optimizations

tests/Feature/Vision/ArtworkVectorRepairQueueTest.php

@@ -0,0 +1,153 @@
+<?php
+
+declare(strict_types=1);
+
+use App\Jobs\BackfillArtworkVectorIndexJob;
+use App\Jobs\SyncArtworkVectorIndexJob;
+use App\Models\Artwork;
+use App\Models\ArtworkEmbedding;
+use Illuminate\Foundation\Testing\RefreshDatabase;
+use Illuminate\Support\Facades\Queue;
+use function Pest\Laravel\artisan;
+
+uses(RefreshDatabase::class);
+
+function jobProperty(object $job, string $property): mixed
+{
+    $reflection = new ReflectionProperty($job, $property);
+    $reflection->setAccessible(true);
+
+    return $reflection->getValue($job);
+}
+
+it('queues a resumable vector repair run from the command', function (): void {
+    Queue::fake();
+
+    artisan('artworks:vectors-repair', [
+        '--after-id' => 25,
+        '--batch' => 50,
+        '--public-only' => true,
+        '--stale-hours' => 24,
+    ])->assertSuccessful();
+
+    Queue::assertPushed(BackfillArtworkVectorIndexJob::class, function (BackfillArtworkVectorIndexJob $job): bool {
+        return jobProperty($job, 'afterId') === 25
+            && jobProperty($job, 'batchSize') === 50
+            && jobProperty($job, 'publicOnly') === true
+            && jobProperty($job, 'staleHours') === 24;
+    });
+});
+
+it('fans out queued vector repair only for artworks with local embeddings', function (): void {
+    Queue::fake();
+
+    $embeddedArtwork = Artwork::factory()->create([
+        'hash' => '1122334455667788',
+        'is_public' => true,
+        'is_approved' => true,
+        'published_at' => now()->subDay(),
+    ]);
+    ArtworkEmbedding::query()->create([
+        'artwork_id' => $embeddedArtwork->id,
+        'model' => 'clip',
+        'model_version' => 'v1',
+        'algo_version' => 'clip-cosine-v1',
+        'dim' => 2,
+        'embedding_json' => json_encode([0.6, 0.8], JSON_THROW_ON_ERROR),
+        'source_hash' => '1122334455667788',
+        'is_normalized' => true,
+        'generated_at' => now(),
+        'meta' => ['source' => 'clip'],
+    ]);
+
+    Artwork::factory()->create([
+        'hash' => 'aabbccddeeff0011',
+        'is_public' => true,
+        'is_approved' => true,
+        'published_at' => now()->subDay(),
+    ]);
+
+    (new BackfillArtworkVectorIndexJob(0, 100, true))->handle();
+
+    Queue::assertPushed(SyncArtworkVectorIndexJob::class, function (SyncArtworkVectorIndexJob $job) use ($embeddedArtwork): bool {
+        return jobProperty($job, 'artworkId') === $embeddedArtwork->id;
+    });
+    Queue::assertNotPushed(BackfillArtworkVectorIndexJob::class);
+});
+
+it('can target only stale or never-indexed artworks during queued repair', function (): void {
+    Queue::fake();
+
+    $staleArtwork = Artwork::factory()->create([
+        'hash' => '1111222233334444',
+        'is_public' => true,
+        'is_approved' => true,
+        'published_at' => now()->subDay(),
+        'last_vector_indexed_at' => now()->subHours(48),
+    ]);
+    ArtworkEmbedding::query()->create([
+        'artwork_id' => $staleArtwork->id,
+        'model' => 'clip',
+        'model_version' => 'v1',
+        'algo_version' => 'clip-cosine-v1',
+        'dim' => 2,
+        'embedding_json' => json_encode([0.6, 0.8], JSON_THROW_ON_ERROR),
+        'source_hash' => '1111222233334444',
+        'is_normalized' => true,
+        'generated_at' => now(),
+        'meta' => ['source' => 'clip'],
+    ]);
+
+    $missingArtwork = Artwork::factory()->create([
+        'hash' => '5555666677778888',
+        'is_public' => true,
+        'is_approved' => true,
+        'published_at' => now()->subDay(),
+        'last_vector_indexed_at' => null,
+    ]);
+    ArtworkEmbedding::query()->create([
+        'artwork_id' => $missingArtwork->id,
+        'model' => 'clip',
+        'model_version' => 'v1',
+        'algo_version' => 'clip-cosine-v1',
+        'dim' => 2,
+        'embedding_json' => json_encode([0.6, 0.8], JSON_THROW_ON_ERROR),
+        'source_hash' => '5555666677778888',
+        'is_normalized' => true,
+        'generated_at' => now(),
+        'meta' => ['source' => 'clip'],
+    ]);
+
+    $freshArtwork = Artwork::factory()->create([
+        'hash' => '9999aaaabbbbcccc',
+        'is_public' => true,
+        'is_approved' => true,
+        'published_at' => now()->subDay(),
+        'last_vector_indexed_at' => now()->subHours(2),
+    ]);
+    ArtworkEmbedding::query()->create([
+        'artwork_id' => $freshArtwork->id,
+        'model' => 'clip',
+        'model_version' => 'v1',
+        'algo_version' => 'clip-cosine-v1',
+        'dim' => 2,
+        'embedding_json' => json_encode([0.6, 0.8], JSON_THROW_ON_ERROR),
+        'source_hash' => '9999aaaabbbbcccc',
+        'is_normalized' => true,
+        'generated_at' => now(),
+        'meta' => ['source' => 'clip'],
+    ]);
+
+    (new BackfillArtworkVectorIndexJob(0, 100, true, 24))->handle();
+
+    Queue::assertPushed(SyncArtworkVectorIndexJob::class, function (SyncArtworkVectorIndexJob $job) use ($staleArtwork): bool {
+        return jobProperty($job, 'artworkId') === $staleArtwork->id;
+    });
+    Queue::assertPushed(SyncArtworkVectorIndexJob::class, function (SyncArtworkVectorIndexJob $job) use ($missingArtwork): bool {
+        return jobProperty($job, 'artworkId') === $missingArtwork->id;
+    });
+    Queue::assertNotPushed(SyncArtworkVectorIndexJob::class, function (SyncArtworkVectorIndexJob $job) use ($freshArtwork): bool {
+        return jobProperty($job, 'artworkId') === $freshArtwork->id;
+    });
+    Queue::assertNotPushed(BackfillArtworkVectorIndexJob::class);
+});