暂存

app/service/ArchiveRepository.php

@@ -2,6 +2,7 @@
 
 namespace app\service;
 
+use Throwable;
 use support\Db;
 
 class ArchiveRepository
@@ -24,8 +25,6 @@ class ArchiveRepository
                     'series' => $archive['series'] ?? null,
                     'tags' => json_encode($archive['tags'] ?? [], JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES),
                     'metadata' => json_encode($archive['metadata'] ?? [], JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES),
-                    'content' => $archive['content'] ?? null,
-                    'raw' => $archive['raw'] ?? null,
                     'chunks' => json_encode($chunkUids, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES),
                 ]
             );
@@ -194,6 +193,79 @@ class ArchiveRepository
         Db::table('archives')->where('archive_uid', $archiveUid)->update($updates);
     }
 
+    public function queuePendingAiMetadataArchives(int $limit): array
+    {
+        $limit = max(1, $limit);
+        $rows = Db::table('archives')
+            ->where(function ($query): void {
+                $query
+                    ->whereNull('title')
+                    ->orWhere('title', '')
+                    ->orWhereNull('summary')
+                    ->orWhere('summary', '')
+                    ->orWhereNull('author')
+                    ->orWhere('author', '')
+                    ->orWhereNull('year')
+                    ->orWhere('year', '<=', 0)
+                    ->orWhereRaw("tags = '[]'::jsonb")
+                    ->orWhereRaw("metadata #>> '{title_source}' = 'fallback'");
+            })
+            ->orderByDesc('updated_time')
+            ->limit(max($limit * 5, 50))
+            ->get()
+            ->all();
+
+        $archiveUids = [];
+        foreach ($rows as $row) {
+            $archive = $this->archiveToArray($row);
+            if (!$this->archiveNeedsMetadata($archive) || !$this->shouldQueueAiMetadata($archive)) {
+                continue;
+            }
+
+            $archiveUids[] = (string) $archive['archive_uid'];
+            $this->markAiMetadataStatus((string) $archive['archive_uid'], 'queued', null);
+
+            if (count($archiveUids) >= $limit) {
+                break;
+            }
+        }
+
+        return $archiveUids;
+    }
+
+    public function markAiMetadataStatus(string $archiveUid, string $status, ?string $error = null, array $extra = []): void
+    {
+        $archive = $this->findArchive($archiveUid);
+        if ($archive === null) {
+            return;
+        }
+
+        $metadata = is_array($archive['metadata'] ?? null) ? $archive['metadata'] : [];
+        $aiMeta = is_array($metadata['ai_enrichment'] ?? null) ? $metadata['ai_enrichment'] : [];
+        $attempt = ((int) ($aiMeta['attempt'] ?? 0)) + ($status === 'processing' ? 1 : 0);
+
+        $metadata['ai_enrichment'] = array_merge($aiMeta, $extra, [
+            'status' => $status,
+            'attempt' => $attempt,
+            'updated_at' => date(DATE_ATOM),
+            'error' => $error,
+        ]);
+
+        Db::table('archives')->where('archive_uid', $archiveUid)->update([
+            'metadata' => json_encode($metadata, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES),
+        ]);
+    }
+
+    public function isAiMetadataRetryable(Throwable $exception): bool
+    {
+        if (!$exception instanceof \app\service\LLM\LLMRequestException) {
+            return true;
+        }
+
+        $statusCode = $exception->statusCode();
+        return $statusCode === null || $statusCode === 429 || $statusCode >= 500;
+    }
+
     public function archiveNeedsMetadata(array $archive): bool
     {
         foreach (['title', 'year', 'author', 'tags', 'summary'] as $field) {
@@ -227,12 +299,29 @@ class ArchiveRepository
             'series' => $archive->series,
             'tags' => json_decode($archive->tags ?? '[]', true) ?: [],
             'metadata' => json_decode($archive->metadata ?? '{}', true) ?: [],
-            'content' => $archive->content,
-            'raw' => $archive->raw,
             'chunks' => json_decode($archive->chunks ?? '[]', true) ?: [],
         ];
     }
 
+    private function shouldQueueAiMetadata(array $archive): bool
+    {
+        $aiMeta = is_array($archive['metadata']['ai_enrichment'] ?? null) ? $archive['metadata']['ai_enrichment'] : [];
+        $status = (string) ($aiMeta['status'] ?? '');
+        if ($status === 'failed_terminal') {
+            return false;
+        }
+
+        $updatedAt = strtotime((string) ($aiMeta['updated_at'] ?? '')) ?: 0;
+        $staleAfter = max(60, (int) config('queue.ai_metadata.stale_after_seconds', 900));
+        $isFresh = $updatedAt > 0 && (time() - $updatedAt) < $staleAfter;
+
+        if (in_array($status, ['queued', 'processing'], true) && $isFresh) {
+            return false;
+        }
+
+        return true;
+    }
+
     private function chunkRowToArray(object $row): array
     {
         return [