Bump version to 0.8.6 in pyproject.toml

- Implement WebSocket route for initialization progress updates
- Create initialization component with progress bar and stages
- Add styles for initialization UI
- Update base template to include initialization component
- Enhance model scanner to broadcast progress during initialization

.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,36 @@
+---
+name: Bug report
+about: Create a report to help us improve
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+### **LoRA Manager Version**
+- Version: `vX.X.X`
+
+### **Environment Information**
+- **Operating System**: (e.g., Windows 11, macOS Ventura, Ubuntu 22.04)
+- **Browser & Version**: (e.g., Chrome 120.0.0, Edge 115.0.0)
+
+### **Issue Description**
+- Describe the issue in detail.
+
+### **Steps to Reproduce**
+1. Open LoRA Manager in [your browser].
+2. Perform [specific action].
+3. Observe the issue.
+
+### **Expected Behavior**
+- What did you expect to happen?
+
+### **Screenshots** *(If applicable)*
+- Upload screenshots or screen recordings.
+
+### **Logs**
+- Provide the **ComfyUI startup log** and any relevant error messages.
+- Check the browser developer console (F12 → Console tab) and attach any errors.
+
+### **Additional Context** *(Optional)*
+- Any other relevant details.

.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,20 @@
+---
+name: Feature request
+about: Suggest an idea for this project
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+**Is your feature request related to a problem? Please describe.**
+A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
+
+**Describe the solution you'd like**
+A clear and concise description of what you want to happen.
+
+**Describe alternatives you've considered**
+A clear and concise description of any alternative solutions or features you've considered.
+
+**Additional context**
+Add any other context or screenshots about the feature request here.

.gitignore

@@ -1,2 +1,5 @@
 __pycache__/
-settings.json
+settings.json
+output/*
+py/run_test.py
+.vscode/

README.md

@@ -1,48 +1,69 @@
 # ComfyUI LoRA Manager
 
-A web-based management interface designed to help you organize and manage your local LoRA models in ComfyUI. Access the interface at: `http://localhost:8188/loras`
+> **Revolutionize your workflow with the ultimate LoRA companion for ComfyUI!**
 
-![Interface Preview](https://github.com/willmiao/ComfyUI-Lora-Manager/blob/main/static/images/Screenshot%202025-01-27%20172349.png)
+[![Discord](https://img.shields.io/discord/1346296675538571315?color=7289DA&label=Discord&logo=discord&logoColor=white)](https://discord.gg/vcqNrWVFvM)
+[![Release](https://img.shields.io/github/v/release/willmiao/ComfyUI-Lora-Manager?include_prereleases&color=blue&logo=github)](https://github.com/willmiao/ComfyUI-Lora-Manager/releases)
+[![Release Date](https://img.shields.io/github/release-date/willmiao/ComfyUI-Lora-Manager?color=green&logo=github)](https://github.com/willmiao/ComfyUI-Lora-Manager/releases)
+
+A comprehensive toolset that streamlines organizing, downloading, and applying LoRA models in ComfyUI. With powerful features like recipe management, checkpoint organization, and one-click workflow integration, working with models becomes faster, smoother, and significantly easier. Access the interface at: `http://localhost:8188/loras`
+
+![Interface Preview](https://github.com/willmiao/ComfyUI-Lora-Manager/blob/main/static/images/screenshot.png)
 
 ## 📺 Tutorial: One-Click LoRA Integration
 Watch this quick tutorial to learn how to use the new one-click LoRA integration feature:
 
 [![One-Click LoRA Integration Tutorial](https://img.youtube.com/vi/qS95OjX3e70/0.jpg)](https://youtu.be/qS95OjX3e70)
+[![LoRA Manager v0.8.0 - New Recipe Feature & Bulk Operations](https://img.youtube.com/vi/noN7f_ER7yo/0.jpg)](https://youtu.be/noN7f_ER7yo)
 
 ---
 
 ## Release Notes
 
-### v0.7.35-beta
-* Added base model filtering
-* Implemented bulk operations (copy syntax, move multiple LoRAs)
-* Added ability to edit LoRA model names in details view
-* Added update checker with notification system
-* Added support modal for user feedback and community links
+### v0.8.6 Major Update
+* **Checkpoint Management** - Added comprehensive management for model checkpoints including scanning, searching, filtering, and deletion
+* **Enhanced Metadata Support** - New capabilities for retrieving and managing checkpoint metadata with improved operations
+* **Improved Initial Loading** - Optimized cache initialization with visual progress indicators for better user experience
 
-### v0.7.33
-* Enhanced LoRA Loader node with visual strength adjustment widgets
-* Added toggle switches for LoRA enable/disable
-* Implemented image tooltips for LoRA preview
-* Added TriggerWord Toggle node with visual word selection
-* Fixed various bugs and improved stability
+### v0.8.5
+* **Enhanced LoRA & Recipe Connectivity** - Added Recipes tab in LoRA details to see all recipes using a specific LoRA
+* **Improved Navigation** - New shortcuts to jump between related LoRAs and Recipes with one-click navigation
+* **Video Preview Controls** - Added "Autoplay Videos on Hover" setting to optimize performance and reduce resource usage
+* **UI Experience Refinements** - Smoother transitions between related content pages
 
-### v0.7.3
-* Added "Lora Loader (LoraManager)" custom node for workflows
-* Implemented one-click LoRA integration
-* Added direct copying of LoRA syntax from manager interface
-* Added automatic preset strength value application
-* Added automatic trigger word loading
+### v0.8.4
+* **Node Layout Improvements** - Fixed layout issues with LoRA Loader and Trigger Words Toggle nodes in newer ComfyUI frontend versions
+* **Recipe LoRA Reconnection** - Added ability to reconnect deleted LoRAs in recipes by clicking the "deleted" badge in recipe details
+* **Bug Fixes & Stability** - Resolved various issues for improved reliability
 
-### v0.7.0
-* Added direct CivitAI integration for downloading LoRAs
-* Implemented version selection for model downloads
-* Added target folder selection for downloads
-* Added context menu with quick actions
-* Added force refresh for CivitAI data
-* Implemented LoRA movement between folders
-* Added personal usage tips and notes for LoRAs
-* Improved performance for details window
+### v0.8.3
+* **Enhanced Workflow Parser** - Rebuilt workflow analysis engine with improved support for ComfyUI core nodes and easier extensibility
+* **Improved Recipe System** - Refined the experimental Save Recipe functionality with better workflow integration
+* **New Save Image Node** - Added experimental node with metadata support for perfect CivitAI compatibility
+  * Supports dynamic filename prefixes with variables [1](https://github.com/nkchocoai/ComfyUI-SaveImageWithMetaData?tab=readme-ov-file#filename_prefix)
+* **Default LoRA Root Setting** - Added configuration option for setting your preferred LoRA directory
+
+### v0.8.2  
+* **Faster Initialization for Forge Users** - Improved first-run efficiency by utilizing existing `.json` and `.civitai.info` files from Forge’s CivitAI helper extension, making migration smoother.  
+* **LoRA Filename Editing** - Added support for renaming LoRA files directly within LoRA Manager.  
+* **Recipe Editing** - Users can now edit recipe names and tags.  
+* **Retain Deleted LoRAs in Recipes** - Deleted LoRAs will remain listed in recipes, allowing future functionality to reconnect them once re-obtained.  
+* **Download Missing LoRAs from Recipes** - Easily fetch missing LoRAs associated with a recipe.
+
+### v0.8.1
+* **Base Model Correction** - Added support for modifying base model associations to fix incorrect metadata for non-CivitAI LoRAs
+* **LoRA Loader Flexibility** - Made CLIP input optional for model-only workflows like Hunyuan video generation
+* **Expanded Recipe Support** - Added compatibility with 3 additional recipe metadata formats
+* **Enhanced Showcase Images** - Generation parameters now displayed alongside LoRA preview images
+* **UI Improvements & Bug Fixes** - Various interface refinements and stability enhancements
+
+### v0.8.0
+* **Introduced LoRA Recipes** - Create, import, save, and share your favorite LoRA combinations
+* **Recipe Management System** - Easily browse, search, and organize your LoRA recipes
+* **Workflow Integration** - Save recipes directly from your workflow with generation parameters preserved
+* **Simplified Workflow Application** - Quickly apply saved recipes to new projects
+* **Enhanced UI & UX** - Improved interface design and user experience
+* **Bug Fixes & Stability** - Resolved various issues and enhanced overall performance
 
 [View Update History](./update_logs.md)
 
@@ -76,11 +97,24 @@ Watch this quick tutorial to learn how to use the new one-click LoRA integration
   - Trigger words at a glance
   - One-click workflow integration with preset values
   
+- 🔄 **Checkpoint Management**
+  - Scan and organize checkpoint models
+  - Filter and search your collection
+  - View and edit metadata
+  - Clean up and manage disk space
+  
+- 🧩 **LoRA Recipes**
+  - Save and share favorite LoRA combinations
+  - Preserve generation parameters for future reference
+  - Quick application to workflows
+  - Import/export functionality for community sharing
+  
 - 💻 **User Friendly**
   - One-click access from ComfyUI menu
   - Context menu for quick actions
   - Custom notes and usage tips
   - Multi-folder support
+  - Visual progress indicators during initialization
 
 ---
 
@@ -121,6 +155,15 @@ pip install requirements.txt
 
 ---
 
+## Credits
+
+This project has been inspired by and benefited from other excellent ComfyUI extensions:
+
+- [ComfyUI-SaveImageWithMetaData](https://github.com/Comfy-Community/ComfyUI-SaveImageWithMetaData) - For the image metadata functionality
+- [rgthree-comfy](https://github.com/rgthree/rgthree-comfy) - For the lora loader functionality
+
+---
+
 ## Contributing
 
 If you have suggestions, bug reports, or improvements, feel free to open an issue or contribute directly to the codebase. Pull requests are always welcome!
@@ -139,12 +182,3 @@ Join our Discord community for support, discussions, and updates:
 [Discord Server](https://discord.gg/vcqNrWVFvM)
 
 ---
-
-## 🗺️ Roadmap
-
-- ✅ One-click integration of LoRAs into ComfyUI workflows with preset strength values
-- 🤝 Improved usage tips retrieval from CivitAI model pages
-- 🔌 Integration with Power LoRA Loader and other management tools
-- 🛡️ Configurable NSFW level settings for content filtering
-
----

__init__.py

@@ -1,14 +1,18 @@
 from .py.lora_manager import LoraManager
 from .py.nodes.lora_loader import LoraManagerLoader
 from .py.nodes.trigger_word_toggle import TriggerWordToggle
+from .py.nodes.lora_stacker import LoraStacker
+from .py.nodes.save_image import SaveImage
 
 NODE_CLASS_MAPPINGS = {
     LoraManagerLoader.NAME: LoraManagerLoader,
-    TriggerWordToggle.NAME: TriggerWordToggle
+    TriggerWordToggle.NAME: TriggerWordToggle,
+    LoraStacker.NAME: LoraStacker,
+    SaveImage.NAME: SaveImage
 }
 
 WEB_DIRECTORY = "./web/comfyui"
 
 # Register routes on import
 LoraManager.add_routes()
-__all__ = ['NODE_CLASS_MAPPINGS', 'WEB_DIRECTORY']
+__all__ = ['NODE_CLASS_MAPPINGS', 'WEB_DIRECTORY']

py/config.py

@@ -17,6 +17,8 @@ class Config:
         # 静态路由映射字典, target to route mapping
         self._route_mappings = {}
         self.loras_roots = self._init_lora_paths()
+        self.checkpoints_roots = self._init_checkpoint_paths()
+        self.temp_directory = folder_paths.get_temp_directory()
         # 在初始化时扫描符号链接
         self._scan_symbolic_links()
 
@@ -38,9 +40,12 @@ class Config:
             return False
 
     def _scan_symbolic_links(self):
-        """扫描所有 LoRA 根目录中的符号链接"""
+        """扫描所有 LoRA 和 Checkpoint 根目录中的符号链接"""
         for root in self.loras_roots:
             self._scan_directory_links(root)
+        
+        for root in self.checkpoints_roots:
+            self._scan_directory_links(root)
 
     def _scan_directory_links(self, root: str):
         """递归扫描目录中的符号链接"""
@@ -72,7 +77,7 @@ class Config:
         """添加静态路由映射"""
         normalized_path = os.path.normpath(path).replace(os.sep, '/')
         self._route_mappings[normalized_path] = route
-        logger.info(f"Added route mapping: {normalized_path} -> {route}")
+        # logger.info(f"Added route mapping: {normalized_path} -> {route}")
 
     def map_path_to_link(self, path: str) -> str:
         """将目标路径映射回符号链接路径"""
@@ -84,12 +89,23 @@ class Config:
                 mapped_path = normalized_path.replace(target_path, link_path, 1)
                 return mapped_path
         return path
+    
+    def map_link_to_path(self, link_path: str) -> str:
+        """将符号链接路径映射回实际路径"""
+        normalized_link = os.path.normpath(link_path).replace(os.sep, '/')
+        # 检查路径是否包含在任何映射的目标路径中
+        for target_path, link_path in self._path_mappings.items():
+            if normalized_link.startswith(target_path):
+                # 如果路径以目标路径开头，则替换为实际路径
+                mapped_path = normalized_link.replace(target_path, link_path, 1)
+                return mapped_path
+        return link_path
 
     def _init_lora_paths(self) -> List[str]:
         """Initialize and validate LoRA paths from ComfyUI settings"""
-        paths = list(set(path.replace(os.sep, "/") 
+        paths = sorted(set(path.replace(os.sep, "/") 
                 for path in folder_paths.get_folder_paths("loras") 
-                if os.path.exists(path)))
+                if os.path.exists(path)), key=lambda p: p.lower())
         print("Found LoRA roots:", "\n - " + "\n - ".join(paths))
         
         if not paths:
@@ -103,6 +119,35 @@ class Config:
         
         return paths
 
+    def _init_checkpoint_paths(self) -> List[str]:
+        """Initialize and validate checkpoint paths from ComfyUI settings"""
+        # Get checkpoint paths from folder_paths
+        checkpoint_paths = folder_paths.get_folder_paths("checkpoints")
+        diffusion_paths = folder_paths.get_folder_paths("diffusers")
+        unet_paths = folder_paths.get_folder_paths("unet")
+        
+        # Combine all checkpoint-related paths
+        all_paths = checkpoint_paths + diffusion_paths + unet_paths
+        
+        # Filter and normalize paths
+        paths = sorted(set(path.replace(os.sep, "/") 
+                for path in all_paths 
+                if os.path.exists(path)), key=lambda p: p.lower())
+        
+        print("Found checkpoint roots:", paths)
+        
+        if not paths:
+            logger.warning("No valid checkpoint folders found in ComfyUI configuration")
+            return []
+        
+        # 初始化路径映射，与 LoRA 路径处理方式相同
+        for path in paths:
+            real_path = os.path.normpath(os.path.realpath(path)).replace(os.sep, '/')
+            if real_path != path:
+                self.add_path_mapping(path, real_path)
+        
+        return paths
+
     def get_preview_static_url(self, preview_path: str) -> str:
         """Convert local preview path to static URL"""
         if not preview_path:

py/lora_manager.py

@@ -1,12 +1,11 @@
 import asyncio
-import os
 from server import PromptServer # type: ignore
 from .config import config
 from .routes.lora_routes import LoraRoutes
 from .routes.api_routes import ApiRoutes
-from .services.lora_scanner import LoraScanner
-from .services.file_monitor import LoraFileMonitor
-from .services.lora_cache import LoraCache
+from .routes.recipe_routes import RecipeRoutes
+from .routes.checkpoints_routes import CheckpointsRoutes
+from .services.service_registry import ServiceRegistry
 import logging
 
 logger = logging.getLogger(__name__)
@@ -19,7 +18,7 @@ class LoraManager:
         """Initialize and register all routes"""
         app = PromptServer.instance.app
 
-        added_targets = set()  # 用于跟踪已添加的目标路径
+        added_targets = set()  # Track already added target paths
         
         # Add static routes for each lora root
         for idx, root in enumerate(config.loras_roots, start=1):
@@ -31,77 +30,143 @@ class LoraManager:
                     if link == root:
                         real_root = target
                         break
-            # 为原始路径添加静态路由
+            # Add static route for original path
             app.router.add_static(preview_path, real_root)
             logger.info(f"Added static route {preview_path} -> {real_root}")
             
-            # 记录路由映射
+            # Record route mapping
             config.add_route_mapping(real_root, preview_path)
             added_targets.add(real_root)
         
-        # 为符号链接的目标路径添加额外的静态路由
-        link_idx = 1
+        # Add static routes for each checkpoint root
+        for idx, root in enumerate(config.checkpoints_roots, start=1):
+            preview_path = f'/checkpoints_static/root{idx}/preview'
+            
+            real_root = root
+            if root in config._path_mappings.values():
+                for target, link in config._path_mappings.items():
+                    if link == root:
+                        real_root = target
+                        break
+            # Add static route for original path
+            app.router.add_static(preview_path, real_root)
+            logger.info(f"Added static route {preview_path} -> {real_root}")
+            
+            # Record route mapping
+            config.add_route_mapping(real_root, preview_path)
+            added_targets.add(real_root)
+        
+        # Add static routes for symlink target paths
+        link_idx = {
+            'lora': 1,
+            'checkpoint': 1
+        }
         
         for target_path, link_path in config._path_mappings.items():
             if target_path not in added_targets:
-                route_path = f'/loras_static/link_{link_idx}/preview'
+                # Determine if this is a checkpoint or lora link based on path
+                is_checkpoint = any(cp_root in link_path for cp_root in config.checkpoints_roots)
+                is_checkpoint = is_checkpoint or any(cp_root in target_path for cp_root in config.checkpoints_roots)
+                
+                if is_checkpoint:
+                    route_path = f'/checkpoints_static/link_{link_idx["checkpoint"]}/preview'
+                    link_idx["checkpoint"] += 1
+                else:
+                    route_path = f'/loras_static/link_{link_idx["lora"]}/preview'
+                    link_idx["lora"] += 1
+                
                 app.router.add_static(route_path, target_path)
                 logger.info(f"Added static route for link target {route_path} -> {target_path}")
                 config.add_route_mapping(target_path, route_path)
                 added_targets.add(target_path)
-                link_idx += 1
         
         # Add static route for plugin assets
         app.router.add_static('/loras_static', config.static_path)
         
         # Setup feature routes
-        routes = LoraRoutes()
+        lora_routes = LoraRoutes()
+        checkpoints_routes = CheckpointsRoutes()
         
-        # Setup file monitoring
-        monitor = LoraFileMonitor(routes.scanner, config.loras_roots)
-        monitor.start()
+        # Initialize routes
+        lora_routes.setup_routes(app)
+        checkpoints_routes.setup_routes(app)
+        ApiRoutes.setup_routes(app)
+        RecipeRoutes.setup_routes(app)
         
-        routes.setup_routes(app)
-        ApiRoutes.setup_routes(app, monitor)
-        
-        # Store monitor in app for cleanup
-        app['lora_monitor'] = monitor
-        
-        # Schedule cache initialization using the application's startup handler
-        app.on_startup.append(lambda app: cls._schedule_cache_init(routes.scanner))
+        # Schedule service initialization 
+        app.on_startup.append(lambda app: cls._initialize_services())
         
         # Add cleanup
         app.on_shutdown.append(cls._cleanup)
         app.on_shutdown.append(ApiRoutes.cleanup)
     
     @classmethod
-    async def _schedule_cache_init(cls, scanner: LoraScanner):
-        """Schedule cache initialization in the running event loop"""
+    async def _initialize_services(cls):
+        """Initialize all services using the ServiceRegistry"""
         try:
-            # 创建低优先级的初始化任务
-            asyncio.create_task(cls._initialize_cache(scanner), name='lora_cache_init')
-        except Exception as e:
-            print(f"LoRA Manager: Error scheduling cache initialization: {e}")
-    
-    @classmethod
-    async def _initialize_cache(cls, scanner: LoraScanner):
-        """Initialize cache in background"""
-        try:
-            # 设置初始缓存占位
-            scanner._cache = LoraCache(
-                raw_data=[],
-                sorted_by_name=[],
-                sorted_by_date=[],
-                folders=[]
-            )
+            logger.info("LoRA Manager: Initializing services via ServiceRegistry")
             
-            # 分阶段加载缓存
-            await scanner.get_cached_data(force_refresh=True)
+            # Initialize CivitaiClient first to ensure it's ready for other services
+            civitai_client = await ServiceRegistry.get_civitai_client()
+            
+            # Get file monitors through ServiceRegistry
+            lora_monitor = await ServiceRegistry.get_lora_monitor()
+            checkpoint_monitor = await ServiceRegistry.get_checkpoint_monitor()
+            
+            # Start monitors
+            lora_monitor.start()
+            logger.info("Lora monitor started")
+            
+            # Make sure checkpoint monitor has paths before starting
+            await checkpoint_monitor.initialize_paths()
+            checkpoint_monitor.start()
+            logger.info("Checkpoint monitor started")
+
+            # Register DownloadManager with ServiceRegistry
+            download_manager = await ServiceRegistry.get_download_manager()
+            
+            # Initialize WebSocket manager
+            ws_manager = await ServiceRegistry.get_websocket_manager()
+            
+            # Initialize scanners in background
+            lora_scanner = await ServiceRegistry.get_lora_scanner()
+            checkpoint_scanner = await ServiceRegistry.get_checkpoint_scanner()
+            
+            # Initialize recipe scanner if needed
+            recipe_scanner = await ServiceRegistry.get_recipe_scanner()
+            
+            # Create low-priority initialization tasks
+            asyncio.create_task(lora_scanner.initialize_in_background(), name='lora_cache_init')
+            asyncio.create_task(checkpoint_scanner.initialize_in_background(), name='checkpoint_cache_init')
+            asyncio.create_task(recipe_scanner.initialize_in_background(), name='recipe_cache_init')
+            
+            logger.info("LoRA Manager: All services initialized and background tasks scheduled")
+                
         except Exception as e:
-            print(f"LoRA Manager: Error initializing cache: {e}")
+            logger.error(f"LoRA Manager: Error initializing services: {e}", exc_info=True)
     
     @classmethod
     async def _cleanup(cls, app):
-        """Cleanup resources"""
-        if 'lora_monitor' in app:
-            app['lora_monitor'].stop()
+        """Cleanup resources using ServiceRegistry"""
+        try:
+            logger.info("LoRA Manager: Cleaning up services")
+            
+            # Get monitors from ServiceRegistry
+            lora_monitor = await ServiceRegistry.get_service("lora_monitor")
+            if lora_monitor:
+                lora_monitor.stop()
+                logger.info("Stopped LoRA monitor")
+                
+            checkpoint_monitor = await ServiceRegistry.get_service("checkpoint_monitor")
+            if checkpoint_monitor:
+                checkpoint_monitor.stop()
+                logger.info("Stopped checkpoint monitor")
+                
+            # Close CivitaiClient gracefully
+            civitai_client = await ServiceRegistry.get_service("civitai_client")
+            if civitai_client:
+                await civitai_client.close()
+                logger.info("Closed CivitaiClient connection")
+                
+        except Exception as e:
+            logger.error(f"Error during cleanup: {e}", exc_info=True)

py/nodes/lora_loader.py

@@ -1,3 +1,4 @@
+import logging
 from nodes import LoraLoader
 from comfy.comfy_types import IO # type: ignore
 from ..services.lora_scanner import LoraScanner
@@ -6,16 +7,18 @@ import asyncio
 import os
 from .utils import FlexibleOptionalInputType, any_type
 
+logger = logging.getLogger(__name__)
+
 class LoraManagerLoader:
     NAME = "Lora Loader (LoraManager)"
-    CATEGORY = "loaders"
+    CATEGORY = "Lora Manager/loaders"
 
     @classmethod
     def INPUT_TYPES(cls):
         return {
             "required": {
                 "model": ("MODEL",),
-                "clip": ("CLIP",),
+                # "clip": ("CLIP",),
                 "text": (IO.STRING, {
                     "multiline": True, 
                     "dynamicPrompts": True, 
@@ -26,8 +29,8 @@ class LoraManagerLoader:
             "optional": FlexibleOptionalInputType(any_type),
         }
 
-    RETURN_TYPES = ("MODEL", "CLIP", IO.STRING)
-    RETURN_NAMES = ("MODEL", "CLIP", "trigger_words")
+    RETURN_TYPES = ("MODEL", "CLIP", IO.STRING, IO.STRING)
+    RETURN_NAMES = ("MODEL", "CLIP", "trigger_words", "loaded_loras")
     FUNCTION = "load_loras"
 
     async def get_lora_info(self, lora_name):
@@ -49,29 +52,73 @@ class LoraManagerLoader:
                             return relative_path, trigger_words
         return lora_name, []  # Fallback if not found
 
-    def load_loras(self, model, clip, text, **kwargs):
-        """Loads multiple LoRAs based on the kwargs input."""
+    def extract_lora_name(self, lora_path):
+        """Extract the lora name from a lora path (e.g., 'IL\\aorunIllstrious.safetensors' -> 'aorunIllstrious')"""
+        # Get the basename without extension
+        basename = os.path.basename(lora_path)
+        return os.path.splitext(basename)[0]
+    
+    def _get_loras_list(self, kwargs):
+        """Helper to extract loras list from either old or new kwargs format"""
+        if 'loras' not in kwargs:
+            return []
+            
+        loras_data = kwargs['loras']
+        # Handle new format: {'loras': {'__value__': [...]}}
+        if isinstance(loras_data, dict) and '__value__' in loras_data:
+            return loras_data['__value__']
+        # Handle old format: {'loras': [...]}
+        elif isinstance(loras_data, list):
+            return loras_data
+        # Unexpected format
+        else:
+            logger.warning(f"Unexpected loras format: {type(loras_data)}")
+            return []
+    
+    def load_loras(self, model, text, **kwargs):
+        """Loads multiple LoRAs based on the kwargs input and lora_stack."""
         loaded_loras = []
         all_trigger_words = []
         
-        if 'loras' in kwargs:
-            for lora in kwargs['loras']:
-                if not lora.get('active', False):
-                    continue
-                    
-                lora_name = lora['name']
-                strength = float(lora['strength'])
+        clip = kwargs.get('clip', None)
+        lora_stack = kwargs.get('lora_stack', None)
+        # First process lora_stack if available
+        if lora_stack:
+            for lora_path, model_strength, clip_strength in lora_stack:
+                # Apply the LoRA using the provided path and strengths
+                model, clip = LoraLoader().load_lora(model, clip, lora_path, model_strength, clip_strength)
                 
-                # Get lora path and trigger words
-                lora_path, trigger_words = asyncio.run(self.get_lora_info(lora_name))
+                # Extract lora name for trigger words lookup
+                lora_name = self.extract_lora_name(lora_path)
+                _, trigger_words = asyncio.run(self.get_lora_info(lora_name))
                 
-                # Apply the LoRA using the resolved path
-                model, clip = LoraLoader().load_lora(model, clip, lora_path, strength, strength)
-                loaded_loras.append(f"{lora_name}: {strength}")
-                
-                # Add trigger words to collection
                 all_trigger_words.extend(trigger_words)
+                loaded_loras.append(f"{lora_name}: {model_strength}")
         
-        trigger_words_text = ", ".join(all_trigger_words) if all_trigger_words else ""
+        # Then process loras from kwargs with support for both old and new formats
+        loras_list = self._get_loras_list(kwargs)
+        for lora in loras_list:
+            if not lora.get('active', False):
+                continue
+                
+            lora_name = lora['name']
+            strength = float(lora['strength'])
+            
+            # Get lora path and trigger words
+            lora_path, trigger_words = asyncio.run(self.get_lora_info(lora_name))
+            
+            # Apply the LoRA using the resolved path
+            model, clip = LoraLoader().load_lora(model, clip, lora_path, strength, strength)
+            loaded_loras.append(f"{lora_name}: {strength}")
+            
+            # Add trigger words to collection
+            all_trigger_words.extend(trigger_words)
+        
+        # use ',, ' to separate trigger words for group mode
+        trigger_words_text = ",, ".join(all_trigger_words) if all_trigger_words else ""
+        
+        # Format loaded_loras as <lora:lora_name:strength> separated by spaces
+        formatted_loras = " ".join([f"<lora:{name.split(':')[0].strip()}:{str(strength).strip()}>" 
+                                  for name, strength in [item.split(':') for item in loaded_loras]])
 
-        return (model, clip, trigger_words_text)
+        return (model, clip, trigger_words_text, formatted_loras)

py/nodes/lora_stacker.py

@@ -0,0 +1,118 @@
+from comfy.comfy_types import IO # type: ignore
+from ..services.lora_scanner import LoraScanner
+from ..config import config
+import asyncio
+import os
+from .utils import FlexibleOptionalInputType, any_type
+import logging
+
+logger = logging.getLogger(__name__)
+
+class LoraStacker:
+    NAME = "Lora Stacker (LoraManager)"
+    CATEGORY = "Lora Manager/stackers"
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "text": (IO.STRING, {
+                    "multiline": True, 
+                    "dynamicPrompts": True, 
+                    "tooltip": "Format: <lora:lora_name:strength> separated by spaces or punctuation",
+                    "placeholder": "LoRA syntax input: <lora:name:strength>"
+                }),
+            },
+            "optional": FlexibleOptionalInputType(any_type),
+        }
+
+    RETURN_TYPES = ("LORA_STACK", IO.STRING, IO.STRING)
+    RETURN_NAMES = ("LORA_STACK", "trigger_words", "active_loras")
+    FUNCTION = "stack_loras"
+
+    async def get_lora_info(self, lora_name):
+        """Get the lora path and trigger words from cache"""
+        scanner = await LoraScanner.get_instance()
+        cache = await scanner.get_cached_data()
+        
+        for item in cache.raw_data:
+            if item.get('file_name') == lora_name:
+                file_path = item.get('file_path')
+                if file_path:
+                    for root in config.loras_roots:
+                        root = root.replace(os.sep, '/')
+                        if file_path.startswith(root):
+                            relative_path = os.path.relpath(file_path, root).replace(os.sep, '/')
+                            # Get trigger words from civitai metadata
+                            civitai = item.get('civitai', {})
+                            trigger_words = civitai.get('trainedWords', []) if civitai else []
+                            return relative_path, trigger_words
+        return lora_name, []  # Fallback if not found
+
+    def extract_lora_name(self, lora_path):
+        """Extract the lora name from a lora path (e.g., 'IL\\aorunIllstrious.safetensors' -> 'aorunIllstrious')"""
+        # Get the basename without extension
+        basename = os.path.basename(lora_path)
+        return os.path.splitext(basename)[0]
+    
+    def _get_loras_list(self, kwargs):
+        """Helper to extract loras list from either old or new kwargs format"""
+        if 'loras' not in kwargs:
+            return []
+            
+        loras_data = kwargs['loras']
+        # Handle new format: {'loras': {'__value__': [...]}}
+        if isinstance(loras_data, dict) and '__value__' in loras_data:
+            return loras_data['__value__']
+        # Handle old format: {'loras': [...]}
+        elif isinstance(loras_data, list):
+            return loras_data
+        # Unexpected format
+        else:
+            logger.warning(f"Unexpected loras format: {type(loras_data)}")
+            return []
+    
+    def stack_loras(self, text, **kwargs):
+        """Stacks multiple LoRAs based on the kwargs input without loading them."""
+        stack = []
+        active_loras = []
+        all_trigger_words = []
+        
+        # Process existing lora_stack if available
+        lora_stack = kwargs.get('lora_stack', None)
+        if lora_stack:
+            stack.extend(lora_stack)
+            # Get trigger words from existing stack entries
+            for lora_path, _, _ in lora_stack:
+                lora_name = self.extract_lora_name(lora_path)
+                _, trigger_words = asyncio.run(self.get_lora_info(lora_name))
+                all_trigger_words.extend(trigger_words)
+        
+        # Process loras from kwargs with support for both old and new formats
+        loras_list = self._get_loras_list(kwargs)
+        for lora in loras_list:
+            if not lora.get('active', False):
+                continue
+                
+            lora_name = lora['name']
+            model_strength = float(lora['strength'])
+            clip_strength = model_strength  # Using same strength for both as in the original loader
+            
+            # Get lora path and trigger words
+            lora_path, trigger_words = asyncio.run(self.get_lora_info(lora_name))
+            
+            # Add to stack without loading
+            # replace '/' with os.sep to avoid different OS path format
+            stack.append((lora_path.replace('/', os.sep), model_strength, clip_strength))
+            active_loras.append((lora_name, model_strength))
+            
+            # Add trigger words to collection
+            all_trigger_words.extend(trigger_words)
+        
+        # use ',, ' to separate trigger words for group mode
+        trigger_words_text = ",, ".join(all_trigger_words) if all_trigger_words else ""
+        # Format active_loras as <lora:lora_name:strength> separated by spaces
+        active_loras_text = " ".join([f"<lora:{name}:{str(strength).strip()}>" 
+                                  for name, strength in active_loras])
+
+        return (stack, trigger_words_text, active_loras_text)

