SkinbaseNova/tests/Feature/Vision/ArtworkVectorGatewayCommandsTest.php

<?php

declare(strict_types=1);

use App\Models\Artwork;
use App\Models\ArtworkEmbedding;
use App\Models\Category;
use App\Models\ContentType;
use App\Models\Tag;
use Illuminate\Foundation\Testing\RefreshDatabase;
use Illuminate\Support\Facades\Http;
use function Pest\Laravel\artisan;

uses(RefreshDatabase::class);

beforeEach(function (): void {
    config()->set('vision.vector_gateway.enabled', true);
    config()->set('vision.vector_gateway.base_url', 'https://vision.klevze.net');
    config()->set('vision.vector_gateway.api_key', 'test-key');
    config()->set('vision.vector_gateway.upsert_endpoint', '/vectors/upsert');
    config()->set('vision.vector_gateway.upsert_file_endpoint', '/vectors/upsert/file');
    config()->set('vision.vector_gateway.search_endpoint', '/vectors/search');
    config()->set('vision.vector_gateway.search_point_endpoint', '/vectors/search/point');
    config()->set('vision.image_variant', 'md');
    config()->set('cdn.files_url', 'https://files.skinbase.org');
});

it('indexes artworks into the vector gateway with artwork metadata', function (): void {
    $contentType = ContentType::query()->create([
        'name' => 'Photography',
        'slug' => 'photography',
        'description' => '',
    ]);

    $category = Category::query()->create([
        'content_type_id' => $contentType->id,
        'parent_id' => null,
        'name' => 'Abstract',
        'slug' => 'abstract',
        'description' => '',
        'is_active' => true,
        'sort_order' => 0,
    ]);

    $artwork = Artwork::factory()->create([
        'hash' => 'aabbcc112233',
        'thumb_ext' => 'webp',
    ]);
    $artwork->categories()->attach($category->id);
    $tag = Tag::query()->create(['name' => 'Skyline', 'slug' => 'skyline']);
    $artwork->tags()->attach($tag->id, ['source' => 'ai', 'confidence' => 0.88]);

    Http::fake([
        'https://files.skinbase.org/*'
            => Http::response(
                'fake-image-bytes',
                200,
                ['Content-Type' => 'image/webp']
            ),

        'https://vision.klevze.net/vectors/upsert/file'
            => Http::response([
                'id' => $artwork->id,
                'collection' => 'images',
                'dim' => 512,
            ], 200),
    ]);

    artisan('artworks:vectors-index', ['--limit' => 1])
        ->assertSuccessful();

    $artwork->refresh();

    expect($artwork->last_vector_indexed_at)
        ->not->toBeNull();

    Http::assertSentCount(2);

    Http::assertSent(function ($request): bool {
        return $request->method() === 'GET'
            && $request->url()
                === 'https://files.skinbase.org/artworks/md/aa/bb/aabbcc112233.webp';
    });

    Http::assertSent(function ($request) use ($artwork): bool {
        if (
            $request->url()
            !== 'https://vision.klevze.net/vectors/upsert/file'
        ) {
            return false;
        }

        $parts = collect(
            $request->data()
        )->keyBy('name');

        $idPart = $parts->get('id', []);
        $metadataPart = $parts->get(
            'metadata_json',
            []
        );
        $filePart = $parts->get('file', []);

        $metadata = json_decode(
            (string) (
                $metadataPart['contents'] ?? ''
            ),
            true
        );

        return $request->hasHeader(
                'X-API-Key',
                'test-key'
            )
            && (string) (
                $idPart['contents'] ?? ''
            ) === (string) $artwork->id
            && (string) (
                $filePart['contents'] ?? ''
            ) === 'fake-image-bytes'
            && (string) (
                $filePart['filename'] ?? ''
            ) === sprintf(
                'artwork-%d.webp',
                $artwork->id
            )
            && is_array($metadata)
            && ($metadata['content_type'] ?? null)
                === 'Photography'
            && ($metadata['category'] ?? null)
                === 'Abstract'
            && ($metadata['tags'] ?? null)
                === ['skyline']
            && array_key_exists(
                'is_public',
                $metadata
            )
            && array_key_exists(
                'is_deleted',
                $metadata
            )
            && array_key_exists(
                'is_nsfw',
                $metadata
            )
            && array_key_exists(
                'category_id',
                $metadata
            )
            && array_key_exists(
                'content_type_id',
                $metadata
            )
            && array_key_exists(
                'status',
                $metadata
            );
    });
});

it('searches similar artworks through the vector gateway', function (): void {
    $contentType = ContentType::query()->create([
        'name' => 'Wallpapers',
        'slug' => 'wallpapers',
        'description' => '',
    ]);

    $category = Category::query()->create([
        'content_type_id' => $contentType->id,
        'parent_id' => null,
        'name' => 'Nature',
        'slug' => 'nature',
        'description' => '',
        'is_active' => true,
        'sort_order' => 0,
    ]);

