
+            if ($existing) {
+                if (! $mediaCacher->isInstagramUrl($existing->image_url)) {
+                    unset($attributes['image_url']);
+                }
+                if (! $mediaCacher->isInstagramUrl($existing->video_url)) {
+                    unset($attributes['video_url']);
+                }
+                if ($existing->carousel_images && ! $mediaCacher->isInstagramUrl($existing->carousel_images[0]['url'] ?? null)) {
+                    unset($attributes['carousel_images']);
+                }
+            }
+
+            Post::updateOrCreate(['id' => $post['id']], $attributes);
        }

        $this->info('Upserted ' . count($posts) . ' posts.');