py/nodes/save_image.py

@@ -0,0 +1,375 @@
+import json
+import os
+import asyncio
+import re
+import numpy as np
+import folder_paths # type: ignore
+from ..services.lora_scanner import LoraScanner
+from ..workflow.parser import WorkflowParser
+from PIL import Image, PngImagePlugin
+import piexif
+from io import BytesIO
+
+class SaveImage:
+    NAME = "Save Image (LoraManager)"
+    CATEGORY = "Lora Manager/utils"
+    DESCRIPTION = "Save images with embedded generation metadata in compatible format"
+
+    def __init__(self):
+        self.output_dir = folder_paths.get_output_directory()
+        self.type = "output"
+        self.prefix_append = ""
+        self.compress_level = 4
+        self.counter = 0
+    
+    # Add pattern format regex for filename substitution
+    pattern_format = re.compile(r"(%[^%]+%)")
+    
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "images": ("IMAGE",),
+                "filename_prefix": ("STRING", {"default": "ComfyUI"}),
+                "file_format": (["png", "jpeg", "webp"],),
+            },
+            "optional": {
+                "custom_prompt": ("STRING", {"default": "", "forceInput": True}),
+                "lossless_webp": ("BOOLEAN", {"default": True}),
+                "quality": ("INT", {"default": 100, "min": 1, "max": 100}),
+                "embed_workflow": ("BOOLEAN", {"default": False}),
+                "add_counter_to_filename": ("BOOLEAN", {"default": True}),
+            },
+            "hidden": {
+                "prompt": "PROMPT",
+                "extra_pnginfo": "EXTRA_PNGINFO",
+            },
+        }
+
+    RETURN_TYPES = ("IMAGE",)
+    RETURN_NAMES = ("images",)
+    FUNCTION = "process_image"
+    OUTPUT_NODE = True
+
+    async def get_lora_hash(self, lora_name):
+        """Get the lora hash from cache"""
+        scanner = await LoraScanner.get_instance()
+        cache = await scanner.get_cached_data()
+        
+        for item in cache.raw_data:
+            if item.get('file_name') == lora_name:
+                return item.get('sha256')
+        return None
+
+    async def format_metadata(self, parsed_workflow, custom_prompt=None):
+        """Format metadata in the requested format similar to userComment example"""
+        if not parsed_workflow:
+            return ""
+        
+        # Extract the prompt and negative prompt
+        prompt = parsed_workflow.get('prompt', '')
+        negative_prompt = parsed_workflow.get('negative_prompt', '')
+        
+        # Override prompt with custom_prompt if provided
+        if custom_prompt:
+            prompt = custom_prompt
+        
+        # Extract loras from the prompt if present
+        loras_text = parsed_workflow.get('loras', '')
+        lora_hashes = {}
+        
+        # If loras are found, add them on a new line after the prompt
+        if loras_text:
+            prompt_with_loras = f"{prompt}\n{loras_text}"
+            
+            # Extract lora names from the format <lora:name:strength>
+            lora_matches = re.findall(r'<lora:([^:]+):([^>]+)>', loras_text)
+            
+            # Get hash for each lora
+            for lora_name, strength in lora_matches:
+                hash_value = await self.get_lora_hash(lora_name)
+                if hash_value:
+                    lora_hashes[lora_name] = hash_value
+        else:
+            prompt_with_loras = prompt
+        
+        # Format the first part (prompt and loras)
+        metadata_parts = [prompt_with_loras]
+        
+        # Add negative prompt
+        if negative_prompt:
+            metadata_parts.append(f"Negative prompt: {negative_prompt}")
+        
+        # Format the second part (generation parameters)
+        params = []
+        
+        # Add standard parameters in the correct order
+        if 'steps' in parsed_workflow:
+            params.append(f"Steps: {parsed_workflow.get('steps')}")
+        
+        if 'sampler' in parsed_workflow:
+            sampler = parsed_workflow.get('sampler')
+            # Convert ComfyUI sampler names to user-friendly names
+            sampler_mapping = {
+                'euler': 'Euler',
+                'euler_ancestral': 'Euler a',
+                'dpm_2': 'DPM2',
+                'dpm_2_ancestral': 'DPM2 a',
+                'heun': 'Heun',
+                'dpm_fast': 'DPM fast',
+                'dpm_adaptive': 'DPM adaptive',
+                'lms': 'LMS',
+                'dpmpp_2s_ancestral': 'DPM++ 2S a',
+                'dpmpp_sde': 'DPM++ SDE',
+                'dpmpp_sde_gpu': 'DPM++ SDE',
+                'dpmpp_2m': 'DPM++ 2M',
+                'dpmpp_2m_sde': 'DPM++ 2M SDE',
+                'dpmpp_2m_sde_gpu': 'DPM++ 2M SDE',
+                'ddim': 'DDIM'
+            }
+            sampler_name = sampler_mapping.get(sampler, sampler)
+            params.append(f"Sampler: {sampler_name}")
+        
+        if 'scheduler' in parsed_workflow:
+            scheduler = parsed_workflow.get('scheduler')
+            scheduler_mapping = {
+                'normal': 'Simple',
+                'karras': 'Karras',
+                'exponential': 'Exponential',
+                'sgm_uniform': 'SGM Uniform',
+                'sgm_quadratic': 'SGM Quadratic'
+            }
+            scheduler_name = scheduler_mapping.get(scheduler, scheduler)
+            params.append(f"Schedule type: {scheduler_name}")
+        
+        # CFG scale (cfg in parsed_workflow)
+        if 'cfg_scale' in parsed_workflow:
+            params.append(f"CFG scale: {parsed_workflow.get('cfg_scale')}")
+        elif 'cfg' in parsed_workflow:
+            params.append(f"CFG scale: {parsed_workflow.get('cfg')}")
+        
+        # Seed
+        if 'seed' in parsed_workflow:
+            params.append(f"Seed: {parsed_workflow.get('seed')}")
+        
+        # Size
+        if 'size' in parsed_workflow:
+            params.append(f"Size: {parsed_workflow.get('size')}")
+        
+        # Model info
+        if 'checkpoint' in parsed_workflow:
+            # Extract basename without path
+            checkpoint = os.path.basename(parsed_workflow.get('checkpoint', ''))
+            # Remove extension if present
+            checkpoint = os.path.splitext(checkpoint)[0]
+            params.append(f"Model: {checkpoint}")
+        
+        # Add LoRA hashes if available
+        if lora_hashes:
+            lora_hash_parts = []
+            for lora_name, hash_value in lora_hashes.items():
+                lora_hash_parts.append(f"{lora_name}: {hash_value}")
+            
+            if lora_hash_parts:
+                params.append(f"Lora hashes: \"{', '.join(lora_hash_parts)}\"")
+        
+        # Combine all parameters with commas
+        metadata_parts.append(", ".join(params))
+        
+        # Join all parts with a new line
+        return "\n".join(metadata_parts)
+
+    # credit to nkchocoai
+    # Add format_filename method to handle pattern substitution
+    def format_filename(self, filename, parsed_workflow):
+        """Format filename with metadata values"""
+        if not parsed_workflow:
+            return filename
+            
+        result = re.findall(self.pattern_format, filename)
+        for segment in result:
+            parts = segment.replace("%", "").split(":")
+            key = parts[0]
+            
+            if key == "seed" and 'seed' in parsed_workflow:
+                filename = filename.replace(segment, str(parsed_workflow.get('seed', '')))
+            elif key == "width" and 'size' in parsed_workflow:
+                size = parsed_workflow.get('size', 'x')
+                w = size.split('x')[0] if isinstance(size, str) else size[0]
+                filename = filename.replace(segment, str(w))
+            elif key == "height" and 'size' in parsed_workflow:
+                size = parsed_workflow.get('size', 'x')
+                h = size.split('x')[1] if isinstance(size, str) else size[1]
+                filename = filename.replace(segment, str(h))
+            elif key == "pprompt" and 'prompt' in parsed_workflow:
+                prompt = parsed_workflow.get('prompt', '').replace("\n", " ")
+                if len(parts) >= 2:
+                    length = int(parts[1])
+                    prompt = prompt[:length]
+                filename = filename.replace(segment, prompt.strip())
+            elif key == "nprompt" and 'negative_prompt' in parsed_workflow:
+                prompt = parsed_workflow.get('negative_prompt', '').replace("\n", " ")
+                if len(parts) >= 2:
+                    length = int(parts[1])
+                    prompt = prompt[:length]
+                filename = filename.replace(segment, prompt.strip())
+            elif key == "model" and 'checkpoint' in parsed_workflow:
+                model = parsed_workflow.get('checkpoint', '')
+                model = os.path.splitext(os.path.basename(model))[0]
+                if len(parts) >= 2:
+                    length = int(parts[1])
+                    model = model[:length]
+                filename = filename.replace(segment, model)
+            elif key == "date":
+                from datetime import datetime
+                now = datetime.now()
+                date_table = {
+                    "yyyy": str(now.year),
+                    "MM": str(now.month).zfill(2),
+                    "dd": str(now.day).zfill(2),
+                    "hh": str(now.hour).zfill(2),
+                    "mm": str(now.minute).zfill(2),
+                    "ss": str(now.second).zfill(2),
+                }
+                if len(parts) >= 2:
+                    date_format = parts[1]
+                    for k, v in date_table.items():
+                        date_format = date_format.replace(k, v)
+                    filename = filename.replace(segment, date_format)
+                else:
+                    date_format = "yyyyMMddhhmmss"
+                    for k, v in date_table.items():
+                        date_format = date_format.replace(k, v)
+                    filename = filename.replace(segment, date_format)
+                    
+        return filename
+
+    def save_images(self, images, filename_prefix, file_format, prompt=None, extra_pnginfo=None, 
+                   lossless_webp=True, quality=100, embed_workflow=False, add_counter_to_filename=True,
+                   custom_prompt=None):
+        """Save images with metadata"""
+        results = []
+        
+        # Parse the workflow using the WorkflowParser
+        parser = WorkflowParser()
+        if prompt:
+            parsed_workflow = parser.parse_workflow(prompt)
+        else:
+            parsed_workflow = {}
+            
+        # Get or create metadata asynchronously
+        metadata = asyncio.run(self.format_metadata(parsed_workflow, custom_prompt))
+        
+        # Process filename_prefix with pattern substitution
+        filename_prefix = self.format_filename(filename_prefix, parsed_workflow)
+        
+        # Get initial save path info once for the batch
+        full_output_folder, filename, counter, subfolder, processed_prefix = folder_paths.get_save_image_path(
+            filename_prefix, self.output_dir, images[0].shape[1], images[0].shape[0]
+        )
+        
+        # Create directory if it doesn't exist
+        if not os.path.exists(full_output_folder):
+            os.makedirs(full_output_folder, exist_ok=True)
+        
+        # Process each image with incrementing counter
+        for i, image in enumerate(images):
+            # Convert the tensor image to numpy array
+            img = 255. * image.cpu().numpy()
+            img = Image.fromarray(np.clip(img, 0, 255).astype(np.uint8))
+            
+            # Generate filename with counter if needed
+            base_filename = filename
+            if add_counter_to_filename:
+                # Use counter + i to ensure unique filenames for all images in batch
+                current_counter = counter + i
+                base_filename += f"_{current_counter:05}"
+                
+            # Set file extension and prepare saving parameters
+            if file_format == "png":
+                file = base_filename + ".png"
+                file_extension = ".png"
+                save_kwargs = {"optimize": True, "compress_level": self.compress_level}
+                pnginfo = PngImagePlugin.PngInfo()
+            elif file_format == "jpeg":
+                file = base_filename + ".jpg"
+                file_extension = ".jpg"
+                save_kwargs = {"quality": quality, "optimize": True}
+            elif file_format == "webp":
+                file = base_filename + ".webp" 
+                file_extension = ".webp"
+                save_kwargs = {"quality": quality, "lossless": lossless_webp}
+            
+            # Full save path
+            file_path = os.path.join(full_output_folder, file)
+            
+            # Save the image with metadata
+            try:
+                if file_format == "png":
+                    if metadata:
+                        pnginfo.add_text("parameters", metadata)
+                    if embed_workflow and extra_pnginfo is not None:
+                        workflow_json = json.dumps(extra_pnginfo["workflow"])
+                        pnginfo.add_text("workflow", workflow_json)
+                    save_kwargs["pnginfo"] = pnginfo
+                    img.save(file_path, format="PNG", **save_kwargs)
+                elif file_format == "jpeg":
+                    # For JPEG, use piexif
+                    if metadata:
+                        try:
+                            exif_dict = {'Exif': {piexif.ExifIFD.UserComment: b'UNICODE\0' + metadata.encode('utf-16be')}}
+                            exif_bytes = piexif.dump(exif_dict)
+                            save_kwargs["exif"] = exif_bytes
+                        except Exception as e:
+                            print(f"Error adding EXIF data: {e}")
+                    img.save(file_path, format="JPEG", **save_kwargs)
+                elif file_format == "webp":
+                    # For WebP, also use piexif for metadata
+                    if metadata:
+                        try:
+                            exif_dict = {'Exif': {piexif.ExifIFD.UserComment: b'UNICODE\0' + metadata.encode('utf-16be')}}
+                            exif_bytes = piexif.dump(exif_dict)
+                            save_kwargs["exif"] = exif_bytes
+                        except Exception as e:
+                            print(f"Error adding EXIF data: {e}")
+                    img.save(file_path, format="WEBP", **save_kwargs)
+                
+                results.append({
+                    "filename": file,
+                    "subfolder": subfolder,
+                    "type": self.type
+                })
+                
+            except Exception as e:
+                print(f"Error saving image: {e}")
+        
+        return results
+
+    def process_image(self, images, filename_prefix="ComfyUI", file_format="png", prompt=None, extra_pnginfo=None,
+                     lossless_webp=True, quality=100, embed_workflow=False, add_counter_to_filename=True,
+                     custom_prompt=""):
+        """Process and save image with metadata"""
+        # Make sure the output directory exists
+        os.makedirs(self.output_dir, exist_ok=True)
+        
+        # Ensure images is always a list of images
+        if len(images.shape) == 3:  # Single image (height, width, channels)
+            images = [images]
+        else:  # Multiple images (batch, height, width, channels)
+            images = [img for img in images]
+        
+        # Save all images
+        results = self.save_images(
+            images, 
+            filename_prefix, 
+            file_format, 
+            prompt, 
+            extra_pnginfo,
+            lossless_webp,
+            quality,
+            embed_workflow,
+            add_counter_to_filename,
+            custom_prompt if custom_prompt.strip() else None
+        )
+        
+        return (images,)

py/nodes/trigger_word_toggle.py

@@ -1,17 +1,22 @@
 import json
+import re
 from server import PromptServer # type: ignore
 from .utils import FlexibleOptionalInputType, any_type
+import logging
+
+logger = logging.getLogger(__name__)
+
 
 class TriggerWordToggle:
     NAME = "TriggerWord Toggle (LoraManager)"
-    CATEGORY = "lora manager"
+    CATEGORY = "Lora Manager/utils"
     DESCRIPTION = "Toggle trigger words on/off"
     
     @classmethod
     def INPUT_TYPES(cls):
         return {
             "required": {
-                "trigger_words": ("STRING", {"defaultInput": True, "forceInput": True}),
+                "group_mode": ("BOOLEAN", {"default": True}),
             },
             "optional": FlexibleOptionalInputType(any_type),
             "hidden": {
@@ -23,7 +28,24 @@ class TriggerWordToggle:
     RETURN_NAMES = ("filtered_trigger_words",)
     FUNCTION = "process_trigger_words"
 
-    def process_trigger_words(self, trigger_words, id, **kwargs):
+    def _get_toggle_data(self, kwargs, key='toggle_trigger_words'):
+        """Helper to extract data from either old or new kwargs format"""
+        if key not in kwargs:
+            return None
+            
+        data = kwargs[key]
+        # Handle new format: {'key': {'__value__': ...}}
+        if isinstance(data, dict) and '__value__' in data:
+            return data['__value__']
+        # Handle old format: {'key': ...}
+        else:
+            return data
+
+    def process_trigger_words(self, id, group_mode, **kwargs):
+        # Handle both old and new formats for trigger_words
+        trigger_words_data = self._get_toggle_data(kwargs, 'trigger_words')
+        trigger_words = trigger_words_data if isinstance(trigger_words_data, str) else ""
+        
         # Send trigger words to frontend
         PromptServer.instance.send_sync("trigger_word_update", {
             "id": id,
@@ -32,31 +54,43 @@ class TriggerWordToggle:
         
         filtered_triggers = trigger_words
         
-        if 'toggle_trigger_words' in kwargs:
+        # Get toggle data with support for both formats
+        trigger_data = self._get_toggle_data(kwargs, 'toggle_trigger_words')
+        if trigger_data:
             try:
-                # Get trigger word toggle data
-                trigger_data = kwargs['toggle_trigger_words']
-                
                 # Convert to list if it's a JSON string
                 if isinstance(trigger_data, str):
                     trigger_data = json.loads(trigger_data)
                 
-                # Create dictionaries to track active state of words
+                # Create dictionaries to track active state of words or groups
                 active_state = {item['text']: item.get('active', False) for item in trigger_data}
                 
-                # Split original trigger words
-                original_words = [word.strip() for word in trigger_words.split(',')]
-                
-                # Filter words: keep those not in toggle_trigger_words or those that are active
-                filtered_words = [word for word in original_words if word not in active_state or active_state[word]]
-                
-                # Join them in the same format as input
-                if filtered_words:
-                    filtered_triggers = ', '.join(filtered_words)
+                if group_mode:
+                    # Split by two or more consecutive commas to get groups
+                    groups = re.split(r',{2,}', trigger_words)
+                    # Remove leading/trailing whitespace from each group
+                    groups = [group.strip() for group in groups]
+                    
+                    # Filter groups: keep those not in toggle_trigger_words or those that are active
+                    filtered_groups = [group for group in groups if group not in active_state or active_state[group]]
+                    
+                    if filtered_groups:
+                        filtered_triggers = ', '.join(filtered_groups)
+                    else:
+                        filtered_triggers = ""
                 else:
-                    filtered_triggers = ""
+                    # Original behavior for individual words mode
+                    original_words = [word.strip() for word in trigger_words.split(',')]
+                    # Filter out empty strings
+                    original_words = [word for word in original_words if word]
+                    filtered_words = [word for word in original_words if word not in active_state or active_state[word]]
+                    
+                    if filtered_words:
+                        filtered_triggers = ', '.join(filtered_words)
+                    else:
+                        filtered_triggers = ""
                     
             except Exception as e:
-                print(f"Error processing trigger words: {e}")
+                logger.error(f"Error processing trigger words: {e}")
             
         return (filtered_triggers,)

py/nodes/utils.py

@@ -4,6 +4,7 @@ class AnyType(str):
   def __ne__(self, __value: object) -> bool:
     return False
 
+# Credit to Regis Gaughan, III (rgthree)
 class FlexibleOptionalInputType(dict):
   """A special class to make flexible nodes that pass data to our python handlers.
 

py/routes/checkpoints_routes.py

@@ -0,0 +1,678 @@
+import os
+import json
+import jinja2
+from aiohttp import web
+import logging
+import asyncio
+
+from ..utils.routes_common import ModelRouteUtils
+from ..utils.constants import NSFW_LEVELS
+from ..services.websocket_manager import ws_manager
+from ..services.service_registry import ServiceRegistry
+from ..config import config
+from ..services.settings_manager import settings
+from ..utils.utils import fuzzy_match
+
+logger = logging.getLogger(__name__)
+
+class CheckpointsRoutes:
+    """API routes for checkpoint management"""
+    
+    def __init__(self):
+        self.scanner = None  # Will be initialized in setup_routes
+        self.template_env = jinja2.Environment(
+            loader=jinja2.FileSystemLoader(config.templates_path),
+            autoescape=True
+        )
+        self.download_manager = None  # Will be initialized in setup_routes
+        self._download_lock = asyncio.Lock()
+        
+    async def initialize_services(self):
+        """Initialize services from ServiceRegistry"""
+        self.scanner = await ServiceRegistry.get_checkpoint_scanner()
+        self.download_manager = await ServiceRegistry.get_download_manager()
+        
+    def setup_routes(self, app):
+        """Register routes with the aiohttp app"""
+        # Schedule service initialization on app startup
+        app.on_startup.append(lambda _: self.initialize_services())
+        
+        app.router.add_get('/checkpoints', self.handle_checkpoints_page)
+        app.router.add_get('/api/checkpoints', self.get_checkpoints)
+        app.router.add_post('/api/checkpoints/fetch-all-civitai', self.fetch_all_civitai)
+        app.router.add_get('/api/checkpoints/base-models', self.get_base_models)
+        app.router.add_get('/api/checkpoints/top-tags', self.get_top_tags)
+        app.router.add_get('/api/checkpoints/scan', self.scan_checkpoints)
+        app.router.add_get('/api/checkpoints/info/{name}', self.get_checkpoint_info)
+        app.router.add_get('/api/checkpoints/roots', self.get_checkpoint_roots)
+        app.router.add_get('/api/checkpoints/civitai/versions/{model_id}', self.get_civitai_versions)  # Add new route
+        
+        # Add new routes for model management similar to LoRA routes
+        app.router.add_post('/api/checkpoints/delete', self.delete_model)
+        app.router.add_post('/api/checkpoints/fetch-civitai', self.fetch_civitai)
+        app.router.add_post('/api/checkpoints/replace-preview', self.replace_preview)
+        app.router.add_post('/api/checkpoints/download', self.download_checkpoint)
+        app.router.add_post('/api/checkpoints/save-metadata', self.save_metadata) # Add new route
+        
+        # Add new WebSocket endpoint for checkpoint progress
+        app.router.add_get('/ws/checkpoint-progress', ws_manager.handle_checkpoint_connection)
+
+    async def get_checkpoints(self, request):
+        """Get paginated checkpoint data"""
+        try:
+            # Parse query parameters
+            page = int(request.query.get('page', '1'))
+            page_size = min(int(request.query.get('page_size', '20')), 100)
+            sort_by = request.query.get('sort_by', 'name')
+            folder = request.query.get('folder', None)
+            search = request.query.get('search', None)
+            fuzzy_search = request.query.get('fuzzy_search', 'false').lower() == 'true'
+            base_models = request.query.getall('base_model', [])
+            tags = request.query.getall('tag', [])
+            
+            # Process search options
+            search_options = {
+                'filename': request.query.get('search_filename', 'true').lower() == 'true',
+                'modelname': request.query.get('search_modelname', 'true').lower() == 'true',
+                'tags': request.query.get('search_tags', 'false').lower() == 'true',
+                'recursive': request.query.get('recursive', 'false').lower() == 'true',
+            }
+            
+            # Process hash filters if provided
+            hash_filters = {}
+            if 'hash' in request.query:
+                hash_filters['single_hash'] = request.query['hash']
+            elif 'hashes' in request.query:
+                try:
+                    hash_list = json.loads(request.query['hashes'])
+                    if isinstance(hash_list, list):
+                        hash_filters['multiple_hashes'] = hash_list
+                except (json.JSONDecodeError, TypeError):
+                    pass
+            
+            # Get data from scanner
+            result = await self.get_paginated_data(
+                page=page,
+                page_size=page_size,
+                sort_by=sort_by,
+                folder=folder,
+                search=search,
+                fuzzy_search=fuzzy_search,
+                base_models=base_models,
+                tags=tags,
+                search_options=search_options,
+                hash_filters=hash_filters
+            )
+            
+            # Format response items
+            formatted_result = {
+                'items': [self._format_checkpoint_response(cp) for cp in result['items']],
+                'total': result['total'],
+                'page': result['page'],
+                'page_size': result['page_size'],
+                'total_pages': result['total_pages']
+            }
+            
+            # Return as JSON
+            return web.json_response(formatted_result)
+            
+        except Exception as e:
+            logger.error(f"Error in get_checkpoints: {e}", exc_info=True)
+            return web.json_response({"error": str(e)}, status=500)
+
+    async def get_paginated_data(self, page, page_size, sort_by='name', 
+                               folder=None, search=None, fuzzy_search=False,
+                               base_models=None, tags=None,
+                               search_options=None, hash_filters=None):
+        """Get paginated and filtered checkpoint data"""
+        cache = await self.scanner.get_cached_data()
+
+        # Get default search options if not provided
+        if search_options is None:
+            search_options = {
+                'filename': True,
+                'modelname': True,
+                'tags': False,
+                'recursive': False,
+            }
+
+        # Get the base data set
+        filtered_data = cache.sorted_by_date if sort_by == 'date' else cache.sorted_by_name
+        
+        # Apply hash filtering if provided (highest priority)
+        if hash_filters:
+            single_hash = hash_filters.get('single_hash')
+            multiple_hashes = hash_filters.get('multiple_hashes')
+            
+            if single_hash:
+                # Filter by single hash
+                single_hash = single_hash.lower()  # Ensure lowercase for matching
+                filtered_data = [
+                    cp for cp in filtered_data
+                    if cp.get('sha256', '').lower() == single_hash
+                ]
+            elif multiple_hashes:
+                # Filter by multiple hashes
+                hash_set = set(hash.lower() for hash in multiple_hashes)  # Convert to set for faster lookup
+                filtered_data = [
+                    cp for cp in filtered_data
+                    if cp.get('sha256', '').lower() in hash_set
+                ]
+            
+            # Jump to pagination
+            total_items = len(filtered_data)
+            start_idx = (page - 1) * page_size
+            end_idx = min(start_idx + page_size, total_items)
+            
+            result = {
+                'items': filtered_data[start_idx:end_idx],
+                'total': total_items,
+                'page': page,
+                'page_size': page_size,
+                'total_pages': (total_items + page_size - 1) // page_size
+            }
+            
+            return result
+        
+        # Apply SFW filtering if enabled in settings
+        if settings.get('show_only_sfw', False):
+            filtered_data = [
+                cp for cp in filtered_data
+                if not cp.get('preview_nsfw_level') or cp.get('preview_nsfw_level') < NSFW_LEVELS['R']
+            ]
+        
+        # Apply folder filtering
+        if folder is not None:
+            if search_options.get('recursive', False):
+                # Recursive folder filtering - include all subfolders
+                filtered_data = [
+                    cp for cp in filtered_data
+                    if cp['folder'].startswith(folder)
+                ]
+            else:
+                # Exact folder filtering
+                filtered_data = [
+                    cp for cp in filtered_data
+                    if cp['folder'] == folder
+                ]
+        
+        # Apply base model filtering
+        if base_models and len(base_models) > 0:
+            filtered_data = [
+                cp for cp in filtered_data
+                if cp.get('base_model') in base_models
+            ]
+        
+        # Apply tag filtering
+        if tags and len(tags) > 0:
+            filtered_data = [
+                cp for cp in filtered_data
+                if any(tag in cp.get('tags', []) for tag in tags)
+            ]
+        
+        # Apply search filtering
+        if search:
+            search_results = []
+            
+            for cp in filtered_data:
+                # Search by file name
+                if search_options.get('filename', True):
+                    if fuzzy_search:
+                        if fuzzy_match(cp.get('file_name', ''), search):
+                            search_results.append(cp)
+                            continue
+                    elif search.lower() in cp.get('file_name', '').lower():
+                        search_results.append(cp)
+                        continue
+                
+                # Search by model name
+                if search_options.get('modelname', True):
+                    if fuzzy_search:
+                        if fuzzy_match(cp.get('model_name', ''), search):
+                            search_results.append(cp)
+                            continue
+                    elif search.lower() in cp.get('model_name', '').lower():
+                        search_results.append(cp)
+                        continue
+                
+                # Search by tags
+                if search_options.get('tags', False) and 'tags' in cp:
+                    if any((fuzzy_match(tag, search) if fuzzy_search else search.lower() in tag.lower()) for tag in cp['tags']):
+                        search_results.append(cp)
+                        continue
+            
+            filtered_data = search_results
+
+        # Calculate pagination
+        total_items = len(filtered_data)
+        start_idx = (page - 1) * page_size
+        end_idx = min(start_idx + page_size, total_items)
+        
+        result = {
+            'items': filtered_data[start_idx:end_idx],
+            'total': total_items,
+            'page': page,
+            'page_size': page_size,
+            'total_pages': (total_items + page_size - 1) // page_size
+        }
+        
+        return result
+
+    def _format_checkpoint_response(self, checkpoint):
+        """Format checkpoint data for API response"""
+        return {
+            "model_name": checkpoint["model_name"],
+            "file_name": checkpoint["file_name"],
+            "preview_url": config.get_preview_static_url(checkpoint.get("preview_url", "")),
+            "preview_nsfw_level": checkpoint.get("preview_nsfw_level", 0),
+            "base_model": checkpoint.get("base_model", ""),
+            "folder": checkpoint["folder"],
+            "sha256": checkpoint.get("sha256", ""),
+            "file_path": checkpoint["file_path"].replace(os.sep, "/"),
+            "file_size": checkpoint.get("size", 0),
+            "modified": checkpoint.get("modified", ""),
+            "tags": checkpoint.get("tags", []),
+            "modelDescription": checkpoint.get("modelDescription", ""),
+            "from_civitai": checkpoint.get("from_civitai", True),
+            "notes": checkpoint.get("notes", ""),
+            "model_type": checkpoint.get("model_type", "checkpoint"),
+            "civitai": ModelRouteUtils.filter_civitai_data(checkpoint.get("civitai", {}))
+        }
+    
+    async def fetch_all_civitai(self, request: web.Request) -> web.Response:
+        """Fetch CivitAI metadata for all checkpoints in the background"""
+        try:
+            cache = await self.scanner.get_cached_data()
+            total = len(cache.raw_data)
+            processed = 0
+            success = 0
+            needs_resort = False
+            
+            # Prepare checkpoints to process
+            to_process = [
+                cp for cp in cache.raw_data 
+                if cp.get('sha256') and (not cp.get('civitai') or 'id' not in cp.get('civitai')) and cp.get('from_civitai', True)
+            ]
+            total_to_process = len(to_process)
+            
+            # Send initial progress
+            await ws_manager.broadcast({
+                'status': 'started',
+                'total': total_to_process,
+                'processed': 0,
+                'success': 0
+            })
+            
+            # Process each checkpoint
+            for cp in to_process:
+                try:
+                    original_name = cp.get('model_name')
+                    if await ModelRouteUtils.fetch_and_update_model(
+                        sha256=cp['sha256'],
+                        file_path=cp['file_path'],
+                        model_data=cp,
+                        update_cache_func=self.scanner.update_single_model_cache
+                    ):
+                        success += 1
+                        if original_name != cp.get('model_name'):
+                            needs_resort = True
+                    
+                    processed += 1
+                    
+                    # Send progress update
+                    await ws_manager.broadcast({
+                        'status': 'processing',
+                        'total': total_to_process,
+                        'processed': processed,
+                        'success': success,
+                        'current_name': cp.get('model_name', 'Unknown')
+                    })
+                    
+                except Exception as e:
+                    logger.error(f"Error fetching CivitAI data for {cp['file_path']}: {e}")
+            
+            if needs_resort:
+                await cache.resort(name_only=True)
+            
+            # Send completion message
+            await ws_manager.broadcast({
+                'status': 'completed',
+                'total': total_to_process,
+                'processed': processed,
+                'success': success
+            })
+                    
+            return web.json_response({
+                "success": True,
+                "message": f"Successfully updated {success} of {processed} processed checkpoints (total: {total})"
+            })
+            
+        except Exception as e:
+            # Send error message
+            await ws_manager.broadcast({
+                'status': 'error',
+                'error': str(e)
+            })
+            logger.error(f"Error in fetch_all_civitai for checkpoints: {e}")
+            return web.Response(text=str(e), status=500)
+        
+    async def get_top_tags(self, request: web.Request) -> web.Response:
+        """Handle request for top tags sorted by frequency"""
+        try:
+            # Parse query parameters
+            limit = int(request.query.get('limit', '20'))
+            
+            # Validate limit
+            if limit < 1 or limit > 100:
+                limit = 20  # Default to a reasonable limit
+                
+            # Get top tags
+            top_tags = await self.scanner.get_top_tags(limit)
+            
+            return web.json_response({
+                'success': True,
+                'tags': top_tags
+            })
+            
+        except Exception as e:
+            logger.error(f"Error getting top tags: {str(e)}", exc_info=True)
+            return web.json_response({
+                'success': False,
+                'error': 'Internal server error'
+            }, status=500)
+        
+    async def get_base_models(self, request: web.Request) -> web.Response:
+        """Get base models used in loras"""
+        try:
+            # Parse query parameters
+            limit = int(request.query.get('limit', '20'))
+            
+            # Validate limit
+            if limit < 1 or limit > 100:
+                limit = 20  # Default to a reasonable limit
+                
+            # Get base models
+            base_models = await self.scanner.get_base_models(limit)
+            
+            return web.json_response({
+                'success': True,
+                'base_models': base_models
+            })
+        except Exception as e:
+            logger.error(f"Error retrieving base models: {e}")
+            return web.json_response({
+                'success': False,
+                'error': str(e)
+            }, status=500)
+
+    async def scan_checkpoints(self, request):
+        """Force a rescan of checkpoint files"""
+        try:
+            await self.scanner.get_cached_data(force_refresh=True)
+            return web.json_response({"status": "success", "message": "Checkpoint scan completed"})
+        except Exception as e:
+            logger.error(f"Error in scan_checkpoints: {e}", exc_info=True)
+            return web.json_response({"error": str(e)}, status=500)
+
+    async def get_checkpoint_info(self, request):
+        """Get detailed information for a specific checkpoint by name"""
+        try:
+            name = request.match_info.get('name', '')
+            checkpoint_info = await self.scanner.get_checkpoint_info_by_name(name)
+            
+            if checkpoint_info:
+                return web.json_response(checkpoint_info)
+            else:
+                return web.json_response({"error": "Checkpoint not found"}, status=404)
+                
+        except Exception as e:
+            logger.error(f"Error in get_checkpoint_info: {e}", exc_info=True)
+            return web.json_response({"error": str(e)}, status=500)
+
+    async def handle_checkpoints_page(self, request: web.Request) -> web.Response:
+        """Handle GET /checkpoints request"""
+        try:
+            # Check if the CheckpointScanner is initializing
+            # It's initializing if the cache object doesn't exist yet,
+            # OR if the scanner explicitly says it's initializing (background task running).
+            is_initializing = (
+                self.scanner._cache is None or
+                (hasattr(self.scanner, '_is_initializing') and self.scanner._is_initializing)
+            )
+
+            if is_initializing:
+                # If still initializing, return loading page
+                template = self.template_env.get_template('checkpoints.html')
+                rendered = template.render(
+                    folders=[],  # 空文件夹列表
+                    is_initializing=True,  # 新增标志
+                    settings=settings,  # Pass settings to template
+                    request=request  # Pass the request object to the template
+                )
+                
+                logger.info("Checkpoints page is initializing, returning loading page")
+            else:
+                # 正常流程 - 获取已经初始化好的缓存数据
+                try:
+                    cache = await self.scanner.get_cached_data(force_refresh=False)
+                    template = self.template_env.get_template('checkpoints.html')
+                    rendered = template.render(
+                        folders=cache.folders,
+                        is_initializing=False,
+                        settings=settings,  # Pass settings to template
+                        request=request  # Pass the request object to the template
+                    )
+                except Exception as cache_error:
+                    logger.error(f"Error loading checkpoints cache data: {cache_error}")
+                    # 如果获取缓存失败，也显示初始化页面
+                    template = self.template_env.get_template('checkpoints.html')
+                    rendered = template.render(
+                        folders=[],
+                        is_initializing=True,
+                        settings=settings,
+                        request=request
+                    )
+                    logger.info("Checkpoints cache error, returning initialization page")
+            
+            return web.Response(
+                text=rendered,
+                content_type='text/html'
+            )
+        except Exception as e:
+            logger.error(f"Error handling checkpoints request: {e}", exc_info=True)
+            return web.Response(
+                text="Error loading checkpoints page",
+                status=500
+            )
+
+    async def delete_model(self, request: web.Request) -> web.Response:
+        """Handle checkpoint model deletion request"""
+        return await ModelRouteUtils.handle_delete_model(request, self.scanner)
+    
+    async def fetch_civitai(self, request: web.Request) -> web.Response:
+        """Handle CivitAI metadata fetch request for checkpoints"""
+        return await ModelRouteUtils.handle_fetch_civitai(request, self.scanner)
+    
+    async def replace_preview(self, request: web.Request) -> web.Response:
+        """Handle preview image replacement for checkpoints"""
+        return await ModelRouteUtils.handle_replace_preview(request, self.scanner)
+
+    async def download_checkpoint(self, request: web.Request) -> web.Response:
+        """Handle checkpoint download request"""
+        async with self._download_lock:
+            # Get the download manager from service registry if not already initialized
+            if self.download_manager is None:
+                self.download_manager = await ServiceRegistry.get_download_manager()
+            
+            try:
+                data = await request.json()
+                
+                # Create progress callback that uses checkpoint-specific WebSocket
+                async def progress_callback(progress):
+                    await ws_manager.broadcast_checkpoint_progress({
+                        'status': 'progress',
+                        'progress': progress
+                    })
+                
+                # Check which identifier is provided
+                download_url = data.get('download_url')
+                model_hash = data.get('model_hash')
+                model_version_id = data.get('model_version_id')
+                
+                # Validate that at least one identifier is provided
+                if not any([download_url, model_hash, model_version_id]):
+                    return web.Response(
+                        status=400, 
+                        text="Missing required parameter: Please provide either 'download_url', 'hash', or 'modelVersionId'"
+                    )
+                
+                result = await self.download_manager.download_from_civitai(
+                    download_url=download_url,
+                    model_hash=model_hash,
+                    model_version_id=model_version_id,
+                    save_dir=data.get('checkpoint_root'),
+                    relative_path=data.get('relative_path', ''),
+                    progress_callback=progress_callback,
+                    model_type="checkpoint"
+                )
+                
+                if not result.get('success', False):
+                    error_message = result.get('error', 'Unknown error')
+                    
+                    # Return 401 for early access errors
+                    if 'early access' in error_message.lower():
+                        logger.warning(f"Early access download failed: {error_message}")
+                        return web.Response(
+                            status=401,
+                            text=f"Early Access Restriction: {error_message}"
+                        )
+                    
+                    return web.Response(status=500, text=error_message)
+                
+                return web.json_response(result)
+                
+            except Exception as e:
+                error_message = str(e)
+                
+                # Check if this might be an early access error
+                if '401' in error_message:
+                    logger.warning(f"Early access error (401): {error_message}")
+                    return web.Response(
+                        status=401,
+                        text="Early Access Restriction: This model requires purchase. Please ensure you have purchased early access and are logged in to Civitai."
+                    )
+                
+                logger.error(f"Error downloading checkpoint: {error_message}")
+                return web.Response(status=500, text=error_message)
+
+    async def get_checkpoint_roots(self, request):
+        """Return the checkpoint root directories"""
+        try:
+            if self.scanner is None:
+                self.scanner = await ServiceRegistry.get_checkpoint_scanner()
+                
+            roots = self.scanner.get_model_roots()
+            return web.json_response({
+                "success": True,
+                "roots": roots
+            })
+        except Exception as e:
+            logger.error(f"Error getting checkpoint roots: {e}", exc_info=True)
+            return web.json_response({
+                "success": False,
+                "error": str(e)
+            }, status=500)
+
+    async def save_metadata(self, request: web.Request) -> web.Response:
+        """Handle saving metadata updates for checkpoints"""
+        try:
+            if self.scanner is None:
+                self.scanner = await ServiceRegistry.get_checkpoint_scanner()
+                
+            data = await request.json()
+            file_path = data.get('file_path')
+            if not file_path:
+                return web.Response(text='File path is required', status=400)
+
+            # Remove file path from data to avoid saving it
+            metadata_updates = {k: v for k, v in data.items() if k != 'file_path'}
+            
+            # Get metadata file path
+            metadata_path = os.path.splitext(file_path)[0] + '.metadata.json'
+            
+            # Load existing metadata
+            metadata = await ModelRouteUtils.load_local_metadata(metadata_path)
+
+            # Update metadata
+            metadata.update(metadata_updates)
+
+            # Save updated metadata
+            with open(metadata_path, 'w', encoding='utf-8') as f:
+                json.dump(metadata, f, indent=2, ensure_ascii=False)
+
+            # Update cache
+            await self.scanner.update_single_model_cache(file_path, file_path, metadata)
+
+            # If model_name was updated, resort the cache
+            if 'model_name' in metadata_updates:
+                cache = await self.scanner.get_cached_data()
+                await cache.resort(name_only=True)
+
+            return web.json_response({'success': True})
+
+        except Exception as e:
+            logger.error(f"Error saving checkpoint metadata: {e}", exc_info=True)
+            return web.Response(text=str(e), status=500)
+
+    async def get_civitai_versions(self, request: web.Request) -> web.Response:
+        """Get available versions for a Civitai checkpoint model with local availability info"""
+        try:
+            if self.scanner is None:
+                self.scanner = await ServiceRegistry.get_checkpoint_scanner()
+                
+            # Get the civitai client from service registry
+            civitai_client = await ServiceRegistry.get_civitai_client()
+                
+            model_id = request.match_info['model_id']
+            response = await civitai_client.get_model_versions(model_id)
+            if not response or not response.get('modelVersions'):
+                return web.Response(status=404, text="Model not found")
+            
+            versions = response.get('modelVersions', [])
+            model_type = response.get('type', '')
+            
+            # Check model type - should be Checkpoint
+            if model_type.lower() != 'checkpoint':
+                return web.json_response({
+                    'error': f"Model type mismatch. Expected Checkpoint, got {model_type}"
+                }, status=400)
+            
+            # Check local availability for each version
+            for version in versions:
+                # Find the primary model file (type="Model" and primary=true) in the files list
+                model_file = next((file for file in version.get('files', []) 
+                                  if file.get('type') == 'Model' and file.get('primary') == True), None)
+                
+                # If no primary file found, try to find any model file
+                if not model_file:
+                    model_file = next((file for file in version.get('files', []) 
+                                      if file.get('type') == 'Model'), None)
+                
+                if model_file:
+                    sha256 = model_file.get('hashes', {}).get('SHA256')
+                    if sha256:
+                        # Set existsLocally and localPath at the version level
+                        version['existsLocally'] = self.scanner.has_hash(sha256)
+                        if version['existsLocally']:
+                            version['localPath'] = self.scanner.get_path_by_hash(sha256)
+                        
+                        # Also set the model file size at the version level for easier access
+                        version['modelSizeKB'] = model_file.get('sizeKB')
+                else:
+                    # No model file found in this version
+                    version['existsLocally'] = False
+                    
+            return web.json_response(versions)
+        except Exception as e:
+            logger.error(f"Error fetching checkpoint model versions: {e}")
+            return web.Response(status=500, text=str(e))