    $source = Artwork::factory()->create([
        'title' => 'Source artwork',
        'hash' => 'aabbcc112233',
        'thumb_ext' => 'webp',
    ]);
    $source->categories()->attach($category->id);

    $similar = Artwork::factory()->create([
        'title' => 'Nearby artwork',
        'hash' => 'ddeeff445566',
        'thumb_ext' => 'webp',
    ]);
    $similar->categories()->attach($category->id);

    Http::fake([
        'https://vision.klevze.net/vectors/search/point'
            => Http::response([
                'results' => [
                    [
                        'id' => $source->id,
                        'score' => 1.0,
                    ],
                    [
                        'id' => $similar->id,
                        'score' => 0.9876,
                    ],
                ],
            ], 200),
    ]);

    artisan('artworks:vectors-search', [
        'artwork_id' => $source->id,
        '--limit' => 5,
    ])
        ->expectsTable(['ID', 'Score', 'Title', 'Content Type', 'Category'], [[
            'id' => $similar->id,
            'score' => '0.9876',
            'title' => 'Nearby artwork',
            'content_type' => 'Wallpapers',
            'category' => 'Nature',
        ]])
        ->assertSuccessful();

    Http::assertSentCount(1);

    Http::assertSent(
        function ($request) use ($source): bool {
            return $request->url()
                === 'https://vision.klevze.net/vectors/search/point'
                && ($request->data()['id'] ?? null)
                    === $source->id;
        }
    );

    Http::assertNotSent(
        fn ($request): bool =>
            str_contains(
                $request->url(),
                'files.skinbase.org'
            )
    );
});

it('can re-upsert only artworks that already have local embeddings', function (): void {
    $contentType = ContentType::query()->create([
        'name' => 'Photography',
        'slug' => 'photography',
        'description' => '',
    ]);

    $category = Category::query()->create([
        'content_type_id' => $contentType->id,
        'parent_id' => null,
        'name' => 'Portraits',
        'slug' => 'portraits',
        'description' => '',
        'is_active' => true,
        'sort_order' => 0,
    ]);

    $embeddedArtwork = Artwork::factory()->create([
        'hash' => '112233445566',
        'thumb_ext' => 'webp',
        'is_public' => true,
        'is_approved' => true,
        'published_at' => now()->subDay(),
    ]);
    $embeddedArtwork->categories()->attach($category->id);

    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' => '112233445566',
        'is_normalized' => true,
        'generated_at' => now(),
        'meta' => ['source' => 'clip'],
    ]);

    $nonEmbeddedArtwork = Artwork::factory()->create([
        'hash' => 'aabbccddeeff',
        'thumb_ext' => 'webp',
        'is_public' => true,
        'is_approved' => true,
        'published_at' => now()->subDay(),
    ]);
    $nonEmbeddedArtwork->categories()->attach($category->id);

    Http::fake([
        'https://files.skinbase.org/*'
            => Http::response(
                'fake-image-bytes',
                200,
                ['Content-Type' => 'image/webp']
            ),

        'https://vision.klevze.net/vectors/upsert/file'
            => Http::response([
                'id' => $embeddedArtwork->id,
                'collection' => 'images',
                'dim' => 512,
            ], 200),
    ]);

    artisan('artworks:vectors-index', [
        '--embedded-only' => true,
        '--limit' => 10,
    ])->assertSuccessful();

    $embeddedArtwork->refresh();
    $nonEmbeddedArtwork->refresh();

    expect($embeddedArtwork->last_vector_indexed_at)
        ->not->toBeNull()
        ->and(
            $nonEmbeddedArtwork->last_vector_indexed_at
        )
        ->toBeNull();

    /*
     * One indexed artwork:
     *   1x CDN image download
     *   1x /vectors/upsert/file
     */
    Http::assertSentCount(2);

    Http::assertSent(
        function ($request): bool {
            return $request->method() === 'GET'
                && $request->url()
                    === 'https://files.skinbase.org/artworks/md/11/22/112233445566.webp';
        }
    );

    Http::assertSent(
        function ($request) use (
            $embeddedArtwork
        ): bool {
            if (
                $request->url()
                !== 'https://vision.klevze.net/vectors/upsert/file'
            ) {
                return false;
            }

            $parts = collect(
                $request->data()
            )->keyBy('name');

            $idPart = $parts->get('id', []);
            $filePart = $parts->get('file', []);

            return (string) (
                $idPart['contents'] ?? ''
            ) === (string) $embeddedArtwork->id
                && (string) (
                    $filePart['contents'] ?? ''
                ) === 'fake-image-bytes'
                && (string) (
                    $filePart['filename'] ?? ''
                ) === sprintf(
                    'artwork-%d.webp',
                    $embeddedArtwork->id
                );
        }
    );
});