Harden vision vector search and embedding jobs.
Add point search, circuit-breaking, and safer gateway limits so artwork similarity lookups fail closed instead of hanging the request path.
This commit is contained in:
@@ -7,9 +7,8 @@ namespace App\Jobs;
|
||||
use App\Models\Artwork;
|
||||
use App\Models\ArtworkEmbedding;
|
||||
use App\Services\Vision\ArtworkEmbeddingClient;
|
||||
use App\Services\Vision\ArtworkVisionImageUrl;
|
||||
use App\Services\Vision\ArtworkVectorIndexService;
|
||||
use App\Services\Vision\VectorService;
|
||||
use App\Services\Vision\ArtworkVisionImageUrl;
|
||||
use Illuminate\Bus\Queueable;
|
||||
use Illuminate\Contracts\Queue\ShouldQueue;
|
||||
use Illuminate\Foundation\Bus\Dispatchable;
|
||||
@@ -49,8 +48,7 @@ final class GenerateArtworkEmbeddingJob implements ShouldQueue
|
||||
ArtworkEmbeddingClient $client,
|
||||
ArtworkVisionImageUrl $imageUrlBuilder,
|
||||
ArtworkVectorIndexService $vectors,
|
||||
): void
|
||||
{
|
||||
): void {
|
||||
if (! (bool) config('recommendations.embedding.enabled', true)) {
|
||||
return;
|
||||
}
|
||||
@@ -78,17 +76,73 @@ final class GenerateArtworkEmbeddingJob implements ShouldQueue
|
||||
->where('model_version', $modelVersion)
|
||||
->first();
|
||||
|
||||
if ($existing && (string) ($existing->source_hash ?? '') === $sourceHash) {
|
||||
if (
|
||||
$existing
|
||||
&& (string) ($existing->source_hash ?? '') === $sourceHash
|
||||
&& $this->vectorIndexCoversEmbedding(
|
||||
$artwork,
|
||||
$existing
|
||||
)
|
||||
) {
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
$lockKey = $this->lockKey($artwork->id, $model, $modelVersion);
|
||||
$lockKey = $this->lockKey(
|
||||
$artwork->id,
|
||||
$model,
|
||||
$modelVersion
|
||||
);
|
||||
|
||||
if (! $this->acquireLock($lockKey)) {
|
||||
return;
|
||||
}
|
||||
|
||||
try {
|
||||
/*
|
||||
* Another worker may have generated or indexed the
|
||||
* embedding between our first read and lock acquisition.
|
||||
* Re-read both records under the lock.
|
||||
*/
|
||||
if (! $this->force) {
|
||||
$artwork->refresh();
|
||||
|
||||
$existing = ArtworkEmbedding::query()
|
||||
->where('artwork_id', $artwork->id)
|
||||
->where('model', $model)
|
||||
->where('model_version', $modelVersion)
|
||||
->first();
|
||||
|
||||
if (
|
||||
$existing
|
||||
&& (string) ($existing->source_hash ?? '')
|
||||
=== $sourceHash
|
||||
) {
|
||||
if (
|
||||
$this->vectorIndexCoversEmbedding(
|
||||
$artwork,
|
||||
$existing
|
||||
)
|
||||
) {
|
||||
return;
|
||||
}
|
||||
|
||||
$storedVector = $this->decodeStoredVector(
|
||||
$existing
|
||||
);
|
||||
|
||||
if ($storedVector !== []) {
|
||||
$this->upsertVectorIndex(
|
||||
$vectors,
|
||||
$artwork,
|
||||
$storedVector
|
||||
);
|
||||
|
||||
return;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
$imageUrl = $imageUrlBuilder->fromArtwork($artwork);
|
||||
if ($imageUrl === null) {
|
||||
return;
|
||||
@@ -121,23 +175,33 @@ final class GenerateArtworkEmbeddingJob implements ShouldQueue
|
||||
]
|
||||
);
|
||||
|
||||
$this->upsertVectorIndex($vectors, $artwork);
|
||||
$this->upsertVectorIndex(
|
||||
$vectors,
|
||||
$artwork,
|
||||
$normalized
|
||||
);
|
||||
} finally {
|
||||
$this->releaseLock($lockKey);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* @param array<int, float> $vector
|
||||
*/
|
||||
private function upsertVectorIndex(
|
||||
ArtworkVectorIndexService $vectors,