py/routes/lora_routes.py

@@ -1,11 +1,11 @@
 import os
 from aiohttp import web
 import jinja2
-from typing import Dict, List
+from typing import Dict
 import logging
-from ..services.lora_scanner import LoraScanner
 from ..config import config
-from ..services.settings_manager import settings  # Add this import
+from ..services.settings_manager import settings
+from ..services.service_registry import ServiceRegistry  # Add ServiceRegistry import
 
 logger = logging.getLogger(__name__)
 logging.getLogger('asyncio').setLevel(logging.CRITICAL)
@@ -14,22 +14,35 @@ class LoraRoutes:
     """Route handlers for LoRA management endpoints"""
     
     def __init__(self):
-        self.scanner = LoraScanner()
+        # Initialize service references as None, will be set during async init
+        self.scanner = None
+        self.recipe_scanner = None
         self.template_env = jinja2.Environment(
             loader=jinja2.FileSystemLoader(config.templates_path),
             autoescape=True
         )
 
+    async def init_services(self):
+        """Initialize services from ServiceRegistry"""
+        self.scanner = await ServiceRegistry.get_lora_scanner()
+        self.recipe_scanner = await ServiceRegistry.get_recipe_scanner()
+    
     def format_lora_data(self, lora: Dict) -> Dict:
         """Format LoRA data for template rendering"""
         return {
             "model_name": lora["model_name"],
             "file_name": lora["file_name"],   
             "preview_url": config.get_preview_static_url(lora["preview_url"]),
+            "preview_nsfw_level": lora.get("preview_nsfw_level", 0),
             "base_model": lora["base_model"],
             "folder": lora["folder"],
             "sha256": lora["sha256"],
             "file_path": lora["file_path"].replace(os.sep, "/"),
+            "size": lora["size"],
+            "tags": lora["tags"],
+            "modelDescription": lora["modelDescription"],
+            "usage_tips": lora["usage_tips"],
+            "notes": lora["notes"],
             "modified": lora["modified"],
             "from_civitai": lora.get("from_civitai", True),
             "civitai": self._filter_civitai_data(lora.get("civitai", {}))
@@ -50,30 +63,49 @@ class LoraRoutes:
     async def handle_loras_page(self, request: web.Request) -> web.Response:
         """Handle GET /loras request"""
         try:
-            # 不等待缓存数据，直接检查缓存状态
+            # Ensure services are initialized
+            await self.init_services()
+            
+            # Check if the LoraScanner is initializing
+            # It's initializing if the cache object doesn't exist yet,
+            # OR if the scanner explicitly says it's initializing (background task running).
             is_initializing = (
-                self.scanner._cache is None and 
-                (self.scanner._initialization_task is not None and 
-                not self.scanner._initialization_task.done())
+                self.scanner._cache is None or
+                (hasattr(self.scanner, '_is_initializing') and self.scanner._is_initializing)
             )
 
             if is_initializing:
-                # 如果正在初始化，返回一个只包含加载提示的页面
+                # If still initializing, return loading page
                 template = self.template_env.get_template('loras.html')
                 rendered = template.render(
-                    folders=[],  # 空文件夹列表
-                    is_initializing=True,  # 新增标志
-                    settings=settings  # Pass settings to template
+                    folders=[],
+                    is_initializing=True,
+                    settings=settings,
+                    request=request
                 )
+                
+                logger.info("Loras page is initializing, returning loading page")
             else:
-                # 正常流程
-                cache = await self.scanner.get_cached_data()
-                template = self.template_env.get_template('loras.html')
-                rendered = template.render(
-                    folders=cache.folders,
-                    is_initializing=False,
-                    settings=settings  # Pass settings to template
-                )
+                # Normal flow - get data from initialized cache
+                try:
+                    cache = await self.scanner.get_cached_data(force_refresh=False)
+                    template = self.template_env.get_template('loras.html')
+                    rendered = template.render(
+                        folders=cache.folders,
+                        is_initializing=False,
+                        settings=settings,
+                        request=request
+                    )
+                except Exception as cache_error:
+                    logger.error(f"Error loading cache data: {cache_error}")
+                    template = self.template_env.get_template('loras.html')
+                    rendered = template.render(
+                        folders=[],
+                        is_initializing=True,
+                        settings=settings,
+                        request=request
+                    )
+                    logger.info("Cache error, returning initialization page")
             
             return web.Response(
                 text=rendered,
@@ -87,6 +119,71 @@ class LoraRoutes:
                 status=500
             )
 
+    async def handle_recipes_page(self, request: web.Request) -> web.Response:
+        """Handle GET /loras/recipes request"""
+        try:
+            # Ensure services are initialized
+            await self.init_services()
+            
+            # Skip initialization check and directly try to get cached data
+            try:
+                # Recipe scanner will initialize cache if needed
+                await self.recipe_scanner.get_cached_data(force_refresh=False)
+                template = self.template_env.get_template('recipes.html')
+                rendered = template.render(
+                    recipes=[],  # Frontend will load recipes via API
+                    is_initializing=False,
+                    settings=settings,
+                    request=request
+                )
+            except Exception as cache_error:
+                logger.error(f"Error loading recipe cache data: {cache_error}")
+                # Still keep error handling - show initializing page on error
+                template = self.template_env.get_template('recipes.html')
+                rendered = template.render(
+                    is_initializing=True,
+                    settings=settings,
+                    request=request
+                )
+                logger.info("Recipe cache error, returning initialization page")
+            
+            return web.Response(
+                text=rendered,
+                content_type='text/html'
+            )
+            
+        except Exception as e:
+            logger.error(f"Error handling recipes request: {e}", exc_info=True)
+            return web.Response(
+                text="Error loading recipes page",
+                status=500
+            )
+
+    def _format_recipe_file_url(self, file_path: str) -> str:
+        """Format file path for recipe image as a URL - same as in recipe_routes"""
+        try:
+            # Return the file URL directly for the first lora root's preview
+            recipes_dir = os.path.join(config.loras_roots[0], "recipes").replace(os.sep, '/')
+            if file_path.replace(os.sep, '/').startswith(recipes_dir):
+                relative_path = os.path.relpath(file_path, config.loras_roots[0]).replace(os.sep, '/')
+                return f"/loras_static/root1/preview/{relative_path}" 
+            
+            # If not in recipes dir, try to create a valid URL from the file path
+            file_name = os.path.basename(file_path)
+            return f"/loras_static/root1/preview/recipes/{file_name}"
+        except Exception as e:
+            logger.error(f"Error formatting recipe file URL: {e}", exc_info=True)
+            return '/loras_static/images/no-preview.png'  # Return default image on error
+
     def setup_routes(self, app: web.Application):
         """Register routes with the application"""
+        # Add an app startup handler to initialize services
+        app.on_startup.append(self._on_startup)
+        
+        # Register routes
         app.router.add_get('/loras', self.handle_loras_page)
+        app.router.add_get('/loras/recipes', self.handle_recipes_page)
+        
+    async def _on_startup(self, app):
+        """Initialize services when the app starts"""
+        await self.init_services()

py/routes/update_routes.py

@@ -24,11 +24,9 @@ class UpdateRoutes:
         try:
             # Read local version from pyproject.toml
             local_version = UpdateRoutes._get_local_version()
-            logger.info(f"Local version: {local_version}")
-            
+
             # Fetch remote version from GitHub
             remote_version, changelog = await UpdateRoutes._get_remote_version()
-            logger.info(f"Remote version: {remote_version}")
             
             # Compare versions
             update_available = UpdateRoutes._compare_versions(
@@ -36,8 +34,6 @@ class UpdateRoutes:
                 remote_version.replace('v', '')
             )
             
-            logger.info(f"Update available: {update_available}")
-            
             return web.json_response({
                 'success': True,
                 'current_version': local_version,

py/services/checkpoint_scanner.py

@@ -0,0 +1,131 @@
+import os
+import logging
+import asyncio
+from typing import List, Dict, Optional, Set
+import folder_paths # type: ignore
+
+from ..utils.models import CheckpointMetadata
+from ..config import config
+from .model_scanner import ModelScanner
+from .model_hash_index import ModelHashIndex
+from .service_registry import ServiceRegistry
+
+logger = logging.getLogger(__name__)
+
+class CheckpointScanner(ModelScanner):
+    """Service for scanning and managing checkpoint files"""
+    
+    _instance = None
+    _lock = asyncio.Lock()
+    
+    def __new__(cls):
+        if cls._instance is None:
+            cls._instance = super().__new__(cls)
+        return cls._instance
+    
+    def __init__(self):
+        if not hasattr(self, '_initialized'):
+            # Define supported file extensions
+            file_extensions = {'.safetensors', '.ckpt', '.pt', '.pth', '.sft', '.gguf'}
+            super().__init__(
+                model_type="checkpoint",
+                model_class=CheckpointMetadata,
+                file_extensions=file_extensions,
+                hash_index=ModelHashIndex()
+            )
+            self._checkpoint_roots = self._init_checkpoint_roots()
+            self._initialized = True
+
+    @classmethod
+    async def get_instance(cls):
+        """Get singleton instance with async support"""
+        async with cls._lock:
+            if cls._instance is None:
+                cls._instance = cls()
+            return cls._instance
+    
+    def _init_checkpoint_roots(self) -> List[str]:
+        """Initialize checkpoint roots from ComfyUI settings"""
+        # Get both checkpoint and diffusion_models paths
+        checkpoint_paths = folder_paths.get_folder_paths("checkpoints")
+        diffusion_paths = folder_paths.get_folder_paths("diffusion_models")
+        
+        # Combine, normalize and deduplicate paths
+        all_paths = set()
+        for path in checkpoint_paths + diffusion_paths:
+            if os.path.exists(path):
+                norm_path = path.replace(os.sep, "/")
+                all_paths.add(norm_path)
+        
+        # Sort for consistent order
+        sorted_paths = sorted(all_paths, key=lambda p: p.lower())
+        
+        return sorted_paths
+    
+    def get_model_roots(self) -> List[str]:
+        """Get checkpoint root directories"""
+        return self._checkpoint_roots
+        
+    async def scan_all_models(self) -> List[Dict]:
+        """Scan all checkpoint directories and return metadata"""
+        all_checkpoints = []
+        
+        # Create scan tasks for each directory
+        scan_tasks = []
+        for root in self._checkpoint_roots:
+            task = asyncio.create_task(self._scan_directory(root))
+            scan_tasks.append(task)
+            
+        # Wait for all tasks to complete
+        for task in scan_tasks:
+            try:
+                checkpoints = await task
+                all_checkpoints.extend(checkpoints)
+            except Exception as e:
+                logger.error(f"Error scanning checkpoint directory: {e}")
+                
+        return all_checkpoints
+    
+    async def _scan_directory(self, root_path: str) -> List[Dict]:
+        """Scan a directory for checkpoint files"""
+        checkpoints = []
+        original_root = root_path
+
+        async def scan_recursive(path: str, visited_paths: set):
+            try:
+                real_path = os.path.realpath(path)
+                if real_path in visited_paths:
+                    logger.debug(f"Skipping already visited path: {path}")
+                    return
+                visited_paths.add(real_path)
+                
+                with os.scandir(path) as it:
+                    entries = list(it)
+                    for entry in entries:
+                        try:
+                            if entry.is_file(follow_symlinks=True):
+                                # Check if file has supported extension
+                                ext = os.path.splitext(entry.name)[1].lower()
+                                if ext in self.file_extensions:
+                                    file_path = entry.path.replace(os.sep, "/")
+                                    await self._process_single_file(file_path, original_root, checkpoints)
+                                    await asyncio.sleep(0)
+                            elif entry.is_dir(follow_symlinks=True):
+                                # For directories, continue scanning with original path
+                                await scan_recursive(entry.path, visited_paths)
+                        except Exception as e:
+                            logger.error(f"Error processing entry {entry.path}: {e}")
+            except Exception as e:
+                logger.error(f"Error scanning {path}: {e}")
+        
+        await scan_recursive(root_path, set())
+        return checkpoints
+
+    async def _process_single_file(self, file_path: str, root_path: str, checkpoints: list):
+        """Process a single checkpoint file and add to results"""
+        try:
+            result = await self._process_model_file(file_path, root_path)
+            if result:
+                checkpoints.append(result)
+        except Exception as e:
+            logger.error(f"Error processing {file_path}: {e}")

py/services/civitai_client.py

@@ -3,6 +3,7 @@ import aiohttp
 import os
 import json
 import logging
+import asyncio
 from email.parser import Parser
 from typing import Optional, Dict, Tuple, List
 from urllib.parse import unquote
@@ -11,20 +12,51 @@ from ..utils.models import LoraMetadata
 logger = logging.getLogger(__name__)
 
 class CivitaiClient:
+    _instance = None
+    _lock = asyncio.Lock()
+    
+    @classmethod
+    async def get_instance(cls):
+        """Get singleton instance of CivitaiClient"""
+        async with cls._lock:
+            if cls._instance is None:
+                cls._instance = cls()
+            return cls._instance
+
     def __init__(self):
+        # Check if already initialized for singleton pattern
+        if hasattr(self, '_initialized'):
+            return
+        self._initialized = True
+        
         self.base_url = "https://civitai.com/api/v1"
         self.headers = {
             'User-Agent': 'ComfyUI-LoRA-Manager/1.0'
         }
         self._session = None
+        # Set default buffer size to 1MB for higher throughput
+        self.chunk_size = 1024 * 1024
     
     @property
     async def session(self) -> aiohttp.ClientSession:
         """Lazy initialize the session"""
         if self._session is None:
-            connector = aiohttp.TCPConnector(ssl=True)
-            trust_env = True  # 允许使用系统环境变量中的代理设置
-            self._session = aiohttp.ClientSession(connector=connector, trust_env=trust_env)
+            # Optimize TCP connection parameters
+            connector = aiohttp.TCPConnector(
+                ssl=True,
+                limit=10,  # Increase parallel connections
+                ttl_dns_cache=300,  # DNS cache time
+                force_close=False,  # Keep connections for reuse
+                enable_cleanup_closed=True
+            )
+            trust_env = True  # Allow using system environment proxy settings
+            # Configure timeout parameters
+            timeout = aiohttp.ClientTimeout(total=None, connect=60, sock_read=60)
+            self._session = aiohttp.ClientSession(
+                connector=connector, 
+                trust_env=trust_env,
+                timeout=timeout
+            )
         return self._session
 
     def _parse_content_disposition(self, header: str) -> str:
@@ -74,8 +106,24 @@ class CivitaiClient:
         session = await self.session
         try:
             headers = self._get_request_headers()
+            
+            # Add Range header to allow resumable downloads
+            headers['Accept-Encoding'] = 'identity'  # Disable compression for better chunked downloads
+            
             async with session.get(url, headers=headers, allow_redirects=True) as response:
                 if response.status != 200:
+                    # Handle 401 unauthorized responses
+                    if response.status == 401:
+                        logger.warning(f"Unauthorized access to resource: {url} (Status 401)")
+                        
+                        return False, "Invalid or missing CivitAI API key, or early access restriction."
+                    
+                    # Handle other client errors that might be permission-related
+                    if response.status == 403:
+                        logger.warning(f"Forbidden access to resource: {url} (Status 403)")
+                        return False, "Access forbidden: You don't have permission to download this file."
+                    
+                    # Generic error response for other status codes
                     return False, f"Download failed with status {response.status}"
 
                 # Get filename from content-disposition header
@@ -89,16 +137,23 @@ class CivitaiClient:
                 # Get total file size for progress calculation
                 total_size = int(response.headers.get('content-length', 0))
                 current_size = 0
+                last_progress_report_time = datetime.now()
 
-                # Stream download to file with progress updates
+                # Stream download to file with progress updates using larger buffer
                 with open(save_path, 'wb') as f:
-                    async for chunk in response.content.iter_chunked(8192):
+                    async for chunk in response.content.iter_chunked(self.chunk_size):
                         if chunk:
                             f.write(chunk)
                             current_size += len(chunk)
-                            if progress_callback and total_size:
+                            
+                            # Limit progress update frequency to reduce overhead
+                            now = datetime.now()
+                            time_diff = (now - last_progress_report_time).total_seconds()
+                            
+                            if progress_callback and total_size and time_diff >= 0.5:
                                 progress = (current_size / total_size) * 100
                                 await progress_callback(progress)
+                                last_progress_report_time = now
                 
                 # Ensure 100% progress is reported
                 if progress_callback:
@@ -106,6 +161,9 @@ class CivitaiClient:
                         
                 return True, save_path
                 
+        except aiohttp.ClientError as e:
+            logger.error(f"Network error during download: {e}")
+            return False, f"Network error: {str(e)}"
         except Exception as e:
             logger.error(f"Download error: {e}")
             return False, str(e)
@@ -143,7 +201,11 @@ class CivitaiClient:
                 if response.status != 200:
                     return None
                 data = await response.json()
-                return data.get('modelVersions', [])
+                # Also return model type along with versions
+                return {
+                    'modelVersions': data.get('modelVersions', []),
+                    'type': data.get('type', '')
+                }
         except Exception as e:
             logger.error(f"Error fetching model versions: {e}")
             return None
@@ -163,8 +225,78 @@ class CivitaiClient:
             logger.error(f"Error fetching model version info: {e}")
             return None
 
+    async def get_model_metadata(self, model_id: str) -> Tuple[Optional[Dict], int]:
+        """Fetch model metadata (description and tags) from Civitai API
+        
+        Args:
+            model_id: The Civitai model ID
+            
+        Returns:
+            Tuple[Optional[Dict], int]: A tuple containing:
+                - A dictionary with model metadata or None if not found
+                - The HTTP status code from the request
+        """
+        try:
+            session = await self.session
+            headers = self._get_request_headers()
+            url = f"{self.base_url}/models/{model_id}"
+            
+            async with session.get(url, headers=headers) as response:
+                status_code = response.status
+                
+                if status_code != 200:
+                    logger.warning(f"Failed to fetch model metadata: Status {status_code}")
+                    return None, status_code
+                
+                data = await response.json()
+                
+                # Extract relevant metadata
+                metadata = {
+                    "description": data.get("description") or "No model description available",
+                    "tags": data.get("tags", [])
+                }
+                
+                if metadata["description"] or metadata["tags"]:
+                    return metadata, status_code
+                else:
+                    logger.warning(f"No metadata found for model {model_id}")
+                    return None, status_code
+                
+        except Exception as e:
+            logger.error(f"Error fetching model metadata: {e}", exc_info=True)
+            return None, 0
+
+    # Keep old method for backward compatibility, delegating to the new one
+    async def get_model_description(self, model_id: str) -> Optional[str]:
+        """Fetch the model description from Civitai API (Legacy method)"""
+        metadata, _ = await self.get_model_metadata(model_id)
+        return metadata.get("description") if metadata else None
+
     async def close(self):
         """Close the session if it exists"""
         if self._session is not None:
             await self._session.close()
-            self._session = None
+            self._session = None
+
+    async def _get_hash_from_civitai(self, model_version_id: str) -> Optional[str]:
+        """Get hash from Civitai API"""
+        try:
+            if not self._session:
+                return None
+            
+            version_info = await self._session.get(f"{self.base_url}/model-versions/{model_version_id}")
+            
+            if not version_info or not version_info.json().get('files'):
+                return None
+            
+            # Get hash from the first file
+            for file_info in version_info.json().get('files', []):
+                if file_info.get('hashes', {}).get('SHA256'):
+                    # Convert hash to lowercase to standardize
+                    hash_value = file_info['hashes']['SHA256'].lower()
+                    return hash_value
+                
+            return None
+        except Exception as e:
+            logger.error(f"Error getting hash from Civitai: {e}")
+            return None

py/services/download_manager.py

@@ -1,20 +1,79 @@
 import logging
 import os
 import json
-from typing import Optional, Dict
+import asyncio
+from typing import Optional, Dict, Any
 from .civitai_client import CivitaiClient
-from .file_monitor import LoraFileMonitor
-from ..utils.models import LoraMetadata
+from ..utils.models import LoraMetadata, CheckpointMetadata
+from ..utils.constants import CARD_PREVIEW_WIDTH
+from ..utils.exif_utils import ExifUtils
+from .service_registry import ServiceRegistry
+
+# Download to temporary file first
+import tempfile
 
 logger = logging.getLogger(__name__)
 
 class DownloadManager:
-    def __init__(self, file_monitor: Optional[LoraFileMonitor] = None):
-        self.civitai_client = CivitaiClient()
-        self.file_monitor = file_monitor
+    _instance = None
+    _lock = asyncio.Lock()
+    
+    @classmethod
+    async def get_instance(cls):
+        """Get singleton instance of DownloadManager"""
+        async with cls._lock:
+            if cls._instance is None:
+                cls._instance = cls()
+            return cls._instance
 
-    async def download_from_civitai(self, download_url: str, save_dir: str, relative_path: str = '', 
-                                  progress_callback=None) -> Dict:
+    def __init__(self):
+        # Check if already initialized for singleton pattern
+        if hasattr(self, '_initialized'):
+            return
+        self._initialized = True
+        
+        self._civitai_client = None  # Will be lazily initialized
+
+    async def _get_civitai_client(self):
+        """Lazily initialize CivitaiClient from registry"""
+        if self._civitai_client is None:
+            self._civitai_client = await ServiceRegistry.get_civitai_client()
+        return self._civitai_client
+
+    async def _get_lora_monitor(self):
+        """Get the lora file monitor from registry"""
+        return await ServiceRegistry.get_lora_monitor()
+        
+    async def _get_checkpoint_monitor(self):
+        """Get the checkpoint file monitor from registry"""
+        return await ServiceRegistry.get_checkpoint_monitor()
+    
+    async def _get_lora_scanner(self):
+        """Get the lora scanner from registry"""
+        return await ServiceRegistry.get_lora_scanner()
+        
+    async def _get_checkpoint_scanner(self):
+        """Get the checkpoint scanner from registry"""
+        return await ServiceRegistry.get_checkpoint_scanner()
+
+    async def download_from_civitai(self, download_url: str = None, model_hash: str = None, 
+                                  model_version_id: str = None, save_dir: str = None, 
+                                  relative_path: str = '', progress_callback=None, 
+                                  model_type: str = "lora") -> Dict:
+        """Download model from Civitai
+        
+        Args:
+            download_url: Direct download URL for the model
+            model_hash: SHA256 hash of the model
+            model_version_id: Civitai model version ID
+            save_dir: Directory to save the model to
+            relative_path: Relative path within save_dir
+            progress_callback: Callback function for progress updates
+            model_type: Type of model ('lora' or 'checkpoint')
+            
+        Returns:
+            Dict with download result
+        """
         try:
             # Update save directory with relative path if provided
             if relative_path:
@@ -22,56 +81,114 @@ class DownloadManager:
                 # Create directory if it doesn't exist
                 os.makedirs(save_dir, exist_ok=True)
 
-            # Get version info
-            version_id = download_url.split('/')[-1]
-            version_info = await self.civitai_client.get_model_version_info(version_id)
+            # Get civitai client
+            civitai_client = await self._get_civitai_client()
+
+            # Get version info based on the provided identifier
+            version_info = None
+            
+            if download_url:
+                # Extract version ID from download URL
+                version_id = download_url.split('/')[-1]
+                version_info = await civitai_client.get_model_version_info(version_id)
+            elif model_version_id:
+                # Use model version ID directly
+                version_info = await civitai_client.get_model_version_info(model_version_id)
+            elif model_hash:
+                # Get model by hash
+                version_info = await civitai_client.get_model_by_hash(model_hash)
+
+            
             if not version_info:
                 return {'success': False, 'error': 'Failed to fetch model metadata'}
 
+            # Check if this is an early access model
+            if version_info.get('earlyAccessEndsAt'):
+                early_access_date = version_info.get('earlyAccessEndsAt', '')
+                # Convert to a readable date if possible
+                try:
+                    from datetime import datetime
+                    date_obj = datetime.fromisoformat(early_access_date.replace('Z', '+00:00'))
+                    formatted_date = date_obj.strftime('%Y-%m-%d')
+                    early_access_msg = f"This model requires early access payment (until {formatted_date}). "
+                except:
+                    early_access_msg = "This model requires early access payment. "
+                
+                early_access_msg += "Please ensure you have purchased early access and are logged in to Civitai."
+                logger.warning(f"Early access model detected: {version_info.get('name', 'Unknown')}")
+                
+                # We'll still try to download, but log a warning and prepare for potential failure
+                if progress_callback:
+                    await progress_callback(1)  # Show minimal progress to indicate we're trying
+
             # Report initial progress
             if progress_callback:
                 await progress_callback(0)
 
-            # 2. 获取文件信息
+            # 2. Get file information
             file_info = next((f for f in version_info.get('files', []) if f.get('primary')), None)
             if not file_info:
                 return {'success': False, 'error': 'No primary file found in metadata'}
 
-            # 3. 准备下载
+            # 3. Prepare download
             file_name = file_info['name']
             save_path = os.path.join(save_dir, file_name)
             file_size = file_info.get('sizeKB', 0) * 1024
 
-            # 4. 通知文件监控系统
-            self.file_monitor.handler.add_ignore_path(
-                save_path.replace(os.sep, '/'),
-                file_size
-            )
+            # 4. Notify file monitor - use normalized path and file size
+            file_monitor = await self._get_lora_monitor() if model_type == "lora" else await self._get_checkpoint_monitor()
+            if file_monitor and file_monitor.handler:
+                file_monitor.handler.add_ignore_path(
+                    save_path.replace(os.sep, '/'),
+                    file_size
+                )
 
-            # 5. 准备元数据
-            metadata = LoraMetadata.from_civitai_info(version_info, file_info, save_path)
+            # 5. Prepare metadata based on model type
+            if model_type == "checkpoint":
+                metadata = CheckpointMetadata.from_civitai_info(version_info, file_info, save_path)
+                logger.info(f"Creating CheckpointMetadata for {file_name}")
+            else:
+                metadata = LoraMetadata.from_civitai_info(version_info, file_info, save_path)
+                logger.info(f"Creating LoraMetadata for {file_name}")
             
-            # 6. 开始下载流程
+            # 5.1 Get and update model tags and description
+            model_id = version_info.get('modelId')
+            if model_id:
+                model_metadata, _ = await civitai_client.get_model_metadata(str(model_id))
+                if model_metadata:
+                    if model_metadata.get("tags"):
+                        metadata.tags = model_metadata.get("tags", [])
+                    if model_metadata.get("description"):
+                        metadata.modelDescription = model_metadata.get("description", "")
+            
+            # 6. Start download process
             result = await self._execute_download(
-                download_url=download_url,
+                download_url=file_info.get('downloadUrl', ''),
                 save_dir=save_dir,
                 metadata=metadata,
                 version_info=version_info,
                 relative_path=relative_path,
-                progress_callback=progress_callback
+                progress_callback=progress_callback,
+                model_type=model_type
             )
 
             return result
 
         except Exception as e:
             logger.error(f"Error in download_from_civitai: {e}", exc_info=True)
+            # Check if this might be an early access error
+            error_str = str(e).lower()
+            if "403" in error_str or "401" in error_str or "unauthorized" in error_str or "early access" in error_str:
+                return {'success': False, 'error': f"Early access restriction: {str(e)}. Please ensure you have purchased early access and are logged in to Civitai."}
             return {'success': False, 'error': str(e)}
 
     async def _execute_download(self, download_url: str, save_dir: str, 
-                              metadata: LoraMetadata, version_info: Dict, 
-                              relative_path: str, progress_callback=None) -> Dict:
+                              metadata, version_info: Dict, 
+                              relative_path: str, progress_callback=None,
+                              model_type: str = "lora") -> Dict:
         """Execute the actual download process including preview images and model files"""
         try:
+            civitai_client = await self._get_civitai_client()
             save_path = metadata.file_path
             metadata_path = os.path.splitext(save_path)[0] + '.metadata.json'
 
