@mesh-sync/worker-backend-client 4.0.3 → 4.0.5

package/README.md

@@ -924,7 +924,7 @@ and stores aggregated statistics in Elasticsearch for trend analysis.
 
 ### model-metadata-generation-request
 
-**Description:** Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM analysis with optional vision capabilities. Worker fetches model data from database and uses thumbnails for enhanced analysis when available. Prerequisites: file download, thumbnail generation (optional for vision), and technical metadata analysis (strongly recommended).
+**Description:** Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM vision analysis of thumbnails by default. Worker queries database by modelId to retrieve: Model (fileName, fileSize, ownerId, storageItem), TechnicalMetadata (geometry, quality, printability), and ThumbnailMedia (required for default vision analysis). Prerequisites: file download completed, technical metadata analysis completed (strongly recommended), thumbnail generation completed (REQUIRED for vision-enhanced analysis, default behavior). Vision analysis can be disabled via ENABLE_VISION_METADATA_GENERATION environment variable (default: true).
 
 **Method:** `client.modelMetadataGenerationRequest(data)`
 
@@ -932,17 +932,9 @@ and stores aggregated statistics in Elasticsearch for trend analysis.
 
 **Fields:**
 
-- `modelId` (string) [✗]: UUID of the model to generate metadata for. Worker queries database for: Model.fileName, Model.fileSize, Model.storageItem (path, provider). For vision-enhanced analysis, worker queries thumbnail_media table for thumbnail paths.
+- `modelId` (string) [✗]: UUID of the model to generate metadata for. Worker queries database to retrieve: 1) Model entity (name, fileName, fileSize, ownerId, storageItem relationship), 2) Model.technicalMetadata relationship (vertices, faces, quality scores, printability metrics), 3) AssociatedMedia.thumbnails (via associatedMediaId) for vision-enhanced analysis. All data fetched from PostgreSQL - no additional parameters required.
 
-- `ownerId` (string) [✗]: UUID of the model owner. Used for authorization and audit logging.
-
-- `parentJobId` (string) [✗]: ID of parent job (file-download, thumbnail-generation, or technical-analysis) for BullMQ dependency tracking. Ensures metadata generation runs after prerequisites complete. Should reference thumbnail-generation job when vision analysis is desired.
-
-- `technicalMetadata` (object) [✗]: Technical analysis results (geometry, quality metrics, printability) passed from backend. Worker-core-lib Model schema lacks technical_metadata column, so backend must provide this data. Used for enhanced LLM prompt context. Include: vertexCount, faceCount, surfaceArea, volume, qualityScore, printabilityScore, detectedIssues, etc.
-
-- `thumbnailPath` (string) [✗]: Path to generated thumbnail in MinIO storage (e.g., 'thumbnails/{modelId}/thumbnail.webp'). If provided, enables vision-enhanced metadata generation. Worker can also query thumbnail_media table, but explicit path improves performance.
-
-- `enableVisionAnalysis` (boolean) [✗]: Flag to enable vision-based metadata enrichment using thumbnail images. Requires thumbnailPath or available thumbnail in database. Vision analysis generates more accurate descriptions, tags, and classifications based on visual appearance.
+- `parentJobId` (string) [✗]: ID of parent thumbnail-generation job for BullMQ dependency tracking. SHOULD be set to thumbnail-generation job ID to ensure thumbnails are ready before metadata generation (required for default vision analysis). If omitted, worker will check for thumbnail existence and wait/retry if ENABLE_VISION_METADATA_GENERATION=true.
 
 
 
@@ -1591,7 +1583,7 @@ and stores aggregated statistics in Elasticsearch for trend analysis.
   - Notifies backend that enriched marketplace metadata generation completed. Backend updates Model entity with generated description, tags, classification, etc.
 
 - `modelMetadataGenerationRequest(data: ModelMetadataGenerationRequestMessage): Promise<JobResponse>`
-  - Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM analysis with optional vision capabilities. Worker fetches model data from database and uses thumbnails for enhanced analysis when available. Prerequisites: file download, thumbnail generation (optional for vision), and technical metadata analysis (strongly recommended).
+  - Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM vision analysis of thumbnails by default. Worker queries database by modelId to retrieve: Model (fileName, fileSize, ownerId, storageItem), TechnicalMetadata (geometry, quality, printability), and ThumbnailMedia (required for default vision analysis). Prerequisites: file download completed, technical metadata analysis completed (strongly recommended), thumbnail generation completed (REQUIRED for vision-enhanced analysis, default behavior). Vision analysis can be disabled via ENABLE_VISION_METADATA_GENERATION environment variable (default: true).
 