|
||||
Artwork $artwork
|
||||
): void
|
||||
{
|
||||
Artwork $artwork,
|
||||
array $vector
|
||||
): void {
|
||||
if (! $vectors->isConfigured()) {
|
||||
return;
|
||||
}
|
||||
|
||||
try {
|
||||
$vectors->upsertArtwork($artwork);
|
||||
$vectors->upsertArtworkVector(
|
||||
$artwork,
|
||||
$vector
|
||||
);
|
||||
} catch (\Throwable $e) {
|
||||
Log::warning('GenerateArtworkEmbeddingJob vector upsert failed', [
|
||||
'artwork_id' => (int) $artwork->id,
|
||||
@@ -146,8 +210,96 @@ final class GenerateArtworkEmbeddingJob implements ShouldQueue
|
||||
}
|
||||
}
|
||||
|
||||
private function vectorIndexCoversEmbedding(
|
||||
Artwork $artwork,
|
||||
ArtworkEmbedding $embedding
|
||||
): bool {
|
||||
$indexedAt = $artwork->last_vector_indexed_at;
|
||||
$generatedAt = $embedding->generated_at;
|
||||
|
||||
if ($indexedAt === null || $generatedAt === null) {
|
||||
return false;
|
||||
}
|
||||
|
||||
return $indexedAt->greaterThanOrEqualTo(
|
||||
$generatedAt
|
||||
);
|
||||
}
|
||||
|
||||
/**
|
||||
* @param array<int, float> $vector
|
||||
* @return array<int, float>
|
||||
*/
|
||||
private function decodeStoredVector(
|
||||
ArtworkEmbedding $embedding
|
||||
): array {
|
||||
try {
|
||||
$decoded = json_decode(
|
||||
(string) $embedding->embedding_json,
|
||||
true,
|
||||
512,
|
||||
JSON_THROW_ON_ERROR
|
||||
);
|
||||
} catch (\Throwable) {
|
||||
return [];
|
||||
}
|
||||
|
||||
if (! is_array($decoded) || $decoded === []) {
|
||||
return [];
|
||||
}
|
||||
|
||||
$vector = [];
|
||||
|
||||
foreach ($decoded as $value) {
|
||||
if (! is_int($value) && ! is_float($value)) {
|
||||
return [];
|
||||
}
|
||||
|
||||
$value = (float) $value;
|
||||
|
||||
if (! is_finite($value)) {
|
||||
return [];
|
||||
}
|
||||
|
||||
$vector[] = $value;
|
||||
}
|
||||
|
||||
$count = count($vector);
|
||||
|
||||
$minDim = max(
|
||||
1,
|
||||
(int) config(
|
||||
'recommendations.embedding.min_dim',
|
||||
64
|
||||
)
|
||||
);
|
||||
|
||||
$maxDim = max(
|
||||
$minDim,
|
||||
(int) config(
|
||||
'recommendations.embedding.max_dim',
|
||||
4096
|
||||
)
|
||||
);
|
||||
|
||||
if (
|
||||
$count < $minDim
|
||||
|| $count > $maxDim
|
||||
) {
|
||||
return [];
|
||||
}
|
||||
|
||||
if (
|
||||
(int) $embedding->dim > 0
|
||||
&& $count !== (int) $embedding->dim
|
||||
) {
|
||||
return [];
|
||||
}
|
||||
|
||||
return $vector;
|
||||
}
|
||||
|
||||
/**
|
||||
* @param array<int, float> $vector
|
||||
* @return array<int, float>
|
||||
*/
|
||||
private function normalize(array $vector): array
|
||||
@@ -162,12 +314,13 @@ final class GenerateArtworkEmbeddingJob implements ShouldQueue
|
||||
}
|
||||
|
||||
$norm = sqrt($sumSquares);
|
||||
|
||||
return array_map(static fn (float $value): float => $value / $norm, $vector);
|
||||
}
|
||||
|
||||
private function lockKey(int $artworkId, string $model, string $version): string
|
||||
{
|
||||
return 'artwork-embedding:lock:' . $artworkId . ':' . $model . ':' . $version;
|
||||
return 'artwork-embedding:lock:'.$artworkId.':'.$model.':'.$version;
|
||||
}
|
||||
|
||||
private function acquireLock(string $key): bool
|
||||
@@ -177,6 +330,7 @@ final class GenerateArtworkEmbeddingJob implements ShouldQueue
|
||||
if ($didSet) {
|
||||
Redis::expire($key, 1800);
|
||||
}
|
||||
|
||||
return (bool) $didSet;
|
||||
} catch (\Throwable) {
|
||||
return true;
|
||||
|
||||
Reference in New Issue
Block a user