@@ -82,19 +199,61 @@ class DownloadManager:
                 if progress_callback:
                     await progress_callback(1)  # 1% progress for starting preview download
 
-                preview_ext = '.mp4' if images[0].get('type') == 'video' else '.png'
-                preview_path = os.path.splitext(save_path)[0] + '.preview' + preview_ext
-                if await self.civitai_client.download_preview_image(images[0]['url'], preview_path):
-                    metadata.preview_url = preview_path.replace(os.sep, '/')
-                    with open(metadata_path, 'w', encoding='utf-8') as f:
-                        json.dump(metadata.to_dict(), f, indent=2, ensure_ascii=False)
+                # Check if it's a video or an image
+                is_video = images[0].get('type') == 'video'
+                
+                if is_video:
+                    # For videos, use .mp4 extension
+                    preview_ext = '.mp4'
+                    preview_path = os.path.splitext(save_path)[0] + preview_ext
+                    
+                    # Download video directly
+                    if await civitai_client.download_preview_image(images[0]['url'], preview_path):
+                        metadata.preview_url = preview_path.replace(os.sep, '/')
+                        metadata.preview_nsfw_level = images[0].get('nsfwLevel', 0)
+                        with open(metadata_path, 'w', encoding='utf-8') as f:
+                            json.dump(metadata.to_dict(), f, indent=2, ensure_ascii=False)
+                else:
+                    # For images, use WebP format for better performance
+                    with tempfile.NamedTemporaryFile(suffix='.png', delete=False) as temp_file:
+                        temp_path = temp_file.name
+                    
+                    # Download the original image to temp path
+                    if await civitai_client.download_preview_image(images[0]['url'], temp_path):
+                        # Optimize and convert to WebP
+                        preview_path = os.path.splitext(save_path)[0] + '.webp'
+                        
+                        # Use ExifUtils to optimize and convert the image
+                        optimized_data, _ = ExifUtils.optimize_image(
+                            image_data=temp_path,
+                            target_width=CARD_PREVIEW_WIDTH,
+                            format='webp',
+                            quality=85,
+                            preserve_metadata=True
+                        )
+                        
+                        # Save the optimized image
+                        with open(preview_path, 'wb') as f:
+                            f.write(optimized_data)
+                            
+                        # Update metadata
+                        metadata.preview_url = preview_path.replace(os.sep, '/')
+                        metadata.preview_nsfw_level = images[0].get('nsfwLevel', 0)
+                        with open(metadata_path, 'w', encoding='utf-8') as f:
+                            json.dump(metadata.to_dict(), f, indent=2, ensure_ascii=False)
+                        
+                        # Remove temporary file
+                        try:
+                            os.unlink(temp_path)
+                        except Exception as e:
+                            logger.warning(f"Failed to delete temp file: {e}")
 
                 # Report preview download completion
                 if progress_callback:
                     await progress_callback(3)  # 3% progress after preview download
 
             # Download model file with progress tracking
-            success, result = await self.civitai_client._download_file(
+            success, result = await civitai_client._download_file(
                 download_url, 
                 save_dir,
                 os.path.basename(save_path),
@@ -108,15 +267,22 @@ class DownloadManager:
                         os.remove(path)
                 return {'success': False, 'error': result}
 
-            # 4. 更新文件信息（大小和修改时间）
+            # 4. Update file information (size and modified time)
             metadata.update_file_info(save_path)
 
-            # 5. 最终更新元数据
+            # 5. Final metadata update
             with open(metadata_path, 'w', encoding='utf-8') as f:
                 json.dump(metadata.to_dict(), f, indent=2, ensure_ascii=False)
 
-            # 6. update lora cache
-            cache = await self.file_monitor.scanner.get_cached_data()
+            # 6. Update cache based on model type
+            if model_type == "checkpoint":
+                scanner = await self._get_checkpoint_scanner()
+                logger.info(f"Updating checkpoint cache for {save_path}")
+            else:
+                scanner = await self._get_lora_scanner()
+                logger.info(f"Updating lora cache for {save_path}")
+                
+            cache = await scanner.get_cached_data()
             metadata_dict = metadata.to_dict()
             metadata_dict['folder'] = relative_path
             cache.raw_data.append(metadata_dict)
@@ -124,6 +290,9 @@ class DownloadManager:
             all_folders = set(cache.folders)
             all_folders.add(relative_path)
             cache.folders = sorted(list(all_folders), key=lambda x: x.lower())
+            
+            # Update the hash index with the new model entry
+            scanner._hash_index.add_entry(metadata_dict['sha256'], metadata_dict['file_path'])
 
             # Report 100% completion
             if progress_callback:

py/services/file_monitor.py

@@ -1,28 +1,42 @@
-from operator import itemgetter
 import os
 import logging
 import asyncio
+import time
 from watchdog.observers import Observer
-from watchdog.events import FileSystemEventHandler, FileCreatedEvent, FileDeletedEvent
-from typing import List
+from watchdog.events import FileSystemEventHandler
+from typing import List, Dict, Set, Optional
 from threading import Lock
-from .lora_scanner import LoraScanner
+
 from ..config import config
+from .service_registry import ServiceRegistry
 
 logger = logging.getLogger(__name__)
 
-class LoraFileHandler(FileSystemEventHandler):
-    """Handler for LoRA file system events"""
+# Configuration constant to control file monitoring functionality
+ENABLE_FILE_MONITORING = False
+
+class BaseFileHandler(FileSystemEventHandler):
+    """Base handler for file system events"""
     
-    def __init__(self, scanner: LoraScanner, loop: asyncio.AbstractEventLoop):
-        self.scanner = scanner
-        self.loop = loop  # 存储事件循环引用
-        self.pending_changes = set()  # 待处理的变更
-        self.lock = Lock()  # 线程安全锁
-        self.update_task = None  # 异步更新任务
-        self._ignore_paths = set()  # Add ignore paths set
-        self._min_ignore_timeout = 5  # minimum timeout in seconds
-        self._download_speed = 1024 * 1024  # assume 1MB/s as base speed
+    def __init__(self, loop: asyncio.AbstractEventLoop):
+        self.loop = loop  # Store event loop reference
+        self.pending_changes = set()  # Pending changes
+        self.lock = Lock()  # Thread-safe lock
+        self.update_task = None  # Async update task
+        self._ignore_paths = set()  # Paths to ignore
+        self._min_ignore_timeout = 5  # Minimum timeout in seconds
+        self._download_speed = 1024 * 1024  # Assume 1MB/s as base speed
+        
+        # Track modified files with timestamps for debouncing
+        self.modified_files: Dict[str, float] = {}
+        self.debounce_timer = None
+        self.debounce_delay = 3.0  # Seconds to wait after last modification
+        
+        # Track files already scheduled for processing
+        self.scheduled_files: Set[str] = set()
+        
+        # File extensions to monitor - should be overridden by subclasses
+        self.file_extensions = set()
 
     def _should_ignore(self, path: str) -> bool:
         """Check if path should be ignored"""
@@ -37,32 +51,152 @@ class LoraFileHandler(FileSystemEventHandler):
         # Short timeout (e.g. 5 seconds) is sufficient to ignore the CREATE event
         timeout = 5
         
-        asyncio.get_event_loop().call_later(
+        self.loop.call_later(
             timeout,
             self._ignore_paths.discard,
             real_path.replace(os.sep, '/')
         )
         
     def on_created(self, event):
-        if event.is_directory or not event.src_path.endswith('.safetensors'):
+        if event.is_directory:
             return
-        if self._should_ignore(event.src_path):
+            
+        # Handle appropriate files based on extensions
+        file_ext = os.path.splitext(event.src_path)[1].lower()
+        if file_ext in self.file_extensions:
+            if self._should_ignore(event.src_path):
+                return
+            
+            # Process this file directly and ignore subsequent modifications
+            normalized_path = os.path.realpath(event.src_path).replace(os.sep, '/')
+            if normalized_path not in self.scheduled_files:
+                logger.info(f"File created: {event.src_path}")
+                self.scheduled_files.add(normalized_path)
+                self._schedule_update('add', event.src_path)
+                
+                # Ignore modifications for a short period after creation
+                self.loop.call_later(
+                    self.debounce_delay * 2,
+                    self.scheduled_files.discard,
+                    normalized_path
+                )
+            
+    def on_modified(self, event):
+        if event.is_directory:
             return
-        logger.info(f"LoRA file created: {event.src_path}")
-        self._schedule_update('add', event.src_path)
+            
+        # Only process files with supported extensions
+        file_ext = os.path.splitext(event.src_path)[1].lower()
+        if file_ext in self.file_extensions:
+            if self._should_ignore(event.src_path):
+                return
+                
+            normalized_path = os.path.realpath(event.src_path).replace(os.sep, '/')
+            
+            # Skip if this file is already scheduled for processing
+            if normalized_path in self.scheduled_files:
+                return
+                
+            # Update the timestamp for this file
+            self.modified_files[normalized_path] = time.time()
+            
+            # Cancel any existing timer
+            if self.debounce_timer:
+                self.debounce_timer.cancel()
+                
+            # Set a new timer to process modified files after debounce period
+            self.debounce_timer = self.loop.call_later(
+                self.debounce_delay, 
+                self.loop.call_soon_threadsafe,
+                self._process_modified_files
+            )
+
+    def _process_modified_files(self):
+        """Process files that have been modified after debounce period"""
+        current_time = time.time()
+        files_to_process = []
+        
+        # Find files that haven't been modified for debounce_delay seconds
+        for file_path, last_modified in list(self.modified_files.items()):
+            if current_time - last_modified >= self.debounce_delay:
+                # Only process if not already scheduled
+                if file_path not in self.scheduled_files:
+                    files_to_process.append(file_path)
+                    self.scheduled_files.add(file_path)
+                    
+                    # Auto-remove from scheduled list after reasonable time
+                    self.loop.call_later(
+                        self.debounce_delay * 2,
+                        self.scheduled_files.discard,
+                        file_path
+                    )
+                
+                del self.modified_files[file_path]
+        
+        # Process stable files
+        for file_path in files_to_process:
+            logger.info(f"Processing modified file: {file_path}")
+            self._schedule_update('add', file_path)
 
     def on_deleted(self, event):
-        if event.is_directory or not event.src_path.endswith('.safetensors'):
+        if event.is_directory:
             return
+            
+        file_ext = os.path.splitext(event.src_path)[1].lower()
+        if file_ext not in self.file_extensions:
+            return
+            
         if self._should_ignore(event.src_path):
             return
-        logger.info(f"LoRA file deleted: {event.src_path}")
+        
+        # Remove from scheduled files if present
+        normalized_path = os.path.realpath(event.src_path).replace(os.sep, '/')
+        self.scheduled_files.discard(normalized_path)
+        
+        logger.info(f"File deleted: {event.src_path}")
         self._schedule_update('remove', event.src_path)
         
-    def _schedule_update(self, action: str, file_path: str): #file_path is a real path
+    def on_moved(self, event):
+        """Handle file move/rename events"""
+        
+        src_ext = os.path.splitext(event.src_path)[1].lower()
+        dest_ext = os.path.splitext(event.dest_path)[1].lower()
+        
+        # If destination has supported extension, treat as new file
+        if dest_ext in self.file_extensions:
+            if self._should_ignore(event.dest_path):
+                return
+                
+            normalized_path = os.path.realpath(event.dest_path).replace(os.sep, '/')
+            
+            # Only process if not already scheduled
+            if normalized_path not in self.scheduled_files:
+                logger.info(f"File renamed/moved to: {event.dest_path}")
+                self.scheduled_files.add(normalized_path)
+                self._schedule_update('add', event.dest_path)
+                
+                # Auto-remove from scheduled list after reasonable time
+                self.loop.call_later(
+                    self.debounce_delay * 2, 
+                    self.scheduled_files.discard,
+                    normalized_path
+                )
+            
+        # If source was a supported file, treat it as deleted
+        if src_ext in self.file_extensions:
+            if self._should_ignore(event.src_path):
+                return
+                
+            normalized_path = os.path.realpath(event.src_path).replace(os.sep, '/')
+            self.scheduled_files.discard(normalized_path)
+            
+            logger.info(f"File moved/renamed from: {event.src_path}")
+            self._schedule_update('remove', event.src_path)
+
+    def _schedule_update(self, action: str, file_path: str):
         """Schedule a cache update"""
         with self.lock:
-            # 使用 config 中的方法映射路径
+            # Use config method to map path
             mapped_path = config.map_path_to_link(file_path)
             normalized_path = mapped_path.replace(os.sep, '/')
             self.pending_changes.add((action, normalized_path))
@@ -73,7 +207,20 @@ class LoraFileHandler(FileSystemEventHandler):
         """Create update task in the event loop"""
         if self.update_task is None or self.update_task.done():
             self.update_task = asyncio.create_task(self._process_changes())
+    
+    async def _process_changes(self, delay: float = 2.0):
+        """Process pending changes with debouncing - should be implemented by subclasses"""
+        raise NotImplementedError("Subclasses must implement _process_changes")
 
+
+class LoraFileHandler(BaseFileHandler):
+    """Handler for LoRA file system events"""
+    
+    def __init__(self, loop: asyncio.AbstractEventLoop):
+        super().__init__(loop)
+        # Set supported file extensions for LoRAs
+        self.file_extensions = {'.safetensors'}
+    
     async def _process_changes(self, delay: float = 2.0):
         """Process pending changes with debouncing"""
         await asyncio.sleep(delay)
@@ -86,32 +233,54 @@ class LoraFileHandler(FileSystemEventHandler):
             if not changes:
                 return
             
-            logger.info(f"Processing {len(changes)} file changes")
+            logger.info(f"Processing {len(changes)} LoRA file changes")
 
-            cache = await self.scanner.get_cached_data()
+            # Get scanner through ServiceRegistry
+            scanner = await ServiceRegistry.get_lora_scanner()
+            cache = await scanner.get_cached_data()
             needs_resort = False
             new_folders = set()
             
             for action, file_path in changes:
                 try:
                     if action == 'add':
+                        # Check if file already exists in cache
+                        existing = next((item for item in cache.raw_data if item['file_path'] == file_path), None)
+                        if existing:
+                            logger.info(f"File {file_path} already in cache, skipping")
+                            continue
+                            
                         # Scan new file
-                        lora_data = await self.scanner.scan_single_lora(file_path)
-                        if lora_data:
-                            cache.raw_data.append(lora_data)
-                            new_folders.add(lora_data['folder'])
+                        model_data = await scanner.scan_single_model(file_path)
+                        if model_data:
+                            # Update tags count
+                            for tag in model_data.get('tags', []):
+                                scanner._tags_count[tag] = scanner._tags_count.get(tag, 0) + 1
+                            
+                            cache.raw_data.append(model_data)
+                            new_folders.add(model_data['folder'])
                             # Update hash index
-                            if 'sha256' in lora_data:
-                                self.scanner._hash_index.add_entry(
-                                    lora_data['sha256'], 
-                                    lora_data['file_path']
+                            if 'sha256' in model_data:
+                                scanner._hash_index.add_entry(
+                                    model_data['sha256'], 
+                                    model_data['file_path']
                                 )
                             needs_resort = True
                             
                     elif action == 'remove':
+                        # Find the model to remove so we can update tags count
+                        model_to_remove = next((item for item in cache.raw_data if item['file_path'] == file_path), None)
+                        if model_to_remove:
+                            # Update tags count by reducing counts
+                            for tag in model_to_remove.get('tags', []):
+                                if tag in scanner._tags_count:
+                                    scanner._tags_count[tag] = max(0, scanner._tags_count[tag] - 1)
+                                    if scanner._tags_count[tag] == 0:
+                                        del scanner._tags_count[tag]
+                        
                         # Remove from cache and hash index
                         logger.info(f"Removing {file_path} from cache")
-                        self.scanner._hash_index.remove_by_path(file_path)
+                        scanner._hash_index.remove_by_path(file_path)
                         cache.raw_data = [
                             item for item in cache.raw_data 
                             if item['file_path'] != file_path
@@ -129,62 +298,245 @@ class LoraFileHandler(FileSystemEventHandler):
                 cache.folders = sorted(list(all_folders), key=lambda x: x.lower())
                 
         except Exception as e:
-            logger.error(f"Error in process_changes: {e}")
+            logger.error(f"Error in process_changes for LoRA: {e}")
 
 
-class LoraFileMonitor:
-    """Monitor for LoRA file changes"""
+class CheckpointFileHandler(BaseFileHandler):
+    """Handler for checkpoint file system events"""
     
-    def __init__(self, scanner: LoraScanner, roots: List[str]):
-        self.scanner = scanner
-        scanner.set_file_monitor(self)
+    def __init__(self, loop: asyncio.AbstractEventLoop):
+        super().__init__(loop)
+        # Set supported file extensions for checkpoints
+        self.file_extensions = {'.safetensors', '.ckpt', '.pt', '.pth', '.sft', '.gguf'}
+    
+    async def _process_changes(self, delay: float = 2.0):
+        """Process pending changes with debouncing for checkpoint files"""
+        await asyncio.sleep(delay)
+        
+        try:
+            with self.lock:
+                changes = self.pending_changes.copy()
+                self.pending_changes.clear()
+            
+            if not changes:
+                return
+            
+            logger.info(f"Processing {len(changes)} checkpoint file changes")
+
+            # Get scanner through ServiceRegistry
+            scanner = await ServiceRegistry.get_checkpoint_scanner()
+            cache = await scanner.get_cached_data()
+            needs_resort = False
+            new_folders = set()
+            
+            for action, file_path in changes:
+                try:
+                    if action == 'add':
+                        # Check if file already exists in cache
+                        existing = next((item for item in cache.raw_data if item['file_path'] == file_path), None)
+                        if existing:
+                            logger.info(f"File {file_path} already in cache, skipping")
+                            continue
+                            
+                        # Scan new file
+                        model_data = await scanner.scan_single_model(file_path)
+                        if model_data:
+                            # Update tags count if applicable
+                            for tag in model_data.get('tags', []):
+                                scanner._tags_count[tag] = scanner._tags_count.get(tag, 0) + 1
+                            
+                            cache.raw_data.append(model_data)
+                            new_folders.add(model_data['folder'])
+                            # Update hash index
+                            if 'sha256' in model_data:
+                                scanner._hash_index.add_entry(
+                                    model_data['sha256'], 
+                                    model_data['file_path']
+                                )
+                            needs_resort = True
+                            
+                    elif action == 'remove':
+                        # Find the model to remove so we can update tags count
+                        model_to_remove = next((item for item in cache.raw_data if item['file_path'] == file_path), None)
+                        if model_to_remove:
+                            # Update tags count by reducing counts
+                            for tag in model_to_remove.get('tags', []):
+                                if tag in scanner._tags_count:
+                                    scanner._tags_count[tag] = max(0, scanner._tags_count[tag] - 1)
+                                    if scanner._tags_count[tag] == 0:
+                                        del scanner._tags_count[tag]
+                        
+                        # Remove from cache and hash index
+                        logger.info(f"Removing {file_path} from checkpoint cache")
+                        scanner._hash_index.remove_by_path(file_path)
+                        cache.raw_data = [
+                            item for item in cache.raw_data 
+                            if item['file_path'] != file_path
+                        ]
+                        needs_resort = True
+                        
+                except Exception as e:
+                    logger.error(f"Error processing checkpoint {action} for {file_path}: {e}")
+            
+            if needs_resort:
+                await cache.resort()
+                
+                # Update folder list
+                all_folders = set(cache.folders) | new_folders
+                cache.folders = sorted(list(all_folders), key=lambda x: x.lower())
+                
+        except Exception as e:
+            logger.error(f"Error in process_changes for checkpoint: {e}")
+
+
+class BaseFileMonitor:
+    """Base class for file monitoring"""
+    
+    def __init__(self, monitor_paths: List[str]):
         self.observer = Observer()
         self.loop = asyncio.get_event_loop()
-        self.handler = LoraFileHandler(scanner, self.loop)
-        
-        # 使用已存在的路径映射
         self.monitor_paths = set()
-        for root in roots:
-            self.monitor_paths.add(os.path.realpath(root).replace(os.sep, '/'))
+        
+        # Process monitor paths
+        for path in monitor_paths:
+            self.monitor_paths.add(os.path.realpath(path).replace(os.sep, '/'))
 
-        # 添加所有已映射的目标路径
+        # Add mapped paths from config
         for target_path in config._path_mappings.keys():
             self.monitor_paths.add(target_path)
-
+    
     def start(self):
-        """Start monitoring"""
-        for path_info in self.monitor_paths:
+        """Start file monitoring"""
+        if not ENABLE_FILE_MONITORING:
+            logger.info("File monitoring is disabled via ENABLE_FILE_MONITORING setting")
+            return
+            
+        for path in self.monitor_paths:
             try:
-                if isinstance(path_info, tuple):
-                    # 对于链接，监控目标路径
-                    _, target_path = path_info
-                    self.observer.schedule(self.handler, target_path, recursive=True)
-                    logger.info(f"Started monitoring target path: {target_path}")
-                else:
-                    # 对于普通路径，直接监控
-                    self.observer.schedule(self.handler, path_info, recursive=True)
-                    logger.info(f"Started monitoring: {path_info}")
+                self.observer.schedule(self.handler, path, recursive=True)
+                logger.info(f"Started monitoring: {path}")
             except Exception as e:
-                logger.error(f"Error monitoring {path_info}: {e}")
+                logger.error(f"Error monitoring {path}: {e}")
                 
         self.observer.start()
-        
+    
     def stop(self):
-        """Stop monitoring"""
+        """Stop file monitoring"""
+        if not ENABLE_FILE_MONITORING:
+            return
+            
         self.observer.stop()
         self.observer.join()
-
+    
     def rescan_links(self):
-        """重新扫描链接（当添加新的链接时调用）"""
+        """Rescan links when new ones are added"""
+        if not ENABLE_FILE_MONITORING:
+            return
+            
+        # Find new paths not yet being monitored
         new_paths = set()
-        for path in self.monitor_paths.copy():
-            self._add_link_targets(path)
+        for path in config._path_mappings.keys():
+            real_path = os.path.realpath(path).replace(os.sep, '/')
+            if real_path not in self.monitor_paths:
+                new_paths.add(real_path)
+                self.monitor_paths.add(real_path)
         
-        # 添加新发现的路径到监控
-        new_paths = self.monitor_paths - set(self.observer.watches.keys())
+        # Add new paths to monitoring
         for path in new_paths:
             try:
                 self.observer.schedule(self.handler, path, recursive=True)
                 logger.info(f"Added new monitoring path: {path}")
             except Exception as e:
-                logger.error(f"Error adding new monitor for {path}: {e}")
+                logger.error(f"Error adding new monitor for {path}: {e}")
+
+
+class LoraFileMonitor(BaseFileMonitor):
+    """Monitor for LoRA file changes"""
+    
+    _instance = None
+    _lock = asyncio.Lock()
+    
+    def __new__(cls, monitor_paths=None):
+        if cls._instance is None:
+            cls._instance = super().__new__(cls)
+        return cls._instance
+    
+    def __init__(self, monitor_paths=None):
+        if not hasattr(self, '_initialized'):
+            if monitor_paths is None:
+                from ..config import config
+                monitor_paths = config.loras_roots
+                
+            super().__init__(monitor_paths)
+            self.handler = LoraFileHandler(self.loop)
+            self._initialized = True
+    
+    @classmethod
+    async def get_instance(cls):
+        """Get singleton instance with async support"""
+        async with cls._lock:
+            if cls._instance is None:
+                from ..config import config
+                cls._instance = cls(config.loras_roots)
+            return cls._instance
+
+
+class CheckpointFileMonitor(BaseFileMonitor):
+    """Monitor for checkpoint file changes"""
+    
+    _instance = None
+    _lock = asyncio.Lock()
+    
+    def __new__(cls, monitor_paths=None):
+        if cls._instance is None:
+            cls._instance = super().__new__(cls)
+        return cls._instance
+    
+    def __init__(self, monitor_paths=None):
+        if not hasattr(self, '_initialized'):
+            if monitor_paths is None:
+                # Get checkpoint roots from scanner
+                monitor_paths = []
+                # We'll initialize monitor paths later when scanner is available
+                
+            super().__init__(monitor_paths or [])
+            self.handler = CheckpointFileHandler(self.loop)
+            self._initialized = True
+    
+    @classmethod
+    async def get_instance(cls):
+        """Get singleton instance with async support"""
+        async with cls._lock:
+            if cls._instance is None:
+                cls._instance = cls([])
+                
+                # Now get checkpoint roots from scanner
+                from .checkpoint_scanner import CheckpointScanner
+                scanner = await CheckpointScanner.get_instance()
+                monitor_paths = scanner.get_model_roots()
+                
+                # Update monitor paths - but don't actually monitor them
+                for path in monitor_paths:
+                    real_path = os.path.realpath(path).replace(os.sep, '/')
+                    cls._instance.monitor_paths.add(real_path)
+                
+            return cls._instance
+    
+    def start(self):
+        """Override start to check global enable flag"""
+        if not ENABLE_FILE_MONITORING:
+            logger.info("Checkpoint file monitoring is disabled via ENABLE_FILE_MONITORING setting")
+            return
+            
+        logger.info("Checkpoint file monitoring is temporarily disabled")
+        # Skip the actual monitoring setup
+        pass
+    
+    async def initialize_paths(self):
+        """Initialize monitor paths from scanner - currently disabled"""
+        if not ENABLE_FILE_MONITORING:
+            logger.info("Checkpoint path initialization skipped (monitoring disabled)")
+            return
+            
+        logger.info("Checkpoint file path initialization skipped (monitoring disabled)")
+        pass

py/services/lora_hash_index.py

@@ -15,11 +15,13 @@ class LoraHashIndex:
         """Add or update a hash -> path mapping"""
         if not sha256 or not file_path:
             return
-        self._hash_to_path[sha256] = file_path
+        # Always store lowercase hashes for consistency
+        self._hash_to_path[sha256.lower()] = file_path
         
     def remove_entry(self, sha256: str) -> None:
         """Remove a hash entry"""
-        self._hash_to_path.pop(sha256, None)
+        if sha256:
+            self._hash_to_path.pop(sha256.lower(), None)
         
     def remove_by_path(self, file_path: str) -> None:
         """Remove entry by file path"""
@@ -30,7 +32,9 @@ class LoraHashIndex:
                 
     def get_path(self, sha256: str) -> Optional[str]:
         """Get file path for a given hash"""
-        return self._hash_to_path.get(sha256)
+        if not sha256:
+            return None
+        return self._hash_to_path.get(sha256.lower())
         
     def get_hash(self, file_path: str) -> Optional[str]:
         """Get hash for a given file path"""
@@ -41,7 +45,9 @@ class LoraHashIndex:
         
     def has_hash(self, sha256: str) -> bool:
         """Check if hash exists in index"""
-        return sha256 in self._hash_to_path
+        if not sha256:
+            return False
+        return sha256.lower() in self._hash_to_path
         
     def clear(self) -> None:
         """Clear all entries"""

py/services/lora_scanner.py

@@ -3,18 +3,22 @@ import os
 import logging
 import asyncio
 import shutil
-from typing import List, Dict, Optional
-from dataclasses import dataclass
-from operator import itemgetter
+import time
+from typing import List, Dict, Optional, Set
+
+from ..utils.models import LoraMetadata
 from ..config import config
-from ..utils.file_utils import load_metadata, get_file_info
-from .lora_cache import LoraCache
-from difflib import SequenceMatcher
+from .model_scanner import ModelScanner
 from .lora_hash_index import LoraHashIndex
+from .settings_manager import settings
+from ..utils.constants import NSFW_LEVELS
+from ..utils.utils import fuzzy_match
+from .service_registry import ServiceRegistry
+import sys
 
 logger = logging.getLogger(__name__)
 
-class LoraScanner:
+class LoraScanner(ModelScanner):
     """Service for scanning and managing LoRA files"""
     
     _instance = None
@@ -26,19 +30,20 @@ class LoraScanner:
         return cls._instance
     
     def __init__(self):
-        # 确保初始化只执行一次
+        # Ensure initialization happens only once
         if not hasattr(self, '_initialized'):
-            self._cache: Optional[LoraCache] = None
-            self._hash_index = LoraHashIndex()
-            self._initialization_lock = asyncio.Lock()
-            self._initialization_task: Optional[asyncio.Task] = None
+            # Define supported file extensions
+            file_extensions = {'.safetensors'}
+            
+            # Initialize parent class
+            super().__init__(
+                model_type="lora",
+                model_class=LoraMetadata, 
+                file_extensions=file_extensions,
+                hash_index=LoraHashIndex()
+            )
             self._initialized = True
-            self.file_monitor = None  # Add this line
-
-    def set_file_monitor(self, monitor):
-        """Set file monitor instance"""
-        self.file_monitor = monitor
-
+    
     @classmethod
     async def get_instance(cls):
         """Get singleton instance with async support"""
@@ -46,119 +51,78 @@ class LoraScanner:
             if cls._instance is None:
                 cls._instance = cls()
             return cls._instance
-
-    async def  get_cached_data(self, force_refresh: bool = False) -> LoraCache:
-        """Get cached LoRA data, refresh if needed"""
-        async with self._initialization_lock:
+    
+    def get_model_roots(self) -> List[str]:
+        """Get lora root directories"""
+        return config.loras_roots
+        
+    async def scan_all_models(self) -> List[Dict]:
+        """Scan all LoRA directories and return metadata"""
+        all_loras = []
+        
+        # Create scan tasks for each directory
+        scan_tasks = []
+        for lora_root in self.get_model_roots():
+            task = asyncio.create_task(self._scan_directory(lora_root))
+            scan_tasks.append(task)
             
-            # 如果缓存未初始化但需要响应请求，返回空缓存
-            if self._cache is None and not force_refresh:
-                return LoraCache(
-                    raw_data=[],
-                    sorted_by_name=[],
-                    sorted_by_date=[],
-                    folders=[]
-                )
-
-            # 如果正在初始化，等待完成
-            if self._initialization_task and not self._initialization_task.done():
-                try:
-                    await self._initialization_task
-                except Exception as e:
-                    logger.error(f"Cache initialization failed: {e}")
-                    self._initialization_task = None
-            
-            if (self._cache is None or force_refresh):
+        # Wait for all tasks to complete
+        for task in scan_tasks:
+            try:
+                loras = await task
+                all_loras.extend(loras)
+            except Exception as e:
+                logger.error(f"Error scanning directory: {e}")
                 
-                # 创建新的初始化任务
-                if not self._initialization_task or self._initialization_task.done():
-                    self._initialization_task = asyncio.create_task(self._initialize_cache())
+        return all_loras
+    
+    async def _scan_directory(self, root_path: str) -> List[Dict]:
+        """Scan a single directory for LoRA files"""
+        loras = []
+        original_root = root_path  # Save original root path
+        
+        async def scan_recursive(path: str, visited_paths: set):
+            """Recursively scan directory, avoiding circular symlinks"""
+            try:
+                real_path = os.path.realpath(path)
+                if real_path in visited_paths:
+                    logger.debug(f"Skipping already visited path: {path}")
+                    return
+                visited_paths.add(real_path)
                 
-                try:
-                    await self._initialization_task
-                except Exception as e:
-                    logger.error(f"Cache initialization failed: {e}")
-                    # 如果缓存已存在，继续使用旧缓存
-                    if self._cache is None:
-                        raise  # 如果没有缓存，则抛出异常
-            
-            return self._cache
+                with os.scandir(path) as it:
+                    entries = list(it)
+                    for entry in entries:
+                        try:
+                            if entry.is_file(follow_symlinks=True) and any(entry.name.endswith(ext) for ext in self.file_extensions):
+                                # Use original path instead of real path
+                                file_path = entry.path.replace(os.sep, "/")
+                                await self._process_single_file(file_path, original_root, loras)
+                                await asyncio.sleep(0)
+                            elif entry.is_dir(follow_symlinks=True):
+                                # For directories, continue scanning with original path
+                                await scan_recursive(entry.path, visited_paths)
+                        except Exception as e:
+                            logger.error(f"Error processing entry {entry.path}: {e}")
+            except Exception as e:
+                logger.error(f"Error scanning {path}: {e}")
 
-    async def _initialize_cache(self) -> None:
-        """Initialize or refresh the cache"""
+        await scan_recursive(root_path, set())
+        return loras
+
+    async def _process_single_file(self, file_path: str, root_path: str, loras: list):
+        """Process a single file and add to results list"""
         try:
-            # Clear existing hash index
-            self._hash_index.clear()
-            
-            # Scan for new data
-            raw_data = await self.scan_all_loras()
-            
-            # Build hash index
-            for lora_data in raw_data:
-                if 'sha256' in lora_data and 'file_path' in lora_data:
-                    self._hash_index.add_entry(lora_data['sha256'], lora_data['file_path'])
-            
-            # Update cache
-            self._cache = LoraCache(
-                raw_data=raw_data,
-                sorted_by_name=[],
-                sorted_by_date=[],
-                folders=[]
-            )
-            
-            # Call resort_cache to create sorted views
-            await self._cache.resort()
-
-            self._initialization_task = None
-            logger.info("LoRA Manager: Cache initialization completed")
+            result = await self._process_model_file(file_path, root_path)
+            if result:
+                loras.append(result)
         except Exception as e:
-            logger.error(f"LoRA Manager: Error initializing cache: {e}")
-            self._cache = LoraCache(
-                raw_data=[],
-                sorted_by_name=[],
-                sorted_by_date=[],
-                folders=[]
-            )
-
-    def fuzzy_match(self, text: str, pattern: str, threshold: float = 0.7) -> bool:
-        """
-        Check if text matches pattern using fuzzy matching.
-        Returns True if similarity ratio is above threshold.
-        """
-        if not pattern or not text:
-            return False
-        
-        # Convert both to lowercase for case-insensitive matching
-        text = text.lower()
-        pattern = pattern.lower()
-        
-        # Split pattern into words
-        search_words = pattern.split()
-        
-        # Check each word
-        for word in search_words:
-            # First check if word is a substring (faster)
-            if word in text:
-                continue
-            
-            # If not found as substring, try fuzzy matching
-            # Check if any part of the text matches this word
-            found_match = False
-            for text_part in text.split():
-                ratio = SequenceMatcher(None, text_part, word).ratio()
-                if ratio >= threshold:
-                    found_match = True
-                    break
-                
-            if not found_match:
-                return False
-        
-        # All words found either as substrings or fuzzy matches
-        return True
-
+            logger.error(f"Error processing {file_path}: {e}")
+    
     async def get_paginated_data(self, page: int, page_size: int, sort_by: str = 'name', 
-                               folder: str = None, search: str = None, fuzzy: bool = False,
-                               recursive: bool = False, base_models: list = None):
+                               folder: str = None, search: str = None, fuzzy_search: bool = False,
+                               base_models: list = None, tags: list = None,
+                               search_options: dict = None, hash_filters: dict = None) -> Dict:
         """Get paginated and filtered lora data
         