 - `modelMetamodelDetectionFound(data: ModelMetamodelDetectionFoundMessage): Promise<JobResponse>`
   - Handles model metamodel detection found with hierarchical relationships.

package/dist/client.d.ts

@@ -236,7 +236,7 @@ export declare class WorkerClient {
      */
     modelMetadataGenerationCompleted(data: types.ModelMetadataGenerationCompletedMessage): Promise<JobResponse>;
     /**
-     * Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM analysis with optional vision capabilities. Worker fetches model data from database and uses thumbnails for enhanced analysis when available. Prerequisites: file download, thumbnail generation (optional for vision), and technical metadata analysis (strongly recommended).
+     * Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM vision analysis of thumbnails by default. Worker queries database by modelId to retrieve: Model (fileName, fileSize, ownerId, storageItem), TechnicalMetadata (geometry, quality, printability), and ThumbnailMedia (required for default vision analysis). Prerequisites: file download completed, technical metadata analysis completed (strongly recommended), thumbnail generation completed (REQUIRED for vision-enhanced analysis, default behavior). Vision analysis can be disabled via ENABLE_VISION_METADATA_GENERATION environment variable (default: true).
      * @param data - The message payload
      * @returns Job response with job ID
      */

package/dist/client.js

@@ -313,7 +313,7 @@ export class WorkerClient {
         return this.sendToQueue(types.MessageTypes.MODEL_METADATA_GENERATION_COMPLETED, data);
     }
     /**
-     * Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM analysis with optional vision capabilities. Worker fetches model data from database and uses thumbnails for enhanced analysis when available. Prerequisites: file download, thumbnail generation (optional for vision), and technical metadata analysis (strongly recommended).
+     * Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM vision analysis of thumbnails by default. Worker queries database by modelId to retrieve: Model (fileName, fileSize, ownerId, storageItem), TechnicalMetadata (geometry, quality, printability), and ThumbnailMedia (required for default vision analysis). Prerequisites: file download completed, technical metadata analysis completed (strongly recommended), thumbnail generation completed (REQUIRED for vision-enhanced analysis, default behavior). Vision analysis can be disabled via ENABLE_VISION_METADATA_GENERATION environment variable (default: true).
      * @param data - The message payload
      * @returns Job response with job ID
      */

package/dist/types.d.ts

@@ -679,21 +679,13 @@ export interface ModelMetadataGenerationCompletedMessage {
     metadata: Record<string, string | number | boolean>;
 }
 /**
- * Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM analysis with optional vision capabilities. Worker fetches model data from database and uses thumbnails for enhanced analysis when available. Prerequisites: file download, thumbnail generation (optional for vision), and technical metadata analysis (strongly recommended).
+ * Generates enriched marketplace metadata (SEO descriptions, tags, categories) for 3D models using LLM vision analysis of thumbnails by default. Worker queries database by modelId to retrieve: Model (fileName, fileSize, ownerId, storageItem), TechnicalMetadata (geometry, quality, printability), and ThumbnailMedia (required for default vision analysis). Prerequisites: file download completed, technical metadata analysis completed (strongly recommended), thumbnail generation completed (REQUIRED for vision-enhanced analysis, default behavior). Vision analysis can be disabled via ENABLE_VISION_METADATA_GENERATION environment variable (default: true).
  */
 export interface ModelMetadataGenerationRequestMessage {
-    /** UUID of the model to generate metadata for. Worker queries database for: Model.fileName, Model.fileSize, Model.storageItem (path, provider). For vision-enhanced analysis, worker queries thumbnail_media table for thumbnail paths. */
+    /** UUID of the model to generate metadata for. Worker queries database to retrieve: 1) Model entity (name, fileName, fileSize, ownerId, storageItem relationship), 2) Model.technicalMetadata relationship (vertices, faces, quality scores, printability metrics), 3) AssociatedMedia.thumbnails (via associatedMediaId) for vision-enhanced analysis. All data fetched from PostgreSQL - no additional parameters required. */
     modelId?: string;
-    /** UUID of the model owner. Used for authorization and audit logging. */
-    ownerId?: string;
-    /** ID of parent job (file-download, thumbnail-generation, or technical-analysis) for BullMQ dependency tracking. Ensures metadata generation runs after prerequisites complete. Should reference thumbnail-generation job when vision analysis is desired. */
+    /** ID of parent thumbnail-generation job for BullMQ dependency tracking. SHOULD be set to thumbnail-generation job ID to ensure thumbnails are ready before metadata generation (required for default vision analysis). If omitted, worker will check for thumbnail existence and wait/retry if ENABLE_VISION_METADATA_GENERATION=true. */
     parentJobId?: string;
-    /** Technical analysis results (geometry, quality metrics, printability) passed from backend. Worker-core-lib Model schema lacks technical_metadata column, so backend must provide this data. Used for enhanced LLM prompt context. Include: vertexCount, faceCount, surfaceArea, volume, qualityScore, printabilityScore, detectedIssues, etc. */
-    technicalMetadata?: Record<string, string | number | boolean>;
-    /** Path to generated thumbnail in MinIO storage (e.g., 'thumbnails/{modelId}/thumbnail.webp'). If provided, enables vision-enhanced metadata generation. Worker can also query thumbnail_media table, but explicit path improves performance. */
-    thumbnailPath?: string;
-    /** Flag to enable vision-based metadata enrichment using thumbnail images. Requires thumbnailPath or available thumbnail in database. Vision analysis generates more accurate descriptions, tags, and classifications based on visual appearance. */
-    enableVisionAnalysis?: boolean;
 }
 /**
  * Handles model metamodel detection found with hierarchical relationships.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mesh-sync/worker-backend-client",
-  "version": "4.0.3",
+  "version": "4.0.5",
   "description": "Auto-generated TypeScript client for worker-backend - provides type-safe methods for enqueueing jobs",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",