- Updated generated_models.py to reflect OpenAPI 3.1 nullable format changes
- Models now use Optional[type] instead of nullable: true
- All affected models regenerated with datamodel-codegen
- Syntax and linting checks pass
- Convert all nullable: true to OpenAPI 3.1 format using type: [type, 'null']
- Fix invalid array schema definition in ManagerMappings using oneOf
- Add default security: [] configuration to satisfy security-defined rule
- All 41 validation errors resolved, spec now passes with 0 errors
- 141 warnings remain (mostly missing operationId and example validation)
- Add proper async worker management to TaskQueue class
- Remove redundant task_worker_thread and task_worker_lock global variables
- Replace manual threading with async task management
- Update is_processing() logic to use TaskQueue state instead of thread status
- Implement automatic worker cleanup when queue processing completes
- Simplify queue start endpoint to use TaskQueue.start_worker()
- Eliminate TaskQueue.ExecutionStatus NamedTuple in favor of generated TaskExecutionStatus Pydantic model
- Remove manual conversion logic between NamedTuple and Pydantic model
- Use single source of truth for task execution status
- Clean up unused imports (Literal, NamedTuple)
- Maintain consistent data model usage throughout TaskQueue
- Replace deprecated openapi-spec-validator with @redocly/cli
- Remove fragile custom regex-based route alignment script
- Use industry-standard OpenAPI validation tooling
- Switch from Python to Node.js for validation pipeline
- New validation catches 41 errors and 141 warnings that old validator missed
- Updated all POST endpoints to use proper Pydantic model validation:
- `/v2/manager/queue/task` - validates QueueTaskItem
- `/v2/manager/queue/install_model` - validates ModelMetadata
- `/v2/manager/queue/reinstall` - validates InstallPackParams
- `/v2/customnode/import_fail_info` - validates cnr_id/url fields
- Added proper error handling with ValidationError for detailed error messages
- Updated TaskQueue.put() to handle both dict and Pydantic model inputs
- Added missing imports: InstallPackParams, ModelMetadata, ValidationError
Benefits:
- Early validation catches invalid data at API boundaries
- Better error messages for clients with specific validation failures
- Type safety throughout the request processing pipeline
- Consistent validation behavior across all endpoints
All ruff checks pass and validation is now enabled by default.
Enhances ComfyUI Manager with robust batch execution tracking and unified data model architecture:
- Implemented automatic batch history serialization with before/after system state snapshots
- Added comprehensive state management capturing installed nodes, models, and ComfyUI version info
- Enhanced task queue with proper client ID handling and WebSocket notifications
- Migrated all data models to OpenAPI-generated Pydantic models for consistency
- Added documentation for new TaskQueue methods (done_count, total_count, finalize)
- Fixed 64 linting errors with proper imports and code cleanup
Technical improvements:
- All models now auto-generated from openapi.yaml ensuring API/implementation consistency
- Batch tracking captures complete system state at operation start and completion
- Enhanced REST endpoints with comprehensive documentation
- Removed manual model files in favor of single source of truth
- Added helper methods for system state capture and batch lifecycle management
- Add client_id field to QueueTaskItem and TaskHistoryItem models
- Implement client-specific WebSocket message routing
- Add client filtering to queue status and history endpoints
- Follow ComfyUI patterns for session management
- Create data_models package for better code organization
Add a custom node to quickly create empty latents in common resolutions and aspect ratios for SD 1.5, SDXL, Flux, Chroma, and HiDream. Choose from curated presets or generate by axis and aspect ratio. Appears in the 'latent' node group.