         Args:
@@ -167,59 +131,125 @@ class LoraScanner:
             sort_by: Sort method ('name' or 'date')
             folder: Filter by folder path
             search: Search term
-            fuzzy: Use fuzzy matching for search
-            recursive: Include subfolders when folder filter is applied
+            fuzzy_search: Use fuzzy matching for search
             base_models: List of base models to filter by
+            tags: List of tags to filter by
+            search_options: Dictionary with search options (filename, modelname, tags, recursive)
+            hash_filters: Dictionary with hash filtering options (single_hash or multiple_hashes)
         """
         cache = await self.get_cached_data()
 
-        # 先获取基础数据集
+        # Get default search options if not provided
+        if search_options is None:
+            search_options = {
+                'filename': True,
+                'modelname': True,
+                'tags': False,
+                'recursive': False,
+            }
+
+        # Get the base data set
         filtered_data = cache.sorted_by_date if sort_by == 'date' else cache.sorted_by_name
         
-        # 应用文件夹过滤
-        if folder is not None:
-            if recursive:
-                # 递归模式：匹配所有以该文件夹开头的路径
+        # Apply hash filtering if provided (highest priority)
+        if hash_filters:
+            single_hash = hash_filters.get('single_hash')
+            multiple_hashes = hash_filters.get('multiple_hashes')
+            
+            if single_hash:
+                # Filter by single hash
+                single_hash = single_hash.lower()  # Ensure lowercase for matching
                 filtered_data = [
-                    item for item in filtered_data 
-                    if item['folder'].startswith(folder + '/') or item['folder'] == folder
+                    lora for lora in filtered_data
+                    if lora.get('sha256', '').lower() == single_hash
+                ]
+            elif multiple_hashes:
+                # Filter by multiple hashes
+                hash_set = set(hash.lower() for hash in multiple_hashes)  # Convert to set for faster lookup
+                filtered_data = [
+                    lora for lora in filtered_data
+                    if lora.get('sha256', '').lower() in hash_set
+                ]
+            
+
+            # Jump to pagination
+            total_items = len(filtered_data)
+            start_idx = (page - 1) * page_size
+            end_idx = min(start_idx + page_size, total_items)
+            
+            result = {
+                'items': filtered_data[start_idx:end_idx],
+                'total': total_items,
+                'page': page,
+                'page_size': page_size,
+                'total_pages': (total_items + page_size - 1) // page_size
+            }
+            
+            return result
+        
+        # Apply SFW filtering if enabled
+        if settings.get('show_only_sfw', False):
+            filtered_data = [
+                lora for lora in filtered_data
+                if not lora.get('preview_nsfw_level') or lora.get('preview_nsfw_level') < NSFW_LEVELS['R']
+            ]
+        
+        # Apply folder filtering
+        if folder is not None:
+            if search_options.get('recursive', False):
+                # Recursive folder filtering - include all subfolders
+                filtered_data = [
+                    lora for lora in filtered_data
+                    if lora['folder'].startswith(folder)
                 ]
             else:
-                # 非递归模式：只匹配确切的文件夹
+                # Exact folder filtering
                 filtered_data = [
-                    item for item in filtered_data 
-                    if item['folder'] == folder
+                    lora for lora in filtered_data
+                    if lora['folder'] == folder
                 ]
         
         # Apply base model filtering
         if base_models and len(base_models) > 0:
             filtered_data = [
-                item for item in filtered_data
-                if item.get('base_model') in base_models
+                lora for lora in filtered_data
+                if lora.get('base_model') in base_models
             ]
         
-        # 应用搜索过滤
+        # Apply tag filtering
+        if tags and len(tags) > 0:
+            filtered_data = [
+                lora for lora in filtered_data
+                if any(tag in lora.get('tags', []) for tag in tags)
+            ]
+        
+        # Apply search filtering
         if search:
-            if fuzzy:
-                filtered_data = [
-                    item for item in filtered_data 
-                    if any(
-                        self.fuzzy_match(str(value), search) 
-                        for value in [
-                            item.get('model_name', ''),
-                            item.get('base_model', '')
-                        ]
-                        if value
-                    )
-                ]
-            else:
-                # Original exact search logic
-                filtered_data = [
-                    item for item in filtered_data 
-                    if search in str(item.get('model_name', '')).lower()
-                ]
+            search_results = []
+            search_opts = search_options or {}
+            
+            for lora in filtered_data:
+                # Search by file name
+                if search_opts.get('filename', True):
+                    if fuzzy_match(lora.get('file_name', ''), search):
+                        search_results.append(lora)
+                        continue
+                        
+                # Search by model name
+                if search_opts.get('modelname', True):
+                    if fuzzy_match(lora.get('model_name', ''), search):
+                        search_results.append(lora)
+                        continue
+                        
+                # Search by tags
+                if search_opts.get('tags', False) and 'tags' in lora:
+                    if any(fuzzy_match(tag, search) for tag in lora['tags']):
+                        search_results.append(lora)
+                        continue
+            
+            filtered_data = search_results
 
-        # 计算分页
+        # Calculate pagination
         total_items = len(filtered_data)
         start_idx = (page - 1) * page_size
         end_idx = min(start_idx + page_size, total_items)
@@ -234,238 +264,6 @@ class LoraScanner:
         
         return result
 
-    def invalidate_cache(self):
-        """Invalidate the current cache"""
-        self._cache = None
-
-    async def scan_all_loras(self) -> List[Dict]:
-        """Scan all LoRA directories and return metadata"""
-        all_loras = []
-        
-        # 分目录异步扫描
-        scan_tasks = []
-        for loras_root in config.loras_roots:
-            task = asyncio.create_task(self._scan_directory(loras_root))
-            scan_tasks.append(task)
-            
-        for task in scan_tasks:
-            try:
-                loras = await task
-                all_loras.extend(loras)
-            except Exception as e:
-                logger.error(f"Error scanning directory: {e}")
-                
-        return all_loras
-
-    async def _scan_directory(self, root_path: str) -> List[Dict]:
-        """Scan a single directory for LoRA files"""
-        loras = []
-        original_root = root_path  # 保存原始根路径
-        
-        async def scan_recursive(path: str, visited_paths: set):
-            """递归扫描目录，避免循环链接"""
-            try:
-                real_path = os.path.realpath(path)
-                if real_path in visited_paths:
-                    logger.debug(f"Skipping already visited path: {path}")
-                    return
-                visited_paths.add(real_path)
-                
-                with os.scandir(path) as it:
-                    entries = list(it)
-                    for entry in entries:
-                        try:
-                            if entry.is_file(follow_symlinks=True) and entry.name.endswith('.safetensors'):
-                                # 使用原始路径而不是真实路径
-                                file_path = entry.path.replace(os.sep, "/")
-                                await self._process_single_file(file_path, original_root, loras)
-                                await asyncio.sleep(0)
-                            elif entry.is_dir(follow_symlinks=True):
-                                # 对于目录，使用原始路径继续扫描
-                                await scan_recursive(entry.path, visited_paths)
-                        except Exception as e:
-                            logger.error(f"Error processing entry {entry.path}: {e}")
-            except Exception as e:
-                logger.error(f"Error scanning {path}: {e}")
-
-        await scan_recursive(root_path, set())
-        return loras
-
-    async def _process_single_file(self, file_path: str, root_path: str, loras: list):
-        """处理单个文件并添加到结果列表"""
-        try:
-            result = await self._process_lora_file(file_path, root_path)
-            if result:
-                loras.append(result)
-        except Exception as e:
-            logger.error(f"Error processing {file_path}: {e}")
-
-    async def _process_lora_file(self, file_path: str, root_path: str) -> Dict:
-        """Process a single LoRA file and return its metadata"""
-        # Try loading existing metadata
-        metadata = await load_metadata(file_path)
-        
-        if metadata is None:
-            # Create new metadata if none exists
-            metadata = await get_file_info(file_path)
-        
-        # Convert to dict and add folder info
-        lora_data = metadata.to_dict()
-        rel_path = os.path.relpath(file_path, root_path)
-        folder = os.path.dirname(rel_path)
-        lora_data['folder'] = folder.replace(os.path.sep, '/')
-        
-        return lora_data
-
-    async def update_preview_in_cache(self, file_path: str, preview_url: str) -> bool:
-        """Update preview URL in cache for a specific lora
-        
-        Args:
-            file_path: The file path of the lora to update
-            preview_url: The new preview URL
-            
-        Returns:
-            bool: True if the update was successful, False if cache doesn't exist or lora wasn't found
-        """
-        if self._cache is None:
-            return False
-
-        return await self._cache.update_preview_url(file_path, preview_url)
-
-    async def scan_single_lora(self, file_path: str) -> Optional[Dict]:
-        """Scan a single LoRA file and return its metadata"""
-        try:
-            if not os.path.exists(os.path.realpath(file_path)):
-                return None
-                
-            # 获取基本文件信息
-            metadata = await get_file_info(file_path)
-            if not metadata:
-                return None
-                
-            folder = self._calculate_folder(file_path)
-                    
-            # 确保 folder 字段存在
-            metadata_dict = metadata.to_dict()
-            metadata_dict['folder'] = folder or ''
-            
-            return metadata_dict
-            
-        except Exception as e:
-            logger.error(f"Error scanning {file_path}: {e}")
-            return None
-    
-    def _calculate_folder(self, file_path: str) -> str:
-        """Calculate the folder path for a LoRA file"""
-        # 使用原始路径计算相对路径
-        for root in config.loras_roots:
-            if file_path.startswith(root):
-                rel_path = os.path.relpath(file_path, root)
-                return os.path.dirname(rel_path).replace(os.path.sep, '/')
-        return ''
-
-    async def move_model(self, source_path: str, target_path: str) -> bool:
-        """Move a model and its associated files to a new location"""
-        try:
-            # 保持原始路径格式
-            source_path = source_path.replace(os.sep, '/')
-            target_path = target_path.replace(os.sep, '/')
-            
-            # 其余代码保持不变
-            base_name = os.path.splitext(os.path.basename(source_path))[0]
-            source_dir = os.path.dirname(source_path)
-            
-            os.makedirs(target_path, exist_ok=True)
-            
-            target_lora = os.path.join(target_path, f"{base_name}.safetensors").replace(os.sep, '/')
-
-            # 使用真实路径进行文件操作
-            real_source = os.path.realpath(source_path)
-            real_target = os.path.realpath(target_lora)
-            
-            file_size = os.path.getsize(real_source)
-            
-            if self.file_monitor:
-                self.file_monitor.handler.add_ignore_path(
-                    real_source,
-                    file_size
-                )
-                self.file_monitor.handler.add_ignore_path(
-                    real_target,
-                    file_size
-                )
-            
-            # 使用真实路径进行文件操作
-            shutil.move(real_source, real_target)
-            
-            # Move associated files
-            source_metadata = os.path.join(source_dir, f"{base_name}.metadata.json")
-            if os.path.exists(source_metadata):
-                target_metadata = os.path.join(target_path, f"{base_name}.metadata.json")
-                shutil.move(source_metadata, target_metadata)
-                metadata = await self._update_metadata_paths(target_metadata, target_lora)
-            
-            # Move preview file if exists
-            preview_extensions = ['.preview.png', '.preview.jpeg', '.preview.jpg', '.preview.mp4',
-                               '.png', '.jpeg', '.jpg', '.mp4']
-            for ext in preview_extensions:
-                source_preview = os.path.join(source_dir, f"{base_name}{ext}")
-                if os.path.exists(source_preview):
-                    target_preview = os.path.join(target_path, f"{base_name}{ext}")
-                    shutil.move(source_preview, target_preview)
-                    break
-            
-            # Update cache
-            await self.update_single_lora_cache(source_path, target_lora, metadata)
-            
-            return True
-            
-        except Exception as e:
-            logger.error(f"Error moving model: {e}", exc_info=True)
-            return False
-        
-    async def update_single_lora_cache(self, original_path: str, new_path: str, metadata: Dict) -> bool:
-        cache = await self.get_cached_data()
-        
-        # Remove old path from hash index if exists
-        self._hash_index.remove_by_path(original_path)
-        
-        # Remove the old entry from raw_data
-        cache.raw_data = [
-            item for item in cache.raw_data 
-            if item['file_path'] != original_path
-        ]
-        
-        if metadata:
-            # If this is an update to an existing path (not a move), ensure folder is preserved
-            if original_path == new_path:
-                # Find the folder from existing entries or calculate it
-                existing_folder = next((item['folder'] for item in cache.raw_data 
-                                      if item['file_path'] == original_path), None)
-                if existing_folder:
-                    metadata['folder'] = existing_folder
-                else:
-                    metadata['folder'] = self._calculate_folder(new_path)
-            else:
-                # For moved files, recalculate the folder
-                metadata['folder'] = self._calculate_folder(new_path)
-            
-            # Add the updated metadata to raw_data
-            cache.raw_data.append(metadata)
-            
-            # Update hash index with new path
-            if 'sha256' in metadata:
-                self._hash_index.add_entry(metadata['sha256'], new_path)
-            
-            # Update folders list
-            all_folders = set(item['folder'] for item in cache.raw_data)
-            cache.folders = sorted(list(all_folders), key=lambda x: x.lower())
-        
-        # Resort cache
-        await cache.resort()
-        
-        return True
-
     async def _update_metadata_paths(self, metadata_path: str, lora_path: str) -> Dict:
         """Update file paths in metadata file"""
         try:
@@ -492,16 +290,101 @@ class LoraScanner:
         except Exception as e:
             logger.error(f"Error updating metadata paths: {e}", exc_info=True)
 
-    # Add new methods for hash index functionality
+    # Lora-specific hash index functionality
     def has_lora_hash(self, sha256: str) -> bool:
         """Check if a LoRA with given hash exists"""
-        return self._hash_index.has_hash(sha256)
+        return self.has_hash(sha256)
         
     def get_lora_path_by_hash(self, sha256: str) -> Optional[str]:
         """Get file path for a LoRA by its hash"""
-        return self._hash_index.get_path(sha256)
+        return self.get_path_by_hash(sha256)
         
     def get_lora_hash_by_path(self, file_path: str) -> Optional[str]:
         """Get hash for a LoRA by its file path"""
-        return self._hash_index.get_hash(file_path)
+        return self.get_hash_by_path(file_path)
+
+    async def get_top_tags(self, limit: int = 20) -> List[Dict[str, any]]:
+        """Get top tags sorted by count"""
+        # Make sure cache is initialized
+        await self.get_cached_data()
+        
+        # Sort tags by count in descending order
+        sorted_tags = sorted(
+            [{"tag": tag, "count": count} for tag, count in self._tags_count.items()],
+            key=lambda x: x['count'],
+            reverse=True
+        )
+        
+        # Return limited number
+        return sorted_tags[:limit]
+        
+    async def get_base_models(self, limit: int = 20) -> List[Dict[str, any]]:
+        """Get base models used in loras sorted by frequency"""
+        # Make sure cache is initialized
+        cache = await self.get_cached_data()
+        
+        # Count base model occurrences
+        base_model_counts = {}
+        for lora in cache.raw_data:
+            if 'base_model' in lora and lora['base_model']:
+                base_model = lora['base_model']
+                base_model_counts[base_model] = base_model_counts.get(base_model, 0) + 1
+        
+        # Sort base models by count
+        sorted_models = [{'name': model, 'count': count} for model, count in base_model_counts.items()]
+        sorted_models.sort(key=lambda x: x['count'], reverse=True)
+        
+        # Return limited number
+        return sorted_models[:limit]
+
+    async def diagnose_hash_index(self):
+        """Diagnostic method to verify hash index functionality"""
+        print("\n\n*** DIAGNOSING LORA HASH INDEX ***\n\n", file=sys.stderr)
+        
+        # First check if the hash index has any entries
+        if hasattr(self, '_hash_index'):
+            index_entries = len(self._hash_index._hash_to_path)
+            print(f"Hash index has {index_entries} entries", file=sys.stderr)
+            
+            # Print a few example entries if available
+            if index_entries > 0:
+                print("\nSample hash index entries:", file=sys.stderr)
+                count = 0
+                for hash_val, path in self._hash_index._hash_to_path.items():
+                    if count < 5:  # Just show the first 5
+                        print(f"Hash: {hash_val[:8]}... -> Path: {path}", file=sys.stderr)
+                        count += 1
+                    else:
+                        break
+        else:
+            print("Hash index not initialized", file=sys.stderr)
+        
+        # Try looking up by a known hash for testing
+        if not hasattr(self, '_hash_index') or not self._hash_index._hash_to_path:
+            print("No hash entries to test lookup with", file=sys.stderr)
+            return
+        
+        test_hash = next(iter(self._hash_index._hash_to_path.keys()))
+        test_path = self._hash_index.get_path(test_hash)
+        print(f"\nTest lookup by hash: {test_hash[:8]}... -> {test_path}", file=sys.stderr)
+        
+        # Also test reverse lookup
+        test_hash_result = self._hash_index.get_hash(test_path)
+        print(f"Test reverse lookup: {test_path} -> {test_hash_result[:8]}...\n\n", file=sys.stderr)
+
+    async def get_lora_info_by_name(self, name):
+        """Get LoRA information by name"""
+        try:
+            # Get cached data
+            cache = await self.get_cached_data()
+            
+            # Find the LoRA by name
+            for lora in cache.raw_data:
+                if lora.get("file_name") == name:
+                    return lora
+                    
+            return None
+        except Exception as e:
+            logger.error(f"Error getting LoRA info by name: {e}", exc_info=True)
+            return None
 

py/services/model_cache.py

@@ -0,0 +1,64 @@
+import asyncio
+from typing import List, Dict
+from dataclasses import dataclass
+from operator import itemgetter
+
+@dataclass
+class ModelCache:
+    """Cache structure for model data"""
+    raw_data: List[Dict]
+    sorted_by_name: List[Dict]
+    sorted_by_date: List[Dict]
+    folders: List[str]
+    
+    def __post_init__(self):
+        self._lock = asyncio.Lock()
+
+    async def resort(self, name_only: bool = False):
+        """Resort all cached data views"""
+        async with self._lock:
+            self.sorted_by_name = sorted(
+                self.raw_data, 
+                key=lambda x: x['model_name'].lower()  # Case-insensitive sort
+            )
+            if not name_only:
+                self.sorted_by_date = sorted(
+                    self.raw_data, 
+                    key=itemgetter('modified'), 
+                    reverse=True
+                )
+            # Update folder list
+            all_folders = set(l['folder'] for l in self.raw_data)
+            self.folders = sorted(list(all_folders), key=lambda x: x.lower())
+
+    async def update_preview_url(self, file_path: str, preview_url: str) -> bool:
+        """Update preview_url for a specific model in all cached data
+        
+        Args:
+            file_path: The file path of the model to update
+            preview_url: The new preview URL
+            
+        Returns:
+            bool: True if the update was successful, False if the model wasn't found
+        """
+        async with self._lock:
+            # Update in raw_data
+            for item in self.raw_data:
+                if item['file_path'] == file_path:
+                    item['preview_url'] = preview_url
+                    break
+            else:
+                return False  # Model not found
+                
+            # Update in sorted lists (references to the same dict objects)
+            for item in self.sorted_by_name:
+                if item['file_path'] == file_path:
+                    item['preview_url'] = preview_url
+                    break
+                    
+            for item in self.sorted_by_date:
+                if item['file_path'] == file_path:
+                    item['preview_url'] = preview_url
+                    break
+                    
+            return True

py/services/model_hash_index.py

@@ -0,0 +1,78 @@
+from typing import Dict, Optional, Set
+
+class ModelHashIndex:
+    """Index for looking up models by hash or path"""
+    
+    def __init__(self):
+        self._hash_to_path: Dict[str, str] = {}
+        self._path_to_hash: Dict[str, str] = {}
+    
+    def add_entry(self, sha256: str, file_path: str) -> None:
+        """Add or update hash index entry"""
+        if not sha256 or not file_path:
+            return
+            
+        # Ensure hash is lowercase for consistency
+        sha256 = sha256.lower()
+        
+        # Remove old path mapping if hash exists
+        if sha256 in self._hash_to_path:
+            old_path = self._hash_to_path[sha256]
+            if old_path in self._path_to_hash:
+                del self._path_to_hash[old_path]
+        
+        # Remove old hash mapping if path exists
+        if file_path in self._path_to_hash:
+            old_hash = self._path_to_hash[file_path]
+            if old_hash in self._hash_to_path:
+                del self._hash_to_path[old_hash]
+        
+        # Add new mappings
+        self._hash_to_path[sha256] = file_path
+        self._path_to_hash[file_path] = sha256
+    
+    def remove_by_path(self, file_path: str) -> None:
+        """Remove entry by file path"""
+        if file_path in self._path_to_hash:
+            hash_val = self._path_to_hash[file_path]
+            if hash_val in self._hash_to_path:
+                del self._hash_to_path[hash_val]
+            del self._path_to_hash[file_path]
+    
+    def remove_by_hash(self, sha256: str) -> None:
+        """Remove entry by hash"""
+        sha256 = sha256.lower()
+        if sha256 in self._hash_to_path:
+            path = self._hash_to_path[sha256]
+            if path in self._path_to_hash:
+                del self._path_to_hash[path]
+            del self._hash_to_path[sha256]
+    
+    def has_hash(self, sha256: str) -> bool:
+        """Check if hash exists in index"""
+        return sha256.lower() in self._hash_to_path
+    
+    def get_path(self, sha256: str) -> Optional[str]:
+        """Get file path for a hash"""
+        return self._hash_to_path.get(sha256.lower())
+    
+    def get_hash(self, file_path: str) -> Optional[str]:
+        """Get hash for a file path"""
+        return self._path_to_hash.get(file_path)
+    
+    def clear(self) -> None:
+        """Clear all entries"""
+        self._hash_to_path.clear()
+        self._path_to_hash.clear()
+    
+    def get_all_hashes(self) -> Set[str]:
+        """Get all hashes in the index"""
+        return set(self._hash_to_path.keys())
+    
+    def get_all_paths(self) -> Set[str]:
+        """Get all file paths in the index"""
+        return set(self._path_to_hash.keys())
+    
+    def __len__(self) -> int:
+        """Get number of entries"""
+        return len(self._hash_to_path)

py/services/model_scanner.py

@@ -0,0 +1,879 @@
+import json
+import os
+import logging
+import asyncio
+import time
+import shutil
+from typing import List, Dict, Optional, Type, Set
+
+from ..utils.models import BaseModelMetadata
+from ..config import config
+from ..utils.file_utils import load_metadata, get_file_info, find_preview_file, save_metadata
+from .model_cache import ModelCache
+from .model_hash_index import ModelHashIndex
+from ..utils.constants import PREVIEW_EXTENSIONS
+from .service_registry import ServiceRegistry
+from .websocket_manager import ws_manager
+
+logger = logging.getLogger(__name__)
+
+class ModelScanner:
+    """Base service for scanning and managing model files"""
+    
+    _lock = asyncio.Lock()
+    
+    def __init__(self, model_type: str, model_class: Type[BaseModelMetadata], file_extensions: Set[str], hash_index: Optional[ModelHashIndex] = None):
+        """Initialize the scanner
+        
+        Args:
+            model_type: Type of model (lora, checkpoint, etc.)
+            model_class: Class used to create metadata instances
+            file_extensions: Set of supported file extensions including the dot (e.g. {'.safetensors'})
+            hash_index: Hash index instance (optional)
+        """
+        self.model_type = model_type
+        self.model_class = model_class
+        self.file_extensions = file_extensions
+        self._cache = None
+        self._hash_index = hash_index or ModelHashIndex()
+        self._tags_count = {}  # Dictionary to store tag counts
+        self._is_initializing = False  # Flag to track initialization state
+        
+        # Register this service
+        asyncio.create_task(self._register_service())
+        
+    async def _register_service(self):
+        """Register this instance with the ServiceRegistry"""
+        service_name = f"{self.model_type}_scanner"
+        await ServiceRegistry.register_service(service_name, self)
+    
+    async def initialize_in_background(self) -> None:
+        """Initialize cache in background using thread pool"""
+        try:
+            # Set initial empty cache to avoid None reference errors
+            if self._cache is None:
+                self._cache = ModelCache(
+                    raw_data=[],
+                    sorted_by_name=[],
+                    sorted_by_date=[],
+                    folders=[]
+                )
+            
+            # Set initializing flag to true
+            self._is_initializing = True
+            
+            # Determine the page type based on model type
+            page_type = 'loras' if self.model_type == 'lora' else 'checkpoints'
+            
+            # First, count all model files to track progress
+            await ws_manager.broadcast_init_progress({
+                'stage': 'scan_folders',
+                'progress': 0,
+                'details': f"Scanning {self.model_type} folders...",
+                'scanner_type': self.model_type,
+                'pageType': page_type
+            })
+            
+            # Count files in a separate thread to avoid blocking
+            loop = asyncio.get_event_loop()
+            total_files = await loop.run_in_executor(
+                None,  # Use default thread pool
+                self._count_model_files  # Run file counting in thread
+            )
+            
+            await ws_manager.broadcast_init_progress({
+                'stage': 'count_models',
+                'progress': 1, # Changed from 10 to 1
+                'details': f"Found {total_files} {self.model_type} files",
+                'scanner_type': self.model_type,
+                'pageType': page_type
+            })
+            
+            start_time = time.time()
+            
+            # Use thread pool to execute CPU-intensive operations with progress reporting
+            await loop.run_in_executor(
+                None,  # Use default thread pool
+                self._initialize_cache_sync,  # Run synchronous version in thread
+                total_files,  # Pass the total file count for progress reporting
+                page_type  # Pass the page type for progress reporting
+            )
+            
+            # Send final progress update
+            await ws_manager.broadcast_init_progress({
+                'stage': 'finalizing',
+                'progress': 99, # Changed from 95 to 99
+                'details': f"Finalizing {self.model_type} cache...",
+                'scanner_type': self.model_type,
+                'pageType': page_type
+            })
+            
+            logger.info(f"{self.model_type.capitalize()} cache initialized in {time.time() - start_time:.2f} seconds. Found {len(self._cache.raw_data)} models")
+            
+            # Send completion message
+            await asyncio.sleep(0.5)  # Small delay to ensure final progress message is sent
+            await ws_manager.broadcast_init_progress({
+                'stage': 'finalizing',
+                'progress': 100,
+                'status': 'complete',
+                'details': f"Completed! Found {len(self._cache.raw_data)} {self.model_type} files.",
+                'scanner_type': self.model_type,
+                'pageType': page_type
+            })
+            
+        except Exception as e:
+            logger.error(f"{self.model_type.capitalize()} Scanner: Error initializing cache in background: {e}")
+        finally:
+            # Always clear the initializing flag when done
+            self._is_initializing = False
+    
+    def _count_model_files(self) -> int:
+        """Count all model files with supported extensions in all roots
+        
+        Returns:
+            int: Total number of model files found
+        """
+        total_files = 0
+        visited_real_paths = set()
+        
+        for root_path in self.get_model_roots():
+            if not os.path.exists(root_path):
+                continue
+                
+            def count_recursive(path):
+                nonlocal total_files
+                try:
+                    real_path = os.path.realpath(path)
+                    if real_path in visited_real_paths:
+                        return
+                    visited_real_paths.add(real_path)
+                    
+                    with os.scandir(path) as it:
+                        for entry in it:
+                            try:
+                                if entry.is_file(follow_symlinks=True):
+                                    ext = os.path.splitext(entry.name)[1].lower()
+                                    if ext in self.file_extensions:
+                                        total_files += 1
+                                elif entry.is_dir(follow_symlinks=True):
+                                    count_recursive(entry.path)
+                            except Exception as e:
+                                logger.error(f"Error counting files in entry {entry.path}: {e}")
+                except Exception as e:
+                    logger.error(f"Error counting files in {path}: {e}")
+            
+            count_recursive(root_path)
+        
+        return total_files
+    
+    def _initialize_cache_sync(self, total_files=0, page_type='loras'):
+        """Synchronous version of cache initialization for thread pool execution"""
+        try:
+            # Create a new event loop for this thread
+            loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(loop)
+            
+            # Create a synchronous method to bypass the async lock
+            def sync_initialize_cache():
+                # Track progress
+                processed_files = 0
+                last_progress_time = time.time()
+                last_progress_percent = 0
+                
+                # We need a wrapper around scan_all_models to track progress
+                # This is a local function that will run in our thread's event loop
+                async def scan_with_progress():
+                    nonlocal processed_files, last_progress_time, last_progress_percent
+                    
+                    # For storing raw model data
+                    all_models = []
+                    
+                    # Process each model root
+                    for root_path in self.get_model_roots():
+                        if not os.path.exists(root_path):
+                            continue
+                            
+                        # Track visited paths to avoid symlink loops
+                        visited_paths = set()
+                        
+                        # Recursively process directory
+                        async def scan_dir_with_progress(path):
+                            nonlocal processed_files, last_progress_time, last_progress_percent
+                            
+                            try:
+                                real_path = os.path.realpath(path)
+                                if real_path in visited_paths:
+                                    return
+                                visited_paths.add(real_path)
+                                
+                                with os.scandir(path) as it:
+                                    entries = list(it)
+                                    for entry in entries:
+                                        try:
+                                            if entry.is_file(follow_symlinks=True):
+                                                ext = os.path.splitext(entry.name)[1].lower()
+                                                if ext in self.file_extensions:
+                                                    file_path = entry.path.replace(os.sep, "/")
+                                                    result = await self._process_model_file(file_path, root_path)
+                                                    if result:
+                                                        all_models.append(result)
+                                                    
+                                                    # Update progress counter
+                                                    processed_files += 1
+                                                    
+                                                    # Update progress periodically (not every file to avoid excessive updates)
+                                                    current_time = time.time()
+                                                    if total_files > 0 and (current_time - last_progress_time > 0.5 or processed_files == total_files):
+                                                        # Adjusted progress calculation
+                                                        progress_percent = min(99, int(1 + (processed_files / total_files) * 98))
+                                                        if progress_percent > last_progress_percent:
+                                                            last_progress_percent = progress_percent
+                                                            last_progress_time = current_time
+                                                            
+                                                            # Send progress update through websocket
+                                                            await ws_manager.broadcast_init_progress({
+                                                                'stage': 'process_models',
+                                                                'progress': progress_percent,
+                                                                'details': f"Processing {self.model_type} files: {processed_files}/{total_files}",
+                                                                'scanner_type': self.model_type,
+                                                                'pageType': page_type
+                                                            })
+                                            
+                                            elif entry.is_dir(follow_symlinks=True):
+                                                await scan_dir_with_progress(entry.path)
+                                                
+                                        except Exception as e:
+                                            logger.error(f"Error processing entry {entry.path}: {e}")
+                            except Exception as e:
+                                logger.error(f"Error scanning {path}: {e}")
+                        
+                        # Process the root path
+                        await scan_dir_with_progress(root_path)
+                    
+                    return all_models
+                
+                # Run the progress-tracking scan function
+                raw_data = loop.run_until_complete(scan_with_progress())
+                
+                # Update hash index and tags count
+                for model_data in raw_data:
+                    if 'sha256' in model_data and 'file_path' in model_data:
+                        self._hash_index.add_entry(model_data['sha256'].lower(), model_data['file_path'])
+                    
+                    # Count tags
+                    if 'tags' in model_data and model_data['tags']:
+                        for tag in model_data['tags']:
+                            self._tags_count[tag] = self._tags_count.get(tag, 0) + 1
+                
+                # Update cache
+                self._cache.raw_data = raw_data
+                loop.run_until_complete(self._cache.resort())
+                
+                return self._cache
+            
+            # Run our sync initialization that avoids lock conflicts
+            return sync_initialize_cache()
+        except Exception as e:
+            logger.error(f"Error in thread-based {self.model_type} cache initialization: {e}")
+        finally:
+            # Clean up the event loop
+            loop.close()
+
+    async def get_cached_data(self, force_refresh: bool = False) -> ModelCache:
+        """Get cached model data, refresh if needed"""
+        # If cache is not initialized, return an empty cache
+        # Actual initialization should be done via initialize_in_background
+        if self._cache is None and not force_refresh:
+            return ModelCache(
+                raw_data=[],
+                sorted_by_name=[],
+                sorted_by_date=[],
+                folders=[]
+            )
+
+        # If force refresh is requested, initialize the cache directly
+        if force_refresh:
+            if self._cache is None:
+                # For initial creation, do a full initialization
+                await self._initialize_cache()
+            else:
+                # For subsequent refreshes, use fast reconciliation
+                await self._reconcile_cache()
+        
+        return self._cache
+
+    async def _initialize_cache(self) -> None:
+        """Initialize or refresh the cache"""
+        self._is_initializing = True  # Set flag
+        try:
+            start_time = time.time()
+            # Clear existing hash index
+            self._hash_index.clear()
+            
+            # Clear existing tags count
+            self._tags_count = {}
+            
+            # Determine the page type based on model type
+            page_type = 'loras' if self.model_type == 'lora' else 'checkpoints'
+            
+            # Scan for new data
+            raw_data = await self.scan_all_models()
+            
+            # Build hash index and tags count
+            for model_data in raw_data:
+                if 'sha256' in model_data and 'file_path' in model_data:
+                    self._hash_index.add_entry(model_data['sha256'].lower(), model_data['file_path'])
+                
+                # Count tags
+                if 'tags' in model_data and model_data['tags']:
+                    for tag in model_data['tags']:
+                        self._tags_count[tag] = self._tags_count.get(tag, 0) + 1
+            
+            # Update cache
+            self._cache = ModelCache(
+                raw_data=raw_data,
+                sorted_by_name=[],
+                sorted_by_date=[],
+                folders=[]
+            )
+            
+            # Resort cache
+            await self._cache.resort()
+
+            logger.info(f"{self.model_type.capitalize()} Scanner: Cache initialization completed in {time.time() - start_time:.2f} seconds, found {len(raw_data)} models")
+        except Exception as e:
+            logger.error(f"{self.model_type.capitalize()} Scanner: Error initializing cache: {e}")
+            # Ensure cache is at least an empty structure on error
+            if self._cache is None:
+                self._cache = ModelCache(
+                    raw_data=[],
+                    sorted_by_name=[],
+                    sorted_by_date=[],
+                    folders=[]
+                )
+        finally:
+            self._is_initializing = False # Unset flag
+
+    async def _reconcile_cache(self) -> None:
+        """Fast cache reconciliation - only process differences between cache and filesystem"""
+        self._is_initializing = True # Set flag for reconciliation duration
+        try:
+            start_time = time.time()
+            logger.info(f"{self.model_type.capitalize()} Scanner: Starting fast cache reconciliation...")
+            
+            # Get current cached file paths
+            cached_paths = {item['file_path'] for item in self._cache.raw_data}
+            path_to_item = {item['file_path']: item for item in self._cache.raw_data}
+            
+            # Track found files and new files
+            found_paths = set()
+            new_files = []
+            
+            # Scan all model roots
+            for root_path in self.get_model_roots():
+                if not os.path.exists(root_path):
+                    continue
+                    
+                # Track visited real paths to avoid symlink loops
+                visited_real_paths = set()
+                
+                # Recursively scan directory
+                for root, _, files in os.walk(root_path, followlinks=True):
+                    real_root = os.path.realpath(root)
+                    if real_root in visited_real_paths:
+                        continue
+                    visited_real_paths.add(real_root)
+                    
+                    for file in files:
+                        ext = os.path.splitext(file)[1].lower()
+                        if ext in self.file_extensions:
+                            # Construct paths exactly as they would be in cache
+                            file_path = os.path.join(root, file).replace(os.sep, '/')
+                            
+                            # Check if this file is already in cache
+                            if file_path in cached_paths:
+                                found_paths.add(file_path)
+                                continue
+                                
+                            # Try case-insensitive match on Windows
+                            if os.name == 'nt':
+                                lower_path = file_path.lower()
+                                matched = False
+                                for cached_path in cached_paths:
+                                    if cached_path.lower() == lower_path:
+                                        found_paths.add(cached_path)
+                                        matched = True
+                                        break
+                                if matched:
+                                    continue
+                            
+                            # This is a new file to process
+                            new_files.append(file_path)
+                    
+                    # Yield control periodically
+                    await asyncio.sleep(0)
+            
+            # Process new files in batches
+            total_added = 0
+            if new_files:
+                logger.info(f"{self.model_type.capitalize()} Scanner: Found {len(new_files)} new files to process")
+                batch_size = 50
+                for i in range(0, len(new_files), batch_size):
+                    batch = new_files[i:i+batch_size]
+                    for path in batch:
+                        try:
+                            model_data = await self.scan_single_model(path)
+                            if model_data:
+                                # Add to cache
+                                self._cache.raw_data.append(model_data)
+                                
+                                # Update hash index if available
+                                if 'sha256' in model_data and 'file_path' in model_data:
+                                    self._hash_index.add_entry(model_data['sha256'].lower(), model_data['file_path'])
+                                
+                                # Update tags count
+                                if 'tags' in model_data and model_data['tags']:
+                                    for tag in model_data['tags']:
+                                        self._tags_count[tag] = self._tags_count.get(tag, 0) + 1
+                                        
+                                total_added += 1
+                        except Exception as e:
+                            logger.error(f"Error adding {path} to cache: {e}")
+                    
+                    # Yield control after each batch
+                    await asyncio.sleep(0)
+            
+            # Find missing files (in cache but not in filesystem)
+            missing_files = cached_paths - found_paths
+            total_removed = 0
+            
+            if missing_files:
+                logger.info(f"{self.model_type.capitalize()} Scanner: Found {len(missing_files)} files to remove from cache")
+                
+                # Process files to remove
+                for path in missing_files:
+                    try:
+                        model_to_remove = path_to_item[path]
+                        
+                        # Update tags count
+                        for tag in model_to_remove.get('tags', []):
+                            if tag in self._tags_count:
+                                self._tags_count[tag] = max(0, self._tags_count[tag] - 1)
+                                if self._tags_count[tag] == 0:
+                                    del self._tags_count[tag]
+                        
+                        # Remove from hash index
+                        self._hash_index.remove_by_path(path)
+                        total_removed += 1
+                    except Exception as e:
+                        logger.error(f"Error removing {path} from cache: {e}")
+                
+                # Update cache data
+                self._cache.raw_data = [item for item in self._cache.raw_data if item['file_path'] not in missing_files]
+            
+            # Resort cache if changes were made
+            if total_added > 0 or total_removed > 0:
+                # Update folders list
+                all_folders = set(item.get('folder', '') for item in self._cache.raw_data)
+                self._cache.folders = sorted(list(all_folders), key=lambda x: x.lower())
+                
+                # Resort cache
+                await self._cache.resort()
+                
+            logger.info(f"{self.model_type.capitalize()} Scanner: Cache reconciliation completed in {time.time() - start_time:.2f} seconds. Added {total_added}, removed {total_removed} models.")
+        except Exception as e:
+            logger.error(f"{self.model_type.capitalize()} Scanner: Error reconciling cache: {e}", exc_info=True)
+        finally:
+            self._is_initializing = False # Unset flag
+
+    # These methods should be implemented in child classes
+    async def scan_all_models(self) -> List[Dict]:
+        """Scan all model directories and return metadata"""
+        raise NotImplementedError("Subclasses must implement scan_all_models")
+    
+    def get_model_roots(self) -> List[str]:
+        """Get model root directories"""
+        raise NotImplementedError("Subclasses must implement get_model_roots")
+
+    async def scan_single_model(self, file_path: str) -> Optional[Dict]:
+        """Scan a single model file and return its metadata"""
+        try:
+            if not os.path.exists(os.path.realpath(file_path)):
+                return None
+                
+            # Get basic file info
+            metadata = await self._get_file_info(file_path)
+            if not metadata:
+                return None
+                
+            folder = self._calculate_folder(file_path)
+                    
+            # Ensure folder field exists
+            metadata_dict = metadata.to_dict()
+            metadata_dict['folder'] = folder or ''
+            
+            return metadata_dict
+            
+        except Exception as e:
+            logger.error(f"Error scanning {file_path}: {e}")
+            return None
+    
+    async def _get_file_info(self, file_path: str) -> Optional[BaseModelMetadata]:
+        """Get model file info and metadata (extensible for different model types)"""
+        return await get_file_info(file_path, self.model_class)
+    
+    def _calculate_folder(self, file_path: str) -> str:
+        """Calculate the folder path for a model file"""
+        for root in self.get_model_roots():
+            if file_path.startswith(root):
+                rel_path = os.path.relpath(file_path, root)
+                return os.path.dirname(rel_path).replace(os.path.sep, '/')
+        return ''
+
+    # Common methods shared between scanners
+    async def _process_model_file(self, file_path: str, root_path: str) -> Dict:
+        """Process a single model file and return its metadata"""
+        metadata = await load_metadata(file_path, self.model_class)
+        
+        if metadata is None:
+            civitai_info_path = f"{os.path.splitext(file_path)[0]}.civitai.info"
+            if os.path.exists(civitai_info_path):
+                try:
+                    with open(civitai_info_path, 'r', encoding='utf-8') as f:
+                        version_info = json.load(f)
+                    
+                    file_info = next((f for f in version_info.get('files', []) if f.get('primary')), None)
+                    if file_info:
+                        file_name = os.path.splitext(os.path.basename(file_path))[0]
+                        file_info['name'] = file_name
+                    
+                        metadata = self.model_class.from_civitai_info(version_info, file_info, file_path)
+                        metadata.preview_url = find_preview_file(file_name, os.path.dirname(file_path))
+                        await save_metadata(file_path, metadata)
+                        logger.debug(f"Created metadata from .civitai.info for {file_path}")
+                except Exception as e:
+                    logger.error(f"Error creating metadata from .civitai.info for {file_path}: {e}")
+            
+            if metadata is None:
+                metadata = await self._get_file_info(file_path)
+        
+        model_data = metadata.to_dict()
+        
+        await self._fetch_missing_metadata(file_path, model_data)
+        rel_path = os.path.relpath(file_path, root_path)
+        folder = os.path.dirname(rel_path)
+        model_data['folder'] = folder.replace(os.path.sep, '/')
+        
+        return model_data
+
+    async def _fetch_missing_metadata(self, file_path: str, model_data: Dict) -> None:
+        """Fetch missing description and tags from Civitai if needed"""
+        try:
+            if model_data.get('civitai_deleted', False):
+                logger.debug(f"Skipping metadata fetch for {file_path}: marked as deleted on Civitai")
+                return
+
+            needs_metadata_update = False
+            model_id = None
+            
+            if model_data.get('civitai'):
+                model_id = model_data['civitai'].get('modelId')
+                
+                if model_id:
+                    model_id = str(model_id)
+                    tags_missing = not model_data.get('tags') or len(model_data.get('tags', [])) == 0
+                    desc_missing = not model_data.get('modelDescription') or model_data.get('modelDescription') in (None, "")
+                    needs_metadata_update = tags_missing or desc_missing
+            
+            if needs_metadata_update and model_id:
+                logger.debug(f"Fetching missing metadata for {file_path} with model ID {model_id}")
+                from ..services.civitai_client import CivitaiClient
+                client = CivitaiClient()
+                
+                model_metadata, status_code = await client.get_model_metadata(model_id)
+                await client.close()
+                
+                if status_code == 404:
+                    logger.warning(f"Model {model_id} appears to be deleted from Civitai (404 response)")
+                    model_data['civitai_deleted'] = True
+                    
+                    metadata_path = os.path.splitext(file_path)[0] + '.metadata.json'
+                    with open(metadata_path, 'w', encoding='utf-8') as f:
+                        json.dump(model_data, f, indent=2, ensure_ascii=False)
+                
+                elif model_metadata:
+                    logger.debug(f"Updating metadata for {file_path} with model ID {model_id}")
+                    
+                    if model_metadata.get('tags') and (not model_data.get('tags') or len(model_data.get('tags', [])) == 0):
+                        model_data['tags'] = model_metadata['tags']
+                    
+                    if model_metadata.get('description') and (not model_data.get('modelDescription') or model_data.get('modelDescription') in (None, "")):
+                        model_data['modelDescription'] = model_metadata['description']
+                    
+                    metadata_path = os.path.splitext(file_path)[0] + '.metadata.json'
+                    with open(metadata_path, 'w', encoding='utf-8') as f:
+                        json.dump(model_data, f, indent=2, ensure_ascii=False)
+        except Exception as e:
+            logger.error(f"Failed to update metadata from Civitai for {file_path}: {e}")
+
+    async def _scan_directory(self, root_path: str) -> List[Dict]:
+        """Base implementation for directory scanning"""
+        models = []
+        original_root = root_path
+
+        async def scan_recursive(path: str, visited_paths: set):
+            try:
+                real_path = os.path.realpath(path)
+                if real_path in visited_paths:
+                    logger.debug(f"Skipping already visited path: {path}")
+                    return
+                visited_paths.add(real_path)
+                
+                with os.scandir(path) as it:
+                    entries = list(it)
+                    for entry in entries:
+                        try:
+                            if entry.is_file(follow_symlinks=True):
+                                ext = os.path.splitext(entry.name)[1].lower()
+                                if ext in self.file_extensions:
+                                    file_path = entry.path.replace(os.sep, "/")
+                                    await self._process_single_file(file_path, original_root, models)
+                                    await asyncio.sleep(0)
+                            elif entry.is_dir(follow_symlinks=True):
+                                await scan_recursive(entry.path, visited_paths)
+                        except Exception as e:
+                            logger.error(f"Error processing entry {entry.path}: {e}")
+            except Exception as e:
+                logger.error(f"Error scanning {path}: {e}")
+        
+        await scan_recursive(root_path, set())
+        return models
+
+    async def _process_single_file(self, file_path: str, root_path: str, models_list: list):
+        """Process a single file and add to results list"""
+        try:
+            result = await self._process_model_file(file_path, root_path)
+            if result:
+                models_list.append(result)
+        except Exception as e:
+            logger.error(f"Error processing {file_path}: {e}")
+    
+    async def move_model(self, source_path: str, target_path: str) -> bool:
+        """Move a model and its associated files to a new location"""
+        try:
+            source_path = source_path.replace(os.sep, '/')
+            target_path = target_path.replace(os.sep, '/')
+            
+            file_ext = os.path.splitext(source_path)[1]
+            
+            if not file_ext or file_ext.lower() not in self.file_extensions:
+                logger.error(f"Invalid file extension for model: {file_ext}")
+                return False
+                
+            base_name = os.path.splitext(os.path.basename(source_path))[0]
+            source_dir = os.path.dirname(source_path)
+            
+            os.makedirs(target_path, exist_ok=True)
+            
+            target_file = os.path.join(target_path, f"{base_name}{file_ext}").replace(os.sep, '/')
+
+            real_source = os.path.realpath(source_path)
+            real_target = os.path.realpath(target_file)
+            
+            file_size = os.path.getsize(real_source)
+            
+            # Get the appropriate file monitor through ServiceRegistry
+            if self.model_type == "lora":
+                monitor = await ServiceRegistry.get_lora_monitor()
+            elif self.model_type == "checkpoint":
+                monitor = await ServiceRegistry.get_checkpoint_monitor()
+            else:
+                monitor = None
+                
+            if monitor:
+                monitor.handler.add_ignore_path(
+                    real_source,
+                    file_size
+                )
+                monitor.handler.add_ignore_path(
+                    real_target,
+                    file_size
+                )
+            
+            shutil.move(real_source, real_target)
+            
+            source_metadata = os.path.join(source_dir, f"{base_name}.metadata.json")
+            metadata = None
+            if os.path.exists(source_metadata):
+                target_metadata = os.path.join(target_path, f"{base_name}.metadata.json")
+                shutil.move(source_metadata, target_metadata)
+                metadata = await self._update_metadata_paths(target_metadata, target_file)
+            
+            for ext in PREVIEW_EXTENSIONS:
+                source_preview = os.path.join(source_dir, f"{base_name}{ext}")
+                if os.path.exists(source_preview):
+                    target_preview = os.path.join(target_path, f"{base_name}{ext}")
+                    shutil.move(source_preview, target_preview)
+                    break
+            
+            await self.update_single_model_cache(source_path, target_file, metadata)
+            
+            return True
+            
+        except Exception as e:
+            logger.error(f"Error moving model: {e}", exc_info=True)
+            return False
+    
+    async def _update_metadata_paths(self, metadata_path: str, model_path: str) -> Dict:
+        """Update file paths in metadata file"""
+        try:
+            with open(metadata_path, 'r', encoding='utf-8') as f:
+                metadata = json.load(f)
+            
+            metadata['file_path'] = model_path.replace(os.sep, '/')
+            
+            if 'preview_url' in metadata:
+                preview_dir = os.path.dirname(model_path)
+                preview_name = os.path.splitext(os.path.basename(metadata['preview_url']))[0]
+                preview_ext = os.path.splitext(metadata['preview_url'])[1]
+                new_preview_path = os.path.join(preview_dir, f"{preview_name}{preview_ext}")
+                metadata['preview_url'] = new_preview_path.replace(os.sep, '/')
+            
+            with open(metadata_path, 'w', encoding='utf-8') as f:
+                json.dump(metadata, f, indent=2, ensure_ascii=False)
+
+            return metadata
+                
+        except Exception as e:
+            logger.error(f"Error updating metadata paths: {e}", exc_info=True)
+            return None
+
+    async def update_single_model_cache(self, original_path: str, new_path: str, metadata: Dict) -> bool:
+        """Update cache after a model has been moved or modified"""
+        cache = await self.get_cached_data()
+        
+        existing_item = next((item for item in cache.raw_data if item['file_path'] == original_path), None)
+        if existing_item and 'tags' in existing_item:
+            for tag in existing_item.get('tags', []):
+                if tag in self._tags_count:
+                    self._tags_count[tag] = max(0, self._tags_count[tag] - 1)
+                    if self._tags_count[tag] == 0:
+                        del self._tags_count[tag]
+        
+        self._hash_index.remove_by_path(original_path)
+        
+        cache.raw_data = [
+            item for item in cache.raw_data 
+            if item['file_path'] != original_path
+        ]
+        
+        if metadata:
+            if original_path == new_path:
+                existing_folder = next((item['folder'] for item in cache.raw_data 
+                                      if item['file_path'] == original_path), None)
+                if existing_folder:
+                    metadata['folder'] = existing_folder
+                else:
+                    metadata['folder'] = self._calculate_folder(new_path)
+            else:
+                metadata['folder'] = self._calculate_folder(new_path)
+            
+            cache.raw_data.append(metadata)
+            
+            if 'sha256' in metadata:
+                self._hash_index.add_entry(metadata['sha256'].lower(), new_path)
+            
+            all_folders = set(item['folder'] for item in cache.raw_data)
+            cache.folders = sorted(list(all_folders), key=lambda x: x.lower())
+            
+            if 'tags' in metadata:
+                for tag in metadata.get('tags', []):
+                    self._tags_count[tag] = self._tags_count.get(tag, 0) + 1
+        
+        await cache.resort()
+        
+        return True
+        
+    def has_hash(self, sha256: str) -> bool:
+        """Check if a model with given hash exists"""
+        return self._hash_index.has_hash(sha256.lower())
+        
+    def get_path_by_hash(self, sha256: str) -> Optional[str]:
+        """Get file path for a model by its hash"""
+        return self._hash_index.get_path(sha256.lower())
+        
+    def get_hash_by_path(self, file_path: str) -> Optional[str]:
+        """Get hash for a model by its file path"""
+        return self._hash_index.get_hash(file_path)
+
+    # TODO: Adjust this method to use metadata instead of finding the file    
+    def get_preview_url_by_hash(self, sha256: str) -> Optional[str]:
+        """Get preview static URL for a model by its hash"""
+        file_path = self._hash_index.get_path(sha256.lower())
+        if not file_path:
+            return None
+            
+        base_name = os.path.splitext(file_path)[0]
+        
+        for ext in PREVIEW_EXTENSIONS:
+            preview_path = f"{base_name}{ext}"
+            if os.path.exists(preview_path):
+                return config.get_preview_static_url(preview_path)
+        
+        return None
+        
+    async def get_top_tags(self, limit: int = 20) -> List[Dict[str, any]]:
+        """Get top tags sorted by count"""
+        await self.get_cached_data()
+        
+        sorted_tags = sorted(
+            [{"tag": tag, "count": count} for tag, count in self._tags_count.items()],
+            key=lambda x: x['count'],
+            reverse=True
+        )
+        
+        return sorted_tags[:limit]
+        
+    async def get_base_models(self, limit: int = 20) -> List[Dict[str, any]]:
+        """Get base models sorted by frequency"""
+        cache = await self.get_cached_data()
+        
+        base_model_counts = {}
+        for model in cache.raw_data:
+            if 'base_model' in model and model['base_model']:
+                base_model = model['base_model']
+                base_model_counts[base_model] = base_model_counts.get(base_model, 0) + 1
+        
+        sorted_models = [{'name': model, 'count': count} for model, count in base_model_counts.items()]
+        sorted_models.sort(key=lambda x: x['count'], reverse=True)
+        
+        return sorted_models[:limit]
+        
+    async def get_model_info_by_name(self, name):
+        """Get model information by name"""
+        try:
+            cache = await self.get_cached_data()
+            
+            for model in cache.raw_data:
+                if model.get("file_name") == name:
+                    return model
+                    
+            return None
+        except Exception as e:
+            logger.error(f"Error getting model info by name: {e}", exc_info=True)
+            return None
+        
+    async def update_preview_in_cache(self, file_path: str, preview_url: str) -> bool:
+        """Update preview URL in cache for a specific lora
+        
+        Args:
+            file_path: The file path of the lora to update
+            preview_url: The new preview URL
+            
+        Returns:
+            bool: True if the update was successful, False if cache doesn't exist or lora wasn't found
+        """
+        if self._cache is None:
+            return False
+
+        return await self._cache.update_preview_url(file_path, preview_url)

py/services/recipe_cache.py

@@ -0,0 +1,85 @@
+import asyncio
+from typing import List, Dict
+from dataclasses import dataclass
+from operator import itemgetter
+
+@dataclass
+class RecipeCache:
+    """Cache structure for Recipe data"""
+    raw_data: List[Dict]
+    sorted_by_name: List[Dict]
+    sorted_by_date: List[Dict]
+    
+    def __post_init__(self):
+        self._lock = asyncio.Lock()
+
+    async def resort(self, name_only: bool = False):
+        """Resort all cached data views"""
+        async with self._lock:
+            self.sorted_by_name = sorted(
+                self.raw_data, 
+                key=lambda x: x.get('title', '').lower()  # Case-insensitive sort
+            )
+            if not name_only:
+                self.sorted_by_date = sorted(
+                    self.raw_data, 
+                    key=itemgetter('created_date', 'file_path'), 
+                    reverse=True
+                )
+    
+    async def update_recipe_metadata(self, recipe_id: str, metadata: Dict) -> bool:
+        """Update metadata for a specific recipe in all cached data
+        
+        Args:
+            recipe_id: The ID of the recipe to update
+            metadata: The new metadata
+            
+        Returns:
+            bool: True if the update was successful, False if the recipe wasn't found
+        """
+
+        # Update in raw_data
+        for item in self.raw_data:
+            if item.get('id') == recipe_id:
+                item.update(metadata)
+                break
+        else:
+            return False  # Recipe not found
+            
+        # Resort to reflect changes
+        await self.resort()
+        return True
+            
+    async def add_recipe(self, recipe_data: Dict) -> None:
+        """Add a new recipe to the cache
+        
+        Args:
+            recipe_data: The recipe data to add
+        """
+        async with self._lock:
+            self.raw_data.append(recipe_data)
+            await self.resort() 
+
+    async def remove_recipe(self, recipe_id: str) -> bool:
+        """Remove a recipe from the cache by ID
+        
+        Args:
+            recipe_id: The ID of the recipe to remove
+            
+        Returns:
+            bool: True if the recipe was found and removed, False otherwise
+        """
+        # Find the recipe in raw_data
+        recipe_index = next((i for i, recipe in enumerate(self.raw_data) 
+                             if recipe.get('id') == recipe_id), None)
+        
+        if recipe_index is None:
+            return False
+        
+        # Remove from raw_data
+        self.raw_data.pop(recipe_index)
+        
+        # Resort to update sorted lists
+        await self.resort()
+        
+        return True 

py/services/recipe_scanner.py

@@ -0,0 +1,807 @@
+import os
+import logging
+import asyncio
+import json
+import time
+from typing import List, Dict, Optional, Any, Tuple
+from ..config import config
+from .recipe_cache import RecipeCache
+from .service_registry import ServiceRegistry
+from .lora_scanner import LoraScanner
+from ..utils.utils import fuzzy_match
+import sys
+
+logger = logging.getLogger(__name__)
+
+class RecipeScanner:
+    """Service for scanning and managing recipe images"""
+    
+    _instance = None
+    _lock = asyncio.Lock()
+    
+    @classmethod
+    async def get_instance(cls, lora_scanner: Optional[LoraScanner] = None):
+        """Get singleton instance of RecipeScanner"""
+        async with cls._lock:
+            if cls._instance is None:
+                if not lora_scanner:
+                    # Get lora scanner from service registry if not provided
+                    lora_scanner = await ServiceRegistry.get_lora_scanner()
+                cls._instance = cls(lora_scanner)
+            return cls._instance
+    
+    def __new__(cls, lora_scanner: Optional[LoraScanner] = None):
+        if cls._instance is None:
+            cls._instance = super().__new__(cls)
+            cls._instance._lora_scanner = lora_scanner
+            cls._instance._civitai_client = None  # Will be lazily initialized
+        return cls._instance
+    
+    def __init__(self, lora_scanner: Optional[LoraScanner] = None):
+        # Ensure initialization only happens once
+        if not hasattr(self, '_initialized'):
+            self._cache: Optional[RecipeCache] = None
+            self._initialization_lock = asyncio.Lock()
+            self._initialization_task: Optional[asyncio.Task] = None
+            self._is_initializing = False
+            if lora_scanner:
+                self._lora_scanner = lora_scanner
+            self._initialized = True
+    
+    async def _get_civitai_client(self):
+        """Lazily initialize CivitaiClient from registry"""
+        if self._civitai_client is None:
+            self._civitai_client = await ServiceRegistry.get_civitai_client()
+        return self._civitai_client
+    
+    async def initialize_in_background(self) -> None:
+        """Initialize cache in background using thread pool"""
+        try:
+            # Set initial empty cache to avoid None reference errors
+            if self._cache is None:
+                self._cache = RecipeCache(
+                    raw_data=[],
+                    sorted_by_name=[],
+                    sorted_by_date=[]
+                )
+            
+            # Mark as initializing to prevent concurrent initializations
+            self._is_initializing = True
+            
+            try:
+                # Start timer
+                start_time = time.time()
+                
+                # Use thread pool to execute CPU-intensive operations
+                loop = asyncio.get_event_loop()
+                cache = await loop.run_in_executor(
+                    None,  # Use default thread pool
+                    self._initialize_recipe_cache_sync  # Run synchronous version in thread
+                )
+                
+                # Calculate elapsed time and log it
+                elapsed_time = time.time() - start_time
+                recipe_count = len(cache.raw_data) if cache and hasattr(cache, 'raw_data') else 0
+                logger.info(f"Recipe cache initialized in {elapsed_time:.2f} seconds. Found {recipe_count} recipes")
+            finally:
+                # Mark initialization as complete regardless of outcome
+                self._is_initializing = False
+        except Exception as e:
+            logger.error(f"Recipe Scanner: Error initializing cache in background: {e}")
+    
+    def _initialize_recipe_cache_sync(self):
+        """Synchronous version of recipe cache initialization for thread pool execution"""
+        try:
+            # Create a new event loop for this thread
+            loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(loop)
+            
+            # Create a synchronous method to bypass the async lock
+            def sync_initialize_cache():
+                # We need to implement scan_all_recipes logic synchronously here
+                # instead of calling the async method to avoid event loop issues
+                recipes = []
+                recipes_dir = self.recipes_dir
+                
+                if not recipes_dir or not os.path.exists(recipes_dir):
+                    logger.warning(f"Recipes directory not found: {recipes_dir}")
+                    return recipes
+                
+                # Get all recipe JSON files in the recipes directory
+                recipe_files = []
+                for root, _, files in os.walk(recipes_dir):
+                    recipe_count = sum(1 for f in files if f.lower().endswith('.recipe.json'))
+                    if recipe_count > 0:
+                        for file in files:
+                            if file.lower().endswith('.recipe.json'):
+                                recipe_files.append(os.path.join(root, file))
+                
+                # Process each recipe file
+                for recipe_path in recipe_files:
+                    try:
+                        with open(recipe_path, 'r', encoding='utf-8') as f:
+                            recipe_data = json.load(f)
+                        
+                        # Validate recipe data
+                        if not recipe_data or not isinstance(recipe_data, dict):
+                            logger.warning(f"Invalid recipe data in {recipe_path}")
+                            continue
+                        
+                        # Ensure required fields exist
+                        required_fields = ['id', 'file_path', 'title']
+                        if not all(field in recipe_data for field in required_fields):
+                            logger.warning(f"Missing required fields in {recipe_path}")
+                            continue
+                        
+                        # Ensure the image file exists
+                        image_path = recipe_data.get('file_path')
+                        if not os.path.exists(image_path):
+                            recipe_dir = os.path.dirname(recipe_path)
+                            image_filename = os.path.basename(image_path)
+                            alternative_path = os.path.join(recipe_dir, image_filename)
+                            if os.path.exists(alternative_path):
+                                recipe_data['file_path'] = alternative_path
+                        
+                        # Ensure loras array exists
+                        if 'loras' not in recipe_data:
+                            recipe_data['loras'] = []
+                        
+                        # Ensure gen_params exists
+                        if 'gen_params' not in recipe_data:
+                            recipe_data['gen_params'] = {}
+                        
+                        # Add to list without async operations
+                        recipes.append(recipe_data)
+                    except Exception as e:
+                        logger.error(f"Error loading recipe file {recipe_path}: {e}")
+                        import traceback
+                        traceback.print_exc(file=sys.stderr)
+                
+                # Update cache with the collected data
+                self._cache.raw_data = recipes
+                
+                # Create a simplified resort function that doesn't use await
+                if hasattr(self._cache, "resort"):
+                    try:
+                        # Sort by name
+                        self._cache.sorted_by_name = sorted(
+                            self._cache.raw_data,
+                            key=lambda x: x.get('title', '').lower()
+                        )
+                        
+                        # Sort by date (modified or created)
+                        self._cache.sorted_by_date = sorted(
+                            self._cache.raw_data,
+                            key=lambda x: x.get('modified', x.get('created_date', 0)),
+                            reverse=True
+                        )
+                    except Exception as e:
+                        logger.error(f"Error sorting recipe cache: {e}")
+                
+                return self._cache
+            
+            # Run our sync initialization that avoids lock conflicts
+            return sync_initialize_cache()
+        except Exception as e:
+            logger.error(f"Error in thread-based recipe cache initialization: {e}")
+            return self._cache if hasattr(self, '_cache') else None
+        finally:
+            # Clean up the event loop
+            loop.close()
+
+    @property
+    def recipes_dir(self) -> str:
+        """Get path to recipes directory"""
+        if not config.loras_roots:
+            return ""
+        
+        # config.loras_roots already sorted case-insensitively, use the first one
+        recipes_dir = os.path.join(config.loras_roots[0], "recipes")
+        os.makedirs(recipes_dir, exist_ok=True)
+        
+        return recipes_dir
+    
+    async def get_cached_data(self, force_refresh: bool = False) -> RecipeCache:
+        """Get cached recipe data, refresh if needed"""
+        # If cache is already initialized and no refresh is needed, return it immediately
+        if self._cache is not None and not force_refresh:
+            return self._cache
+
+        # If another initialization is already in progress, wait for it to complete
+        if self._is_initializing and not force_refresh:
+            return self._cache or RecipeCache(raw_data=[], sorted_by_name=[], sorted_by_date=[])
+
+        # If force refresh is requested, initialize the cache directly
+        if force_refresh:
+            # Try to acquire the lock with a timeout to prevent deadlocks
+            try:
+                async with self._initialization_lock:
+                    # Mark as initializing to prevent concurrent initializations
+                    self._is_initializing = True
+                    
+                    try:
+                        # Scan for recipe data directly
+                        raw_data = await self.scan_all_recipes()
+                        
+                        # Update cache
+                        self._cache = RecipeCache(
+                            raw_data=raw_data,
+                            sorted_by_name=[],
+                            sorted_by_date=[]
+                        )
+                        
+                        # Resort cache
+                        await self._cache.resort()
+                        
+                        return self._cache
+                    
+                    except Exception as e:
+                        logger.error(f"Recipe Manager: Error initializing cache: {e}", exc_info=True)
+                        # Create empty cache on error
+                        self._cache = RecipeCache(
+                            raw_data=[],
+                            sorted_by_name=[],
+                            sorted_by_date=[]
+                        )
+                        return self._cache
+                    finally:
+                        # Mark initialization as complete
+                        self._is_initializing = False
+            
+            except Exception as e:
+                logger.error(f"Unexpected error in get_cached_data: {e}")
+        
+        # Return the cache (may be empty or partially initialized)
+        return self._cache or RecipeCache(raw_data=[], sorted_by_name=[], sorted_by_date=[])
+    
+    async def scan_all_recipes(self) -> List[Dict]:
+        """Scan all recipe JSON files and return metadata"""
+        recipes = []
+        recipes_dir = self.recipes_dir
+        
+        if not recipes_dir or not os.path.exists(recipes_dir):
+            logger.warning(f"Recipes directory not found: {recipes_dir}")
+            return recipes
+        
+        # Get all recipe JSON files in the recipes directory
+        recipe_files = []
+        for root, _, files in os.walk(recipes_dir):
+            recipe_count = sum(1 for f in files if f.lower().endswith('.recipe.json'))
+            if recipe_count > 0:
+                for file in files:
+                    if file.lower().endswith('.recipe.json'):
+                        recipe_files.append(os.path.join(root, file))
+        
+        # Process each recipe file
+        for recipe_path in recipe_files:
+            recipe_data = await self._load_recipe_file(recipe_path)
+            if recipe_data:
+                recipes.append(recipe_data)
+        
+        return recipes
+    
+    async def _load_recipe_file(self, recipe_path: str) -> Optional[Dict]:
+        """Load recipe data from a JSON file"""
+        try:
+            with open(recipe_path, 'r', encoding='utf-8') as f:
+                recipe_data = json.load(f)
+            
+            # Validate recipe data
+            if not recipe_data or not isinstance(recipe_data, dict):
+                logger.warning(f"Invalid recipe data in {recipe_path}")
+                return None
+            
+            # Ensure required fields exist
+            required_fields = ['id', 'file_path', 'title']
+            for field in required_fields:
+                if field not in recipe_data:
+                    logger.warning(f"Missing required field '{field}' in {recipe_path}")
+                    return None
+            
+            # Ensure the image file exists
+            image_path = recipe_data.get('file_path')
+            if not os.path.exists(image_path):
+                logger.warning(f"Recipe image not found: {image_path}")
+                # Try to find the image in the same directory as the recipe
+                recipe_dir = os.path.dirname(recipe_path)
+                image_filename = os.path.basename(image_path)
+                alternative_path = os.path.join(recipe_dir, image_filename)
+                if os.path.exists(alternative_path):
+                    recipe_data['file_path'] = alternative_path
+                else:
+                    logger.warning(f"Could not find alternative image path for {image_path}")
+            
+            # Ensure loras array exists
+            if 'loras' not in recipe_data:
+                recipe_data['loras'] = []
+            
+            # Ensure gen_params exists
+            if 'gen_params' not in recipe_data:
+                recipe_data['gen_params'] = {}
+            
+            # Update lora information with local paths and availability
+            await self._update_lora_information(recipe_data)
+            
+            return recipe_data
+        except Exception as e:
+            logger.error(f"Error loading recipe file {recipe_path}: {e}")
+            import traceback
+            traceback.print_exc(file=sys.stderr)
+            return None
+    
+    async def _update_lora_information(self, recipe_data: Dict) -> bool:
+        """Update LoRA information with hash and file_name
+        
+        Returns:
+            bool: True if metadata was updated
+        """
+        if not recipe_data.get('loras'):
+            return False
+        
+        metadata_updated = False
+        
+        for lora in recipe_data['loras']:
+            # Skip if already has complete information
+            if 'hash' in lora and 'file_name' in lora and lora['file_name']:
+                continue
+                
+            # If has modelVersionId but no hash, look in lora cache first, then fetch from Civitai
+            if 'modelVersionId' in lora and not lora.get('hash'):
+                model_version_id = lora['modelVersionId']
+
+                # Try to find in lora cache first
+                hash_from_cache = await self._find_hash_in_lora_cache(model_version_id)
+                if hash_from_cache:
+                    lora['hash'] = hash_from_cache
+                    metadata_updated = True
+                else:
+                    # If not in cache, fetch from Civitai
+                    hash_from_civitai = await self._get_hash_from_civitai(model_version_id)
+                    if hash_from_civitai:
+                        lora['hash'] = hash_from_civitai
+                        metadata_updated = True
+                    else:
+                        logger.debug(f"Could not get hash for modelVersionId {model_version_id}")
+            
+            # If has hash but no file_name, look up in lora library
+            if 'hash' in lora and (not lora.get('file_name') or not lora['file_name']):
+                hash_value = lora['hash']
+                
+                if self._lora_scanner.has_lora_hash(hash_value):
+                    lora_path = self._lora_scanner.get_lora_path_by_hash(hash_value)
+                    if lora_path:
+                        file_name = os.path.splitext(os.path.basename(lora_path))[0]
+                        lora['file_name'] = file_name
+                        metadata_updated = True
+                else:
+                    # Lora not in library
+                    lora['file_name'] = ''
+                    metadata_updated = True
+        
+        return metadata_updated
+    
+    async def _find_hash_in_lora_cache(self, model_version_id: str) -> Optional[str]:
+        """Find hash in lora cache based on modelVersionId"""
+        try:
+            # Get all loras from cache
+            if not self._lora_scanner:
+                return None
+                
+            cache = await self._lora_scanner.get_cached_data()
+            if not cache or not cache.raw_data:
+                return None
+                
+            # Find lora with matching civitai.id
+            for lora in cache.raw_data:
+                civitai_data = lora.get('civitai', {})
+                if civitai_data and str(civitai_data.get('id', '')) == str(model_version_id):
+                    return lora.get('sha256')
+                    
+            return None
+        except Exception as e:
+            logger.error(f"Error finding hash in lora cache: {e}")
+            return None
+    
+    async def _get_hash_from_civitai(self, model_version_id: str) -> Optional[str]:
+        """Get hash from Civitai API"""
+        try:
+            # Get CivitaiClient from ServiceRegistry
+            civitai_client = await self._get_civitai_client()
+            if not civitai_client:
+                logger.error("Failed to get CivitaiClient from ServiceRegistry")
+                return None
+                
+            version_info = await civitai_client.get_model_version_info(model_version_id)
+            
+            if not version_info or not version_info.get('files'):
+                logger.debug(f"No files found in version info for ID: {model_version_id}")
+                return None
+                
+            # Get hash from the first file
+            for file_info in version_info.get('files', []):
+                if file_info.get('hashes', {}).get('SHA256'):
+                    return file_info['hashes']['SHA256']
+                    
+            logger.debug(f"No SHA256 hash found in version info for ID: {model_version_id}")
+            return None
+        except Exception as e:
+            logger.error(f"Error getting hash from Civitai: {e}")
+            return None
+
+    async def _get_model_version_name(self, model_version_id: str) -> Optional[str]:
+        """Get model version name from Civitai API"""
+        try:
+            # Get CivitaiClient from ServiceRegistry
+            civitai_client = await self._get_civitai_client()
+            if not civitai_client:
+                return None
+                
+            version_info = await civitai_client.get_model_version_info(model_version_id)
+            
+            if version_info and 'name' in version_info:
+                return version_info['name']
+                    
+            logger.debug(f"No version name found for modelVersionId {model_version_id}")
+            return None
+        except Exception as e:
+            logger.error(f"Error getting model version name from Civitai: {e}")
+            return None
+
+    async def _determine_base_model(self, loras: List[Dict]) -> Optional[str]:
+        """Determine the most common base model among LoRAs"""
+        base_models = {}
+        
+        # Count occurrences of each base model
+        for lora in loras:
+            if 'hash' in lora:
+                lora_path = self._lora_scanner.get_lora_path_by_hash(lora['hash'])
+                if lora_path:
+                    base_model = await self._get_base_model_for_lora(lora_path)
+                    if base_model:
+                        base_models[base_model] = base_models.get(base_model, 0) + 1
+        
+        # Return the most common base model
+        if base_models:
+            return max(base_models.items(), key=lambda x: x[1])[0]
+        return None
+
+    async def _get_base_model_for_lora(self, lora_path: str) -> Optional[str]:
+        """Get base model for a LoRA from cache"""
+        try:
+            if not self._lora_scanner:
+                return None
+            
+            cache = await self._lora_scanner.get_cached_data()
+            if not cache or not cache.raw_data:
+                return None
+            
+            # Find matching lora in cache
+            for lora in cache.raw_data:
+                if lora.get('file_path') == lora_path:
+                    return lora.get('base_model')
+                
+            return None
+        except Exception as e:
+            logger.error(f"Error getting base model for lora: {e}")
+            return None
+
+    async def get_paginated_data(self, page: int, page_size: int, sort_by: str = 'date', search: str = None, filters: dict = None, search_options: dict = None, lora_hash: str = None, bypass_filters: bool = True):
+        """Get paginated and filtered recipe data
+        
+        Args:
+            page: Current page number (1-based)
+            page_size: Number of items per page
+            sort_by: Sort method ('name' or 'date')
+            search: Search term
+            filters: Dictionary of filters to apply
+            search_options: Dictionary of search options to apply
+            lora_hash: Optional SHA256 hash of a LoRA to filter recipes by
+            bypass_filters: If True, ignore other filters when a lora_hash is provided
+        """
+        cache = await self.get_cached_data()
+
+        # Get base dataset
+        filtered_data = cache.sorted_by_date if sort_by == 'date' else cache.sorted_by_name
+        
+        # Special case: Filter by LoRA hash (takes precedence if bypass_filters is True)
+        if lora_hash:
+            # Filter recipes that contain this LoRA hash
+            filtered_data = [
+                item for item in filtered_data
+                if 'loras' in item and any(
+                    lora.get('hash', '').lower() == lora_hash.lower() 
+                    for lora in item['loras']
+                )
+            ]
+            
+            if bypass_filters:
+                # Skip other filters if bypass_filters is True
+                pass
+            # Otherwise continue with normal filtering after applying LoRA hash filter
+        
+        # Skip further filtering if we're only filtering by LoRA hash with bypass enabled
+        if not (lora_hash and bypass_filters):
+            # Apply search filter
+            if search:
+                # Default search options if none provided
+                if not search_options:
+                    search_options = {
+                        'title': True,
+                        'tags': True,
+                        'lora_name': True,
+                        'lora_model': True
+                    }
+                
+                # Build the search predicate based on search options
+                def matches_search(item):
+                    # Search in title if enabled
+                    if search_options.get('title', True):
+                        if fuzzy_match(str(item.get('title', '')), search):
+                            return True
+                    
+                    # Search in tags if enabled
+                    if search_options.get('tags', True) and 'tags' in item:
+                        for tag in item['tags']:
+                            if fuzzy_match(tag, search):
+                                return True
+                    
+                    # Search in lora file names if enabled
+                    if search_options.get('lora_name', True) and 'loras' in item:
+                        for lora in item['loras']:
+                            if fuzzy_match(str(lora.get('file_name', '')), search):
+                                return True
+                    
+                    # Search in lora model names if enabled
+                    if search_options.get('lora_model', True) and 'loras' in item:
+                        for lora in item['loras']:
+                            if fuzzy_match(str(lora.get('modelName', '')), search):
+                                return True
+                    
+                    # No match found
+                    return False
+                
+                # Filter the data using the search predicate
+                filtered_data = [item for item in filtered_data if matches_search(item)]
+            
+            # Apply additional filters
+            if filters:
+                # Filter by base model
+                if 'base_model' in filters and filters['base_model']:
+                    filtered_data = [
+                        item for item in filtered_data
+                        if item.get('base_model', '') in filters['base_model']
+                    ]
+                
+                # Filter by tags
+                if 'tags' in filters and filters['tags']:
+                    filtered_data = [
+                        item for item in filtered_data
+                        if any(tag in item.get('tags', []) for tag in filters['tags'])
+                    ]
+
+        # Calculate pagination
+        total_items = len(filtered_data)
+        start_idx = (page - 1) * page_size
+        end_idx = min(start_idx + page_size, total_items)
+        
+        # Get paginated items
+        paginated_items = filtered_data[start_idx:end_idx]
+        
+        # Add inLibrary information for each lora
+        for item in paginated_items:
+            if 'loras' in item:
+                for lora in item['loras']:
+                    if 'hash' in lora and lora['hash']:
+                        lora['inLibrary'] = self._lora_scanner.has_lora_hash(lora['hash'].lower())
+                        lora['preview_url'] = self._lora_scanner.get_preview_url_by_hash(lora['hash'].lower())
+                        lora['localPath'] = self._lora_scanner.get_lora_path_by_hash(lora['hash'].lower())
+        
+        result = {
+            'items': paginated_items,
+            'total': total_items,
+            'page': page,
+            'page_size': page_size,
+            'total_pages': (total_items + page_size - 1) // page_size
+        }
+        
+        return result
+    
+    async def get_recipe_by_id(self, recipe_id: str) -> dict:
+        """Get a single recipe by ID with all metadata and formatted URLs
+        
+        Args:
+            recipe_id: The ID of the recipe to retrieve
+            
+        Returns:
+            Dict containing the recipe data or None if not found
+        """
+        if not recipe_id:
+            return None
+            
+        # Get all recipes from cache
+        cache = await self.get_cached_data()
+        
+        # Find the recipe with the specified ID
+        recipe = next((r for r in cache.raw_data if str(r.get('id', '')) == recipe_id), None)
+        
+        if not recipe:
+            return None
+            
+        # Format the recipe with all needed information
+        formatted_recipe = {**recipe}  # Copy all fields
+        
+        # Format file path to URL
+        if 'file_path' in formatted_recipe:
+            formatted_recipe['file_url'] = self._format_file_url(formatted_recipe['file_path'])
+            
+        # Format dates for display
+        for date_field in ['created_date', 'modified']:
+            if date_field in formatted_recipe:
+                formatted_recipe[f"{date_field}_formatted"] = self._format_timestamp(formatted_recipe[date_field])
+                
+        # Add lora metadata
+        if 'loras' in formatted_recipe:
+            for lora in formatted_recipe['loras']:
+                if 'hash' in lora and lora['hash']:
+                    lora_hash = lora['hash'].lower()
+                    lora['inLibrary'] = self._lora_scanner.has_lora_hash(lora_hash)
+                    lora['preview_url'] = self._lora_scanner.get_preview_url_by_hash(lora_hash)
+                    lora['localPath'] = self._lora_scanner.get_lora_path_by_hash(lora_hash)
+                    
+        return formatted_recipe
+        
+    def _format_file_url(self, file_path: str) -> str:
+        """Format file path as URL for serving in web UI"""
+        if not file_path:
+            return '/loras_static/images/no-preview.png'
+            
+        try:
+            # Format file path as a URL that will work with static file serving
+            recipes_dir = os.path.join(config.loras_roots[0], "recipes").replace(os.sep, '/')
+            if file_path.replace(os.sep, '/').startswith(recipes_dir):
+                relative_path = os.path.relpath(file_path, config.loras_roots[0]).replace(os.sep, '/')
+                return f"/loras_static/root1/preview/{relative_path}"
+                
+            # If not in recipes dir, try to create a valid URL from the file name
+            file_name = os.path.basename(file_path)
+            return f"/loras_static/root1/preview/recipes/{file_name}"
+        except Exception as e:
+            logger.error(f"Error formatting file URL: {e}")
+            return '/loras_static/images/no-preview.png'
+    
+    def _format_timestamp(self, timestamp: float) -> str:
+        """Format timestamp for display"""
+        from datetime import datetime
+        return datetime.fromtimestamp(timestamp).strftime('%Y-%m-%d %H:%M:%S')
+
+    async def update_recipe_metadata(self, recipe_id: str, metadata: dict) -> bool:
+        """Update recipe metadata (like title and tags) in both file system and cache
+        
+        Args:
+            recipe_id: The ID of the recipe to update
+            metadata: Dictionary containing metadata fields to update (title, tags, etc.)
+            
+        Returns:
+            bool: True if successful, False otherwise
+        """
+        import os
+        import json
+        
+        # First, find the recipe JSON file path
+        recipe_json_path = os.path.join(self.recipes_dir, f"{recipe_id}.recipe.json")
+        
+        if not os.path.exists(recipe_json_path):
+            return False
+            
+        try:
+            # Load existing recipe data
+            with open(recipe_json_path, 'r', encoding='utf-8') as f:
+                recipe_data = json.load(f)
+                
+            # Update fields
+            for key, value in metadata.items():
+                recipe_data[key] = value
+                
+            # Save updated recipe
+            with open(recipe_json_path, 'w', encoding='utf-8') as f:
+                json.dump(recipe_data, f, indent=4, ensure_ascii=False)
+                
+            # Update the cache if it exists
+            if self._cache is not None:
+                await self._cache.update_recipe_metadata(recipe_id, metadata)
+                
+            # If the recipe has an image, update its EXIF metadata
+            from ..utils.exif_utils import ExifUtils
+            image_path = recipe_data.get('file_path')
+            if image_path and os.path.exists(image_path):
+                ExifUtils.append_recipe_metadata(image_path, recipe_data)
+                
+            return True
+        except Exception as e:
+            import logging
+            logging.getLogger(__name__).error(f"Error updating recipe metadata: {e}", exc_info=True)
+            return False
+
+    async def update_lora_filename_by_hash(self, hash_value: str, new_file_name: str) -> Tuple[int, int]:
+        """Update file_name in all recipes that contain a LoRA with the specified hash.
+        
+        Args:
+            hash_value: The SHA256 hash value of the LoRA
+            new_file_name: The new file_name to set
+            
+        Returns:
+            Tuple[int, int]: (number of recipes updated in files, number of recipes updated in cache)
+        """
+        if not hash_value or not new_file_name:
+            return 0, 0
+            
+        # Always use lowercase hash for consistency
+        hash_value = hash_value.lower()
+        
+        # Get recipes directory
+        recipes_dir = self.recipes_dir
+        if not recipes_dir or not os.path.exists(recipes_dir):
+            logger.warning(f"Recipes directory not found: {recipes_dir}")
+            return 0, 0
+            
+        # Check if cache is initialized
+        cache_initialized = self._cache is not None
+        cache_updated_count = 0
+        file_updated_count = 0
+        
+        # Get all recipe JSON files in the recipes directory
+        recipe_files = []
+        for root, _, files in os.walk(recipes_dir):
+            for file in files:
+                if file.lower().endswith('.recipe.json'):
+                    recipe_files.append(os.path.join(root, file))
+        
+        # Process each recipe file
+        for recipe_path in recipe_files:
+            try:
+                # Load the recipe data
+                with open(recipe_path, 'r', encoding='utf-8') as f:
+                    recipe_data = json.load(f)
+                
+                # Skip if no loras or invalid structure
+                if not recipe_data or not isinstance(recipe_data, dict) or 'loras' not in recipe_data:
+                    continue
+                
+                # Check if any lora has matching hash
+                file_updated = False
+                for lora in recipe_data.get('loras', []):
+                    if 'hash' in lora and lora['hash'].lower() == hash_value:
+                        # Update file_name
+                        old_file_name = lora.get('file_name', '')
+                        lora['file_name'] = new_file_name
+                        file_updated = True
+                        logger.info(f"Updated file_name in recipe {recipe_path}: {old_file_name} -> {new_file_name}")
+                
+                # If updated, save the file
+                if file_updated:
+                    with open(recipe_path, 'w', encoding='utf-8') as f:
+                        json.dump(recipe_data, f, indent=4, ensure_ascii=False)
+                    file_updated_count += 1
+                    
+                    # Also update in cache if it exists
+                    if cache_initialized:
+                        recipe_id = recipe_data.get('id')
+                        if recipe_id:
+                            for cache_item in self._cache.raw_data:
+                                if cache_item.get('id') == recipe_id:
+                                    # Replace loras array with updated version
+                                    cache_item['loras'] = recipe_data['loras']
+                                    cache_updated_count += 1
+                                    break
+            
+            except Exception as e:
+                logger.error(f"Error updating recipe file {recipe_path}: {e}")
+                import traceback
+                traceback.print_exc(file=sys.stderr)
+        
+        # Resort cache if updates were made
+        if cache_initialized and cache_updated_count > 0:
+            await self._cache.resort()
+            logger.info(f"Resorted recipe cache after updating {cache_updated_count} items")
+            
+        return file_updated_count, cache_updated_count

py/services/service_registry.py

@@ -0,0 +1,124 @@
+import asyncio
+import logging
+from typing import Optional, Dict, Any, TypeVar, Type
+
+logger = logging.getLogger(__name__)
+
+T = TypeVar('T')  # Define a type variable for service types
+
+class ServiceRegistry:
+    """Centralized registry for service singletons"""
+    
+    _instance = None
+    _services: Dict[str, Any] = {}
+    _lock = asyncio.Lock()
+    
+    @classmethod
+    def get_instance(cls):
+        """Get singleton instance of the registry"""
+        if cls._instance is None:
+            cls._instance = cls()
+        return cls._instance
+    
+    @classmethod
+    async def register_service(cls, service_name: str, service_instance: Any) -> None:
+        """Register a service instance with the registry"""
+        registry = cls.get_instance()
+        async with cls._lock:
+            registry._services[service_name] = service_instance
+            logger.debug(f"Registered service: {service_name}")
+    
+    @classmethod
+    async def get_service(cls, service_name: str) -> Any:
+        """Get a service instance by name"""
+        registry = cls.get_instance()
+        async with cls._lock:
+            if service_name not in registry._services:
+                logger.debug(f"Service {service_name} not found in registry")
+                return None
+            return registry._services[service_name]
+    
+    # Convenience methods for common services
+    @classmethod
+    async def get_lora_scanner(cls):
+        """Get the LoraScanner instance"""
+        from .lora_scanner import LoraScanner
+        scanner = await cls.get_service("lora_scanner")
+        if scanner is None:
+            scanner = await LoraScanner.get_instance()
+            await cls.register_service("lora_scanner", scanner)
+        return scanner
+        
+    @classmethod
+    async def get_checkpoint_scanner(cls):
+        """Get the CheckpointScanner instance"""
+        from .checkpoint_scanner import CheckpointScanner
+        scanner = await cls.get_service("checkpoint_scanner")
+        if scanner is None:
+            scanner = await CheckpointScanner.get_instance()
+            await cls.register_service("checkpoint_scanner", scanner)
+        return scanner
+    
+    @classmethod
+    async def get_lora_monitor(cls):
+        """Get the LoraFileMonitor instance"""
+        from .file_monitor import LoraFileMonitor
+        monitor = await cls.get_service("lora_monitor")
+        if monitor is None:
+            monitor = await LoraFileMonitor.get_instance()
+            await cls.register_service("lora_monitor", monitor)
+        return monitor
+    
+    @classmethod
+    async def get_checkpoint_monitor(cls):
+        """Get the CheckpointFileMonitor instance"""
+        from .file_monitor import CheckpointFileMonitor
+        monitor = await cls.get_service("checkpoint_monitor")
+        if monitor is None:
+            monitor = await CheckpointFileMonitor.get_instance()
+            await cls.register_service("checkpoint_monitor", monitor)
+        return monitor
+
+    @classmethod
+    async def get_civitai_client(cls):
+        """Get the CivitaiClient instance"""
+        from .civitai_client import CivitaiClient
+        client = await cls.get_service("civitai_client")
+        if client is None:
+            client = await CivitaiClient.get_instance()
+            await cls.register_service("civitai_client", client)
+        return client
+
+    @classmethod
+    async def get_download_manager(cls):
+        """Get the DownloadManager instance"""
+        from .download_manager import DownloadManager
+        manager = await cls.get_service("download_manager")
+        if manager is None:
+            # We'll let DownloadManager.get_instance handle file_monitor parameter
+            manager = await DownloadManager.get_instance()
+            await cls.register_service("download_manager", manager)
+        return manager
+
+    @classmethod
+    async def get_recipe_scanner(cls):
+        """Get the RecipeScanner instance"""
+        from .recipe_scanner import RecipeScanner
+        scanner = await cls.get_service("recipe_scanner")
+        if scanner is None:
+            lora_scanner = await cls.get_lora_scanner()
+            scanner = RecipeScanner(lora_scanner)
+            await cls.register_service("recipe_scanner", scanner)
+        return scanner
+
+    @classmethod
+    async def get_websocket_manager(cls):
+        """Get the WebSocketManager instance"""
+        from .websocket_manager import ws_manager
+        manager = await cls.get_service("websocket_manager")
+        if manager is None:
+            # ws_manager is already a global instance in websocket_manager.py
+            from .websocket_manager import ws_manager
+            await cls.register_service("websocket_manager", ws_manager)
+            manager = ws_manager
+        return manager

py/services/settings_manager.py

@@ -37,7 +37,8 @@ class SettingsManager:
     def _get_default_settings(self) -> Dict[str, Any]:
         """Return default settings"""
         return {
-            "civitai_api_key": ""
+            "civitai_api_key": "",
+            "show_only_sfw": False
         }
 
     def get(self, key: str, default: Any = None) -> Any:

py/services/websocket_manager.py

@@ -9,6 +9,8 @@ class WebSocketManager:
     
     def __init__(self):
         self._websockets: Set[web.WebSocketResponse] = set()
+        self._init_websockets: Set[web.WebSocketResponse] = set()  # New set for initialization progress clients
+        self._checkpoint_websockets: Set[web.WebSocketResponse] = set()  # New set for checkpoint download progress
         
     async def handle_connection(self, request: web.Request) -> web.WebSocketResponse:
         """Handle new WebSocket connection"""
@@ -23,6 +25,34 @@ class WebSocketManager:
         finally:
             self._websockets.discard(ws)
         return ws
+    
+    async def handle_init_connection(self, request: web.Request) -> web.WebSocketResponse:
+        """Handle new WebSocket connection for initialization progress"""
+        ws = web.WebSocketResponse()
+        await ws.prepare(request)
+        self._init_websockets.add(ws)
+        
+        try:
+            async for msg in ws:
+                if msg.type == web.WSMsgType.ERROR:
+                    logger.error(f'Init WebSocket error: {ws.exception()}')
+        finally:
+            self._init_websockets.discard(ws)
+        return ws
+        
+    async def handle_checkpoint_connection(self, request: web.Request) -> web.WebSocketResponse:
+        """Handle new WebSocket connection for checkpoint download progress"""
+        ws = web.WebSocketResponse()
+        await ws.prepare(request)
+        self._checkpoint_websockets.add(ws)
+        
+        try:
+            async for msg in ws:
+                if msg.type == web.WSMsgType.ERROR:
+                    logger.error(f'Checkpoint WebSocket error: {ws.exception()}')
+        finally:
+            self._checkpoint_websockets.discard(ws)
+        return ws
         
     async def broadcast(self, data: Dict):
         """Broadcast message to all connected clients"""
@@ -34,10 +64,48 @@ class WebSocketManager:
                 await ws.send_json(data)
             except Exception as e:
                 logger.error(f"Error sending progress: {e}")
+    
+    async def broadcast_init_progress(self, data: Dict):
+        """Broadcast initialization progress to connected clients"""
+        if not self._init_websockets:
+            return
+            
+        # Ensure data has all required fields
+        if 'stage' not in data:
+            data['stage'] = 'processing'
+        if 'progress' not in data:
+            data['progress'] = 0
+        if 'details' not in data:
+            data['details'] = 'Processing...'
+            
+        for ws in self._init_websockets:
+            try:
+                await ws.send_json(data)
+            except Exception as e:
+                logger.error(f"Error sending initialization progress: {e}")
+    
+    async def broadcast_checkpoint_progress(self, data: Dict):
+        """Broadcast checkpoint download progress to connected clients"""
+        if not self._checkpoint_websockets:
+            return
+            
+        for ws in self._checkpoint_websockets:
+            try:
+                await ws.send_json(data)
+            except Exception as e:
+                logger.error(f"Error sending checkpoint progress: {e}")
                 
     def get_connected_clients_count(self) -> int:
         """Get number of connected clients"""
         return len(self._websockets)
 
+    def get_init_clients_count(self) -> int:
+        """Get number of initialization progress clients"""
+        return len(self._init_websockets)
+
+    def get_checkpoint_clients_count(self) -> int:
+        """Get number of checkpoint progress clients"""
+        return len(self._checkpoint_websockets)
+
 # Global instance
-ws_manager = WebSocketManager() 
+ws_manager = WebSocketManager()

py/utils/constants.py

@@ -0,0 +1,25 @@
+NSFW_LEVELS = {
+    "PG": 1,
+    "PG13": 2,
+    "R": 4,
+    "X": 8,
+    "XXX": 16,
+    "Blocked": 32, # Probably not actually visible through the API without being logged in on model owner account?
+}
+
+# preview extensions
+PREVIEW_EXTENSIONS = [
+    '.webp',
+    '.preview.webp',
+    '.preview.png', 
+    '.preview.jpeg', 
+    '.preview.jpg', 
+    '.preview.mp4',
+    '.png', 
+    '.jpeg', 
+    '.jpg', 
+    '.mp4'
+]
+
+# Card preview image width
+CARD_PREVIEW_WIDTH = 480

py/utils/exif_utils.py

@@ -0,0 +1,315 @@
+import piexif
+import json
+import logging
+from typing import Optional
+from io import BytesIO
+import os
+from PIL import Image
+
+logger = logging.getLogger(__name__)
+
+class ExifUtils:
+    """Utility functions for working with EXIF data in images"""
+    
+    @staticmethod
+    def extract_image_metadata(image_path: str) -> Optional[str]:
+        """Extract metadata from image including UserComment or parameters field
+        
+        Args:
+            image_path (str): Path to the image file
+            
+        Returns:
+            Optional[str]: Extracted metadata or None if not found
+        """
+        try:
+            # First try to open the image
+            with Image.open(image_path) as img:
+                # Method 1: Check for parameters in image info
+                if hasattr(img, 'info') and 'parameters' in img.info:
+                    return img.info['parameters']
+                
+                # Method 2: Check EXIF UserComment field
+                if img.format not in ['JPEG', 'TIFF', 'WEBP']:
+                    # For non-JPEG/TIFF/WEBP images, try to get EXIF through PIL
+                    exif = img._getexif()
+                    if exif and piexif.ExifIFD.UserComment in exif:
+                        user_comment = exif[piexif.ExifIFD.UserComment]
+                        if isinstance(user_comment, bytes):
+                            if user_comment.startswith(b'UNICODE\0'):
+                                return user_comment[8:].decode('utf-16be')
+                            return user_comment.decode('utf-8', errors='ignore')
+                        return user_comment
+                
+                # For JPEG/TIFF/WEBP, use piexif
+                try:
+                    exif_dict = piexif.load(image_path)
+                    
+                    if piexif.ExifIFD.UserComment in exif_dict.get('Exif', {}):
+                        user_comment = exif_dict['Exif'][piexif.ExifIFD.UserComment]
+                        if isinstance(user_comment, bytes):
+                            if user_comment.startswith(b'UNICODE\0'):
+                                user_comment = user_comment[8:].decode('utf-16be')
+                            else:
+                                user_comment = user_comment.decode('utf-8', errors='ignore')
+                        return user_comment
+                except Exception as e:
+                    logger.debug(f"Error loading EXIF data: {e}")
+                
+                # Method 3: Check PNG metadata for workflow info (for ComfyUI images)
+                if img.format == 'PNG':
+                    # Look for workflow or prompt metadata in PNG chunks
+                    for key in img.info:
+                        if key in ['workflow', 'prompt', 'parameters']:
+                            return img.info[key]
+                
+                return None
+                
+        except Exception as e:
+            logger.error(f"Error extracting image metadata: {e}", exc_info=True)
+            return None
+    
+    @staticmethod
+    def update_image_metadata(image_path: str, metadata: str) -> str:
+        """Update metadata in image's EXIF data or parameters fields
+        
+        Args:
+            image_path (str): Path to the image file
+            metadata (str): Metadata string to save
+            
+        Returns:
+            str: Path to the updated image
+        """
+        try:
+            # Load the image and check its format
+            with Image.open(image_path) as img:
+                img_format = img.format
+                
+                # For PNG, try to update parameters directly
+                if img_format == 'PNG':
+                    # We'll save with parameters in the PNG info
+                    info_dict = {'parameters': metadata}
+                    img.save(image_path, format='PNG', pnginfo=info_dict)
+                    return image_path
+                
+                # For WebP format, use PIL's exif parameter directly
+                elif img_format == 'WEBP':
+                    exif_dict = {'Exif': {piexif.ExifIFD.UserComment: b'UNICODE\0' + metadata.encode('utf-16be')}}
+                    exif_bytes = piexif.dump(exif_dict)
+                    
+                    # Save with the exif data
+                    img.save(image_path, format='WEBP', exif=exif_bytes, quality=85)
+                    return image_path
+                
+                # For other formats, use standard EXIF approach
+                else:
+                    try:
+                        exif_dict = piexif.load(img.info.get('exif', b''))
+                    except:
+                        exif_dict = {'0th':{}, 'Exif':{}, 'GPS':{}, 'Interop':{}, '1st':{}}
+                    
+                    # If no Exif dictionary exists, create one
+                    if 'Exif' not in exif_dict:
+                        exif_dict['Exif'] = {}
+                    
+                    # Update the UserComment field - use UNICODE format
+                    unicode_bytes = metadata.encode('utf-16be')
+                    metadata_bytes = b'UNICODE\0' + unicode_bytes
+                    
+                    exif_dict['Exif'][piexif.ExifIFD.UserComment] = metadata_bytes
+                    
+                    # Convert EXIF dict back to bytes
+                    exif_bytes = piexif.dump(exif_dict)
+                    
+                    # Save the image with updated EXIF data
+                    img.save(image_path, exif=exif_bytes)
+                    
+            return image_path
+        except Exception as e:
+            logger.error(f"Error updating metadata in {image_path}: {e}")
+            return image_path
+            
+    @staticmethod
+    def append_recipe_metadata(image_path, recipe_data) -> str:
+        """Append recipe metadata to an image's EXIF data"""
+        try:
+            # First, extract existing metadata
+            metadata = ExifUtils.extract_image_metadata(image_path)
+            
+            # Check if there's already recipe metadata
+            if metadata:
+                # Remove any existing recipe metadata
+                metadata = ExifUtils.remove_recipe_metadata(metadata)
+            
+            # Prepare simplified loras data
+            simplified_loras = []
+            for lora in recipe_data.get("loras", []):
+                simplified_lora = {
+                    "file_name": lora.get("file_name", ""),
+                    "hash": lora.get("hash", "").lower() if lora.get("hash") else "",
+                    "strength": float(lora.get("strength", 1.0)),
+                    "modelVersionId": lora.get("modelVersionId", ""),
+                    "modelName": lora.get("modelName", ""),
+                    "modelVersionName": lora.get("modelVersionName", ""),
+                }
+                simplified_loras.append(simplified_lora)            
+            
+            # Create recipe metadata JSON
+            recipe_metadata = {
+                'title': recipe_data.get('title', ''),
+                'base_model': recipe_data.get('base_model', ''),
+                'loras': simplified_loras,
+                'gen_params': recipe_data.get('gen_params', {}),
+                'tags': recipe_data.get('tags', [])
+            }
+            
+            # Convert to JSON string
+            recipe_metadata_json = json.dumps(recipe_metadata)
+            
+            # Create the recipe metadata marker
+            recipe_metadata_marker = f"Recipe metadata: {recipe_metadata_json}"
+            
+            # Append to existing metadata or create new one
+            new_metadata = f"{metadata} \n {recipe_metadata_marker}" if metadata else recipe_metadata_marker
+            
+            # Write back to the image
+            return ExifUtils.update_image_metadata(image_path, new_metadata)
+        except Exception as e:
+            logger.error(f"Error appending recipe metadata: {e}", exc_info=True)
+            return image_path
+
+    @staticmethod
+    def remove_recipe_metadata(user_comment):
+        """Remove recipe metadata from user comment"""
+        if not user_comment:
+            return ""
+        
+        # Find the recipe metadata marker
+        recipe_marker_index = user_comment.find("Recipe metadata: ")
+        if recipe_marker_index == -1:
+            return user_comment
+        
+        # If recipe metadata is not at the start, remove the preceding ", "
+        if recipe_marker_index >= 2 and user_comment[recipe_marker_index-2:recipe_marker_index] == ", ":
+            recipe_marker_index -= 2
+        
+        # Remove the recipe metadata part
+        # First, find where the metadata ends (next line or end of string)
+        next_line_index = user_comment.find("\n", recipe_marker_index)
+        if next_line_index == -1:
+            # Metadata is at the end of the string
+            return user_comment[:recipe_marker_index].rstrip()
+        else:
+            # Metadata is in the middle of the string
+            return user_comment[:recipe_marker_index] + user_comment[next_line_index:]
+            
+    @staticmethod
+    def optimize_image(image_data, target_width=250, format='webp', quality=85, preserve_metadata=True):
+        """
+        Optimize an image by resizing and converting to WebP format
+        
+        Args:
+            image_data: Binary image data or path to image file
+            target_width: Width to resize the image to (preserves aspect ratio)
+            format: Output format (default: webp)
+            quality: Output quality (0-100)
+            preserve_metadata: Whether to preserve EXIF metadata
+            
+        Returns:
+            Tuple of (optimized_image_data, extension)
+        """
+        try:
+            # Extract metadata if needed
+            metadata = None
+            if preserve_metadata:
+                if isinstance(image_data, str) and os.path.exists(image_data):
+                    # It's a file path
+                    metadata = ExifUtils.extract_image_metadata(image_data)
+                    img = Image.open(image_data)
+                else:
+                    # It's binary data
+                    temp_img = BytesIO(image_data)
+                    img = Image.open(temp_img)
+                    # Save to a temporary file to extract metadata
+                    import tempfile
+                    with tempfile.NamedTemporaryFile(suffix='.jpg', delete=False) as temp_file:
+                        temp_path = temp_file.name
+                        temp_file.write(image_data)
+                    metadata = ExifUtils.extract_image_metadata(temp_path)
+                    os.unlink(temp_path)
+            else:
+                # Just open the image without extracting metadata
+                if isinstance(image_data, str) and os.path.exists(image_data):
+                    img = Image.open(image_data)
+                else:
+                    img = Image.open(BytesIO(image_data))
+            
+            # Calculate new height to maintain aspect ratio
+            width, height = img.size
+            new_height = int(height * (target_width / width))
+            
+            # Resize the image
+            resized_img = img.resize((target_width, new_height), Image.LANCZOS)
+            
+            # Save to BytesIO in the specified format
+            output = BytesIO()
+            
+            # WebP format
+            if format.lower() == 'webp':
+                resized_img.save(output, format='WEBP', quality=quality)
+                extension = '.webp'
+            # JPEG format
+            elif format.lower() in ('jpg', 'jpeg'):
+                resized_img.save(output, format='JPEG', quality=quality)
+                extension = '.jpg'
+            # PNG format
+            elif format.lower() == 'png':
+                resized_img.save(output, format='PNG', optimize=True)
+                extension = '.png'
+            else:
+                # Default to WebP
+                resized_img.save(output, format='WEBP', quality=quality)
+                extension = '.webp'
+            
+            # Get the optimized image data
+            optimized_data = output.getvalue()
+            
+            # If we need to preserve metadata, write it to a temporary file
+            if preserve_metadata and metadata:
+                # For WebP format, we'll directly save with metadata
+                if format.lower() == 'webp':
+                    # Create a new BytesIO with metadata
+                    output_with_metadata = BytesIO()
+                    
+                    # Create EXIF data with user comment
+                    exif_dict = {'Exif': {piexif.ExifIFD.UserComment: b'UNICODE\0' + metadata.encode('utf-16be')}}
+                    exif_bytes = piexif.dump(exif_dict)
+                    
+                    # Save with metadata
+                    resized_img.save(output_with_metadata, format='WEBP', exif=exif_bytes, quality=quality)
+                    optimized_data = output_with_metadata.getvalue()
+                else:
+                    # For other formats, use the temporary file approach
+                    import tempfile
+                    with tempfile.NamedTemporaryFile(suffix=extension, delete=False) as temp_file:
+                        temp_path = temp_file.name
+                        temp_file.write(optimized_data)
+                    
+                    # Add the metadata back
+                    ExifUtils.update_image_metadata(temp_path, metadata)
+                    
+                    # Read the file with metadata
+                    with open(temp_path, 'rb') as f:
+                        optimized_data = f.read()
+                    
+                    # Clean up
+                    os.unlink(temp_path)
+            
+            return optimized_data, extension
+            
+        except Exception as e:
+            logger.error(f"Error optimizing image: {e}", exc_info=True)
+            # Return original data if optimization fails
+            if isinstance(image_data, str) and os.path.exists(image_data):
+                with open(image_data, 'rb') as f:
+                    return f.read(), os.path.splitext(image_data)[1]
+            return image_data, '.jpg'

py/utils/file_utils.py

@@ -2,10 +2,14 @@ import logging
 import os
 import hashlib
 import json
-from typing import Dict, Optional
+import time
+from typing import Dict, Optional, Type
 
-from .lora_metadata import extract_lora_metadata
-from .models import LoraMetadata
+from .model_utils import determine_base_model
+from .lora_metadata import extract_lora_metadata, extract_checkpoint_metadata
+from .models import BaseModelMetadata, LoraMetadata, CheckpointMetadata
+from .constants import PREVIEW_EXTENSIONS, CARD_PREVIEW_WIDTH
+from .exif_utils import ExifUtils
 
 logger = logging.getLogger(__name__)
 
@@ -13,35 +17,56 @@ async def calculate_sha256(file_path: str) -> str:
     """Calculate SHA256 hash of a file"""
     sha256_hash = hashlib.sha256()
     with open(file_path, "rb") as f:
-        for byte_block in iter(lambda: f.read(4096), b""):
+        for byte_block in iter(lambda: f.read(128 * 1024), b""):
             sha256_hash.update(byte_block)
     return sha256_hash.hexdigest()
 
-def _find_preview_file(base_name: str, dir_path: str) -> str:
+def find_preview_file(base_name: str, dir_path: str) -> str:
     """Find preview file for given base name in directory"""
-    preview_patterns = [
-        f"{base_name}.preview.png",
-        f"{base_name}.preview.jpg",
-        f"{base_name}.preview.jpeg",
-        f"{base_name}.preview.mp4",
-        f"{base_name}.png",
-        f"{base_name}.jpg", 
-        f"{base_name}.jpeg",
-        f"{base_name}.mp4"
-    ]
     
-    for pattern in preview_patterns:
-        full_pattern = os.path.join(dir_path, pattern)
+    for ext in PREVIEW_EXTENSIONS:
+        full_pattern = os.path.join(dir_path, f"{base_name}{ext}")
         if os.path.exists(full_pattern):
+            # Check if this is an image and not already webp
+            if ext.lower().endswith(('.jpg', '.jpeg', '.png')) and not ext.lower().endswith('.webp'):
+                try:
+                    # Optimize the image to webp format
+                    webp_path = os.path.join(dir_path, f"{base_name}.webp")
+                    
+                    # Use ExifUtils to optimize the image
+                    with open(full_pattern, 'rb') as f:
+                        image_data = f.read()
+                    
+                    optimized_data, _ = ExifUtils.optimize_image(
+                        image_data=image_data,
+                        target_width=CARD_PREVIEW_WIDTH,
+                        format='webp',
+                        quality=85,
+                        preserve_metadata=True
+                    )
+                    
+                    # Save the optimized webp file
+                    with open(webp_path, 'wb') as f:
+                        f.write(optimized_data)
+                    
+                    logger.debug(f"Optimized preview image from {full_pattern} to {webp_path}")
+                    return webp_path.replace(os.sep, "/")
+                except Exception as e:
+                    logger.error(f"Error optimizing preview image {full_pattern}: {e}")
+                    # Fall back to original file if optimization fails
+                    return full_pattern.replace(os.sep, "/")
+            
+            # Return the original path for webp images or non-image files
             return full_pattern.replace(os.sep, "/")
+    
     return ""
 
 def normalize_path(path: str) -> str:
     """Normalize file path to use forward slashes"""
     return path.replace(os.sep, "/") if path else path
 
-async def get_file_info(file_path: str) -> Optional[LoraMetadata]:
-    """Get basic file information as LoraMetadata object"""
+async def get_file_info(file_path: str, model_class: Type[BaseModelMetadata] = LoraMetadata) -> Optional[BaseModelMetadata]:
+    """Get basic file information as a model metadata object"""
     # First check if file actually exists and resolve symlinks
     try:
         real_path = os.path.realpath(file_path)
@@ -54,26 +79,81 @@ async def get_file_info(file_path: str) -> Optional[LoraMetadata]:
     base_name = os.path.splitext(os.path.basename(file_path))[0]
     dir_path = os.path.dirname(file_path)
     
-    preview_url = _find_preview_file(base_name, dir_path)
+    preview_url = find_preview_file(base_name, dir_path)
+    
+    # Check if a .json file exists with SHA256 hash to avoid recalculation
+    json_path = f"{os.path.splitext(file_path)[0]}.json"
+    sha256 = None
+    if os.path.exists(json_path):
+        try:
+            with open(json_path, 'r', encoding='utf-8') as f:
+                json_data = json.load(f)
+                if 'sha256' in json_data:
+                    sha256 = json_data['sha256'].lower()
+                    logger.debug(f"Using SHA256 from .json file for {file_path}")
+        except Exception as e:
+            logger.error(f"Error reading .json file for {file_path}: {e}")
+    
+    # If SHA256 is still not found, check for a .sha256 file
+    if sha256 is None:
+        sha256_file = f"{os.path.splitext(file_path)[0]}.sha256"
+        if os.path.exists(sha256_file):
+            try:
+                with open(sha256_file, 'r', encoding='utf-8') as f:
+                    sha256 = f.read().strip().lower()
+                    logger.debug(f"Using SHA256 from .sha256 file for {file_path}")
+            except Exception as e:
+                logger.error(f"Error reading .sha256 file for {file_path}: {e}")
 
     try:
-        metadata = LoraMetadata(
-            file_name=base_name,
-            model_name=base_name,
-            file_path=normalize_path(file_path),
-            size=os.path.getsize(real_path),
-            modified=os.path.getmtime(real_path),
-            sha256=await calculate_sha256(real_path),
-            base_model="Unknown",  # Will be updated later
-            usage_tips="",
-            notes="",
-            from_civitai=True,
-            preview_url=normalize_path(preview_url),
-        )
+        # If we didn't get SHA256 from the .json file, calculate it
+        if not sha256:
+            start_time = time.time()
+            sha256 = await calculate_sha256(real_path)
+            logger.debug(f"Calculated SHA256 for {file_path} in {time.time() - start_time:.2f} seconds")
+        
+        # Create default metadata based on model class
+        if model_class == CheckpointMetadata:
+            metadata = CheckpointMetadata(
+                file_name=base_name,
+                model_name=base_name,
+                file_path=normalize_path(file_path),
+                size=os.path.getsize(real_path),
+                modified=os.path.getmtime(real_path),
+                sha256=sha256,
+                base_model="Unknown",  # Will be updated later
+                preview_url=normalize_path(preview_url),
+                tags=[],
+                modelDescription="",
+                model_type="checkpoint"
+            )
+            
+            # Extract checkpoint-specific metadata
+            # model_info = await extract_checkpoint_metadata(real_path)
+            # metadata.base_model = model_info['base_model']
+            # if 'model_type' in model_info:
+            #     metadata.model_type = model_info['model_type']
+            
+        else:  # Default to LoraMetadata
+            metadata = LoraMetadata(
+                file_name=base_name,
+                model_name=base_name,
+                file_path=normalize_path(file_path),
+                size=os.path.getsize(real_path),
+                modified=os.path.getmtime(real_path),
+                sha256=sha256,
+                base_model="Unknown",  # Will be updated later
+                usage_tips="{}",
+                preview_url=normalize_path(preview_url),
+                tags=[],
+                modelDescription=""
+            )
+            
+            # Extract lora-specific metadata
+            model_info = await extract_lora_metadata(real_path)
+            metadata.base_model = model_info['base_model']
 
-        # create metadata file
-        base_model_info = await extract_lora_metadata(real_path)
-        metadata.base_model = base_model_info['base_model']
+        # Save metadata to file
         await save_metadata(file_path, metadata)
         
         return metadata
@@ -81,7 +161,7 @@ async def get_file_info(file_path: str) -> Optional[LoraMetadata]:
         logger.error(f"Error getting file info for {file_path}: {e}")
         return None
 
-async def save_metadata(file_path: str, metadata: LoraMetadata) -> None:
+async def save_metadata(file_path: str, metadata: BaseModelMetadata) -> None:
     """Save metadata to .metadata.json file"""
     metadata_path = f"{os.path.splitext(file_path)[0]}.metadata.json"
     try:
@@ -94,7 +174,7 @@ async def save_metadata(file_path: str, metadata: LoraMetadata) -> None:
     except Exception as e:
         print(f"Error saving metadata to {metadata_path}: {str(e)}")
 
-async def load_metadata(file_path: str) -> Optional[LoraMetadata]:
+async def load_metadata(file_path: str, model_class: Type[BaseModelMetadata] = LoraMetadata) -> Optional[BaseModelMetadata]:
     """Load metadata from .metadata.json file"""
     metadata_path = f"{os.path.splitext(file_path)[0]}.metadata.json"
     try:
@@ -103,28 +183,61 @@ async def load_metadata(file_path: str) -> Optional[LoraMetadata]:
                 data = json.load(f)
                 
                 needs_update = False
-                
-                if data['file_path'] != normalize_path(data['file_path']):
-                    data['file_path'] = normalize_path(data['file_path'])
+
+                # Check and normalize base model name
+                normalized_base_model = determine_base_model(data['base_model'])
+                if data['base_model'] != normalized_base_model:
+                    data['base_model'] = normalized_base_model
                     needs_update = True
                 
+                # Compare paths without extensions
+                stored_path_base = os.path.splitext(data['file_path'])[0]
+                current_path_base = os.path.splitext(normalize_path(file_path))[0]
+                if stored_path_base != current_path_base:
+                    data['file_path'] = normalize_path(file_path)
+                    needs_update = True
+                
+                # TODO: optimize preview image to webp format if not already done
                 preview_url = data.get('preview_url', '')
                 if not preview_url or not os.path.exists(preview_url):
                     base_name = os.path.splitext(os.path.basename(file_path))[0]
                     dir_path = os.path.dirname(file_path)
-                    new_preview_url = normalize_path(_find_preview_file(base_name, dir_path))
+                    new_preview_url = normalize_path(find_preview_file(base_name, dir_path))
                     if new_preview_url != preview_url:
                         data['preview_url'] = new_preview_url
                         needs_update = True
-                elif preview_url != normalize_path(preview_url):
-                    data['preview_url'] = normalize_path(preview_url)
+                else:
+                    # Compare preview paths without extensions
+                    stored_preview_base = os.path.splitext(preview_url)[0]
+                    current_preview_base = os.path.splitext(normalize_path(preview_url))[0]
+                    if stored_preview_base != current_preview_base:
+                        data['preview_url'] = normalize_path(preview_url)
+                        needs_update = True
+
+                # Ensure all fields are present
+                if 'tags' not in data:
+                    data['tags'] = []
+                    needs_update = True
+                    
+                if 'modelDescription' not in data:
+                    data['modelDescription'] = ""
+                    needs_update = True
+                    
+                # For checkpoint metadata
+                if model_class == CheckpointMetadata and 'model_type' not in data:
+                    data['model_type'] = "checkpoint"
+                    needs_update = True
+                
+                # For lora metadata
+                if model_class == LoraMetadata and 'usage_tips' not in data:
+                    data['usage_tips'] = "{}"
                     needs_update = True
                 
                 if needs_update:
                     with open(metadata_path, 'w', encoding='utf-8') as f:
                         json.dump(data, f, indent=2, ensure_ascii=False)
                 
-                return LoraMetadata.from_dict(data)
+                return model_class.from_dict(data)
                 
     except Exception as e:
         print(f"Error loading metadata from {metadata_path}: {str(e)}")

py/utils/lora_metadata.py

@@ -1,6 +1,7 @@
 from safetensors import safe_open
 from typing import Dict
 from .model_utils import determine_base_model
+import os
 
 async def extract_lora_metadata(file_path: str) -> Dict:
     """Extract essential metadata from safetensors file"""
@@ -13,4 +14,67 @@ async def extract_lora_metadata(file_path: str) -> Dict:
                 return {"base_model": base_model}
     except Exception as e:
         print(f"Error reading metadata from {file_path}: {str(e)}")
-    return {"base_model": "Unknown"} 
+    return {"base_model": "Unknown"}
+
+async def extract_checkpoint_metadata(file_path: str) -> dict:
+    """Extract metadata from a checkpoint file to determine model type and base model"""
+    try:
+        # Analyze filename for clues about the model
+        filename = os.path.basename(file_path).lower()
+        
+        model_info = {
+            'base_model': 'Unknown',
+            'model_type': 'checkpoint'
+        }
+        
+        # Detect base model from filename
+        if 'xl' in filename or 'sdxl' in filename:
+            model_info['base_model'] = 'SDXL'
+        elif 'sd3' in filename:
+            model_info['base_model'] = 'SD3'  
+        elif 'sd2' in filename or 'v2' in filename:
+            model_info['base_model'] = 'SD2.x'
+        elif 'sd1' in filename or 'v1' in filename:
+            model_info['base_model'] = 'SD1.5'
+        
+        # Detect model type from filename
+        if 'inpaint' in filename:
+            model_info['model_type'] = 'inpainting'
+        elif 'anime' in filename:
+            model_info['model_type'] = 'anime'
+        elif 'realistic' in filename:
+            model_info['model_type'] = 'realistic'
+        
+        # Try to peek at the safetensors file structure if available
+        if file_path.endswith('.safetensors'):
+            import json
+            import struct
+            
+            with open(file_path, 'rb') as f:
+                header_size = struct.unpack('<Q', f.read(8))[0]
+                header_json = f.read(header_size)
+                header = json.loads(header_json)
+                
+                # Look for specific keys to identify model type
+                metadata = header.get('__metadata__', {})
+                if metadata:
+                    # Try to determine if it's SDXL
+                    if any(key.startswith('conditioner.embedders.1') for key in header):
+                        model_info['base_model'] = 'SDXL'
+                    
+                    # Look for model type info
+                    if metadata.get('modelspec.architecture') == 'SD-XL':
+                        model_info['base_model'] = 'SDXL'
+                    elif metadata.get('modelspec.architecture') == 'SD-3':
+                        model_info['base_model'] = 'SD3'
+                    
+                    # Check for specific use case
+                    if metadata.get('modelspec.purpose') == 'inpainting':
+                        model_info['model_type'] = 'inpainting'
+        
+        return model_info
+        
+    except Exception as e:
+        logger.error(f"Error extracting checkpoint metadata for {file_path}: {e}")
+        # Return default values
+        return {'base_model': 'Unknown', 'model_type': 'checkpoint'}

py/utils/model_utils.py

@@ -2,13 +2,15 @@ from typing import Optional
 
 # Base model mapping based on version string
 BASE_MODEL_MAPPING = {
+    "sd_1.5": "SD 1.5",
     "sd-v1-5": "SD 1.5",
     "sd-v2-1": "SD 2.1",
     "sdxl": "SDXL 1.0",
     "sd-v2": "SD 2.0",
     "flux1": "Flux.1 D",
     "flux.1 d": "Flux.1 D",
-    "illustrious": "IL",
+    "illustrious": "Illustrious",
+    "il": "Illustrious",
     "pony": "Pony",
     "Hunyuan Video": "Hunyuan Video"
 }

py/utils/models.py

@@ -1,51 +1,38 @@
 from dataclasses import dataclass, asdict
-from typing import Dict, Optional
+from typing import Dict, Optional, List
 from datetime import datetime
 import os
 from .model_utils import determine_base_model
 
 @dataclass
-class LoraMetadata:
-    """Represents the metadata structure for a Lora model"""
-    file_name: str              # The filename without extension of the lora
-    model_name: str             # The lora's name defined by the creator, initially same as file_name
-    file_path: str              # Full path to the safetensors file
+class BaseModelMetadata:
+    """Base class for all model metadata structures"""
+    file_name: str              # The filename without extension
+    model_name: str             # The model's name defined by the creator
+    file_path: str              # Full path to the model file
     size: int                   # File size in bytes
     modified: float             # Last modified timestamp
     sha256: str                 # SHA256 hash of the file
-    base_model: str             # Base model (SD1.5/SD2.1/SDXL/etc.)
+    base_model: str             # Base model type (SD1.5/SD2.1/SDXL/etc.)
     preview_url: str            # Preview image URL
-    usage_tips: str = "{}"      # Usage tips for the model, json string
+    preview_nsfw_level: int = 0 # NSFW level of the preview image
     notes: str = ""             # Additional notes
-    from_civitai: bool = True  # Whether the lora is from Civitai
+    from_civitai: bool = True   # Whether from Civitai
     civitai: Optional[Dict] = None  # Civitai API data if available
+    tags: List[str] = None      # Model tags
+    modelDescription: str = ""  # Full model description
+
+    def __post_init__(self):
+        # Initialize empty lists to avoid mutable default parameter issue
+        if self.tags is None:
+            self.tags = []
 
     @classmethod
-    def from_dict(cls, data: Dict) -> 'LoraMetadata':
-        """Create LoraMetadata instance from dictionary"""
-        # Create a copy of the data to avoid modifying the input
+    def from_dict(cls, data: Dict) -> 'BaseModelMetadata':
+        """Create instance from dictionary"""
         data_copy = data.copy()
         return cls(**data_copy)
 
-    @classmethod
-    def from_civitai_info(cls, version_info: Dict, file_info: Dict, save_path: str) -> 'LoraMetadata':
-        """Create LoraMetadata instance from Civitai version info"""
-        file_name = file_info['name']
-        base_model = determine_base_model(version_info.get('baseModel', ''))
-        
-        return cls(
-            file_name=os.path.splitext(file_name)[0],
-            model_name=version_info.get('model').get('name', os.path.splitext(file_name)[0]),
-            file_path=save_path.replace(os.sep, '/'),
-            size=file_info.get('sizeKB', 0) * 1024,
-            modified=datetime.now().timestamp(),
-            sha256=file_info['hashes'].get('SHA256', ''),
-            base_model=base_model,
-            preview_url=None,  # Will be updated after preview download
-            from_civitai=True,
-            civitai=version_info
-        )
-
     def to_dict(self) -> Dict:
         """Convert to dictionary for JSON serialization"""
         return asdict(self)
@@ -66,3 +53,55 @@ class LoraMetadata:
             self.modified = os.path.getmtime(file_path)
             self.file_path = file_path.replace(os.sep, '/')
 
+@dataclass
+class LoraMetadata(BaseModelMetadata):
+    """Represents the metadata structure for a Lora model"""
+    usage_tips: str = "{}"      # Usage tips for the model, json string
+
+    @classmethod
+    def from_civitai_info(cls, version_info: Dict, file_info: Dict, save_path: str) -> 'LoraMetadata':
+        """Create LoraMetadata instance from Civitai version info"""
+        file_name = file_info['name']
+        base_model = determine_base_model(version_info.get('baseModel', ''))
+        
+        return cls(
+            file_name=os.path.splitext(file_name)[0],
+            model_name=version_info.get('model').get('name', os.path.splitext(file_name)[0]),
+            file_path=save_path.replace(os.sep, '/'),
+            size=file_info.get('sizeKB', 0) * 1024,
+            modified=datetime.now().timestamp(),
+            sha256=file_info['hashes'].get('SHA256', '').lower(),
+            base_model=base_model,
+            preview_url=None,  # Will be updated after preview download
+            preview_nsfw_level=0, # Will be updated after preview download
+            from_civitai=True,
+            civitai=version_info
+        )
+
+@dataclass
+class CheckpointMetadata(BaseModelMetadata):
+    """Represents the metadata structure for a Checkpoint model"""
+    model_type: str = "checkpoint"  # Model type (checkpoint, inpainting, etc.)
+
+    @classmethod
+    def from_civitai_info(cls, version_info: Dict, file_info: Dict, save_path: str) -> 'CheckpointMetadata':
+        """Create CheckpointMetadata instance from Civitai version info"""
+        file_name = file_info['name']
+        base_model = determine_base_model(version_info.get('baseModel', ''))
+        model_type = version_info.get('type', 'checkpoint')
+        
+        return cls(
+            file_name=os.path.splitext(file_name)[0],
+            model_name=version_info.get('model').get('name', os.path.splitext(file_name)[0]),
+            file_path=save_path.replace(os.sep, '/'),
+            size=file_info.get('sizeKB', 0) * 1024,
+            modified=datetime.now().timestamp(),
+            sha256=file_info['hashes'].get('SHA256', '').lower(),
+            base_model=base_model,
+            preview_url=None,  # Will be updated after preview download
+            preview_nsfw_level=0,
+            from_civitai=True,
+            civitai=version_info,
+            model_type=model_type
+        )
+

py/utils/routes_common.py

@@ -0,0 +1,503 @@
+import os
+import json
+import logging
+from typing import Dict, List, Callable, Awaitable
+from aiohttp import web
+
+from .model_utils import determine_base_model
+from .constants import PREVIEW_EXTENSIONS, CARD_PREVIEW_WIDTH
+from ..config import config
+from ..services.civitai_client import CivitaiClient
+from ..utils.exif_utils import ExifUtils
+from ..services.download_manager import DownloadManager
+
+logger = logging.getLogger(__name__)
+
+
+class ModelRouteUtils:
+    """Shared utilities for model routes (LoRAs, Checkpoints, etc.)"""
+
+    @staticmethod
+    async def load_local_metadata(metadata_path: str) -> Dict:
+        """Load local metadata file"""
+        if os.path.exists(metadata_path):
+            try:
+                with open(metadata_path, 'r', encoding='utf-8') as f:
+                    return json.load(f)
+            except Exception as e:
+                logger.error(f"Error loading metadata from {metadata_path}: {e}")
+        return {}
+
+    @staticmethod
+    async def handle_not_found_on_civitai(metadata_path: str, local_metadata: Dict) -> None:
+        """Handle case when model is not found on CivitAI"""
+        local_metadata['from_civitai'] = False
+        with open(metadata_path, 'w', encoding='utf-8') as f:
+            json.dump(local_metadata, f, indent=2, ensure_ascii=False)
+
+    @staticmethod
+    async def update_model_metadata(metadata_path: str, local_metadata: Dict, 
+                                  civitai_metadata: Dict, client: CivitaiClient) -> None:
+        """Update local metadata with CivitAI data"""
+        local_metadata['civitai'] = civitai_metadata
+        
+        # Update model name if available
+        if 'model' in civitai_metadata:
+            if civitai_metadata.get('model', {}).get('name'):
+                local_metadata['model_name'] = civitai_metadata['model']['name']
+        
+            # Fetch additional model metadata (description and tags) if we have model ID
+            model_id = civitai_metadata['modelId']
+            if model_id:
+                model_metadata, _ = await client.get_model_metadata(str(model_id))
+                if model_metadata:
+                    local_metadata['modelDescription'] = model_metadata.get('description', '')
+                    local_metadata['tags'] = model_metadata.get('tags', [])
+        
+        # Update base model
+        local_metadata['base_model'] = determine_base_model(civitai_metadata.get('baseModel'))
+        
+        # Update preview if needed
+        if not local_metadata.get('preview_url') or not os.path.exists(local_metadata['preview_url']):
+            first_preview = next((img for img in civitai_metadata.get('images', [])), None)
+            if first_preview:
+                # Determine if content is video or image
+                is_video = first_preview['type'] == 'video'
+                
+                if is_video:
+                    # For videos use .mp4 extension
+                    preview_ext = '.mp4'
+                else:
+                    # For images use .webp extension
+                    preview_ext = '.webp'
+                
+                base_name = os.path.splitext(os.path.splitext(os.path.basename(metadata_path))[0])[0]
+                preview_filename = base_name + preview_ext
+                preview_path = os.path.join(os.path.dirname(metadata_path), preview_filename)
+                
+                if is_video:
+                    # Download video as is
+                    if await client.download_preview_image(first_preview['url'], preview_path):
+                        local_metadata['preview_url'] = preview_path.replace(os.sep, '/')
+                        local_metadata['preview_nsfw_level'] = first_preview.get('nsfwLevel', 0)
+                else:
+                    # For images, download and then optimize to WebP
+                    temp_path = preview_path + ".temp"
+                    if await client.download_preview_image(first_preview['url'], temp_path):
+                        try:
+                            # Read the downloaded image
+                            with open(temp_path, 'rb') as f:
+                                image_data = f.read()
+                            
+                            # Optimize and convert to WebP
+                            optimized_data, _ = ExifUtils.optimize_image(
+                                image_data=image_data,
+                                target_width=CARD_PREVIEW_WIDTH,
+                                format='webp',
+                                quality=85,
+                                preserve_metadata=True
+                            )
+                            
+                            # Save the optimized WebP image
+                            with open(preview_path, 'wb') as f:
+                                f.write(optimized_data)
+                                
+                            # Update metadata
+                            local_metadata['preview_url'] = preview_path.replace(os.sep, '/')
+                            local_metadata['preview_nsfw_level'] = first_preview.get('nsfwLevel', 0)
+                            
+                            # Remove the temporary file
+                            if os.path.exists(temp_path):
+                                os.remove(temp_path)
+                                
+                        except Exception as e:
+                            logger.error(f"Error optimizing preview image: {e}")
+                            # If optimization fails, try to use the downloaded image directly
+                            if os.path.exists(temp_path):
+                                os.rename(temp_path, preview_path)
+                                local_metadata['preview_url'] = preview_path.replace(os.sep, '/')
+                                local_metadata['preview_nsfw_level'] = first_preview.get('nsfwLevel', 0)
+
+        # Save updated metadata
+        with open(metadata_path, 'w', encoding='utf-8') as f:
+            json.dump(local_metadata, f, indent=2, ensure_ascii=False)
+
+    @staticmethod
+    async def fetch_and_update_model(
+        sha256: str, 
+        file_path: str, 
+        model_data: dict,
+        update_cache_func: Callable[[str, str, Dict], Awaitable[bool]]
+    ) -> bool:
+        """Fetch and update metadata for a single model
+        
+        Args:
+            sha256: SHA256 hash of the model file
+            file_path: Path to the model file
+            model_data: The model object in cache to update
+            update_cache_func: Function to update the cache with new metadata
+            
+        Returns:
+            bool: True if successful, False otherwise
+        """
+        client = CivitaiClient()
+        try:
+            metadata_path = os.path.splitext(file_path)[0] + '.metadata.json'
+            
+            # Check if model metadata exists
+            local_metadata = await ModelRouteUtils.load_local_metadata(metadata_path)
+
+            # Fetch metadata from Civitai
+            civitai_metadata = await client.get_model_by_hash(sha256)
+            if not civitai_metadata:
+                # Mark as not from CivitAI if not found
+                local_metadata['from_civitai'] = False
+                model_data['from_civitai'] = False
+                with open(metadata_path, 'w', encoding='utf-8') as f:
+                    json.dump(local_metadata, f, indent=2, ensure_ascii=False)
+                return False
+
+            # Update metadata
+            await ModelRouteUtils.update_model_metadata(
+                metadata_path, 
+                local_metadata, 
+                civitai_metadata, 
+                client
+            )
+            
+            # Update cache object directly
+            model_data.update({
+                'model_name': local_metadata.get('model_name'),
+                'preview_url': local_metadata.get('preview_url'),
+                'from_civitai': True,
+                'civitai': civitai_metadata
+            })
+            
+            # Update cache using the provided function
+            await update_cache_func(file_path, file_path, local_metadata)
+                
+            return True
+
+        except Exception as e:
+            logger.error(f"Error fetching CivitAI data: {e}")
+            return False
+        finally:
+            await client.close()
+    
+    @staticmethod
+    def filter_civitai_data(data: Dict) -> Dict:
+        """Filter relevant fields from CivitAI data"""
+        if not data:
+            return {}
+            
+        fields = [
+            "id", "modelId", "name", "createdAt", "updatedAt", 
+            "publishedAt", "trainedWords", "baseModel", "description",
+            "model", "images"
+        ]
+        return {k: data[k] for k in fields if k in data}
+
+    @staticmethod
+    async def delete_model_files(target_dir: str, file_name: str, file_monitor=None) -> List[str]:
+        """Delete model and associated files
+        
+        Args:
+            target_dir: Directory containing the model files
+            file_name: Base name of the model file without extension
+            file_monitor: Optional file monitor to ignore delete events
+            
+        Returns:
+            List of deleted file paths
+        """
+        patterns = [
+            f"{file_name}.safetensors",  # Required
+            f"{file_name}.metadata.json",
+        ]
+        
+        # Add all preview file extensions
+        for ext in PREVIEW_EXTENSIONS:
+            patterns.append(f"{file_name}{ext}")
+        
+        deleted = []
+        main_file = patterns[0]
+        main_path = os.path.join(target_dir, main_file).replace(os.sep, '/')
+        
+        if os.path.exists(main_path):
+            # Notify file monitor to ignore delete event if available
+            if file_monitor:
+                file_monitor.handler.add_ignore_path(main_path, 0)
+            
+            # Delete file
+            os.remove(main_path)
+            deleted.append(main_path)
+        else:
+            logger.warning(f"Model file not found: {main_file}")
+            
+        # Delete optional files
+        for pattern in patterns[1:]:
+            path = os.path.join(target_dir, pattern)
+            if os.path.exists(path):
+                try:
+                    os.remove(path)
+                    deleted.append(pattern)
+                except Exception as e:
+                    logger.warning(f"Failed to delete {pattern}: {e}")
+                    
+        return deleted
+    
+    @staticmethod
+    def get_multipart_ext(filename):
+        """Get extension that may have multiple parts like .metadata.json"""
+        parts = filename.split(".")
+        if len(parts) > 2:  # If contains multi-part extension
+            return "." + ".".join(parts[-2:])  # Take the last two parts, like ".metadata.json"
+        return os.path.splitext(filename)[1]  # Otherwise take the regular extension, like ".safetensors"
+
+    # New common endpoint handlers
+
+    @staticmethod
+    async def handle_delete_model(request: web.Request, scanner) -> web.Response:
+        """Handle model deletion request
+        
+        Args:
+            request: The aiohttp request
+            scanner: The model scanner instance with cache management methods
+            
+        Returns:
+            web.Response: The HTTP response
+        """
+        try:
+            data = await request.json()
+            file_path = data.get('file_path')
+            if not file_path:
+                return web.Response(text='Model path is required', status=400)
+
+            target_dir = os.path.dirname(file_path)
+            file_name = os.path.splitext(os.path.basename(file_path))[0]
+            
+            # Get the file monitor from the scanner if available
+            file_monitor = getattr(scanner, 'file_monitor', None)
+            
+            deleted_files = await ModelRouteUtils.delete_model_files(
+                target_dir, 
+                file_name,
+                file_monitor
+            )
+            
+            # Remove from cache
+            cache = await scanner.get_cached_data()
+            cache.raw_data = [item for item in cache.raw_data if item['file_path'] != file_path]
+            await cache.resort()
+
+            # Update hash index if available
+            if hasattr(scanner, '_hash_index') and scanner._hash_index:
+                scanner._hash_index.remove_by_path(file_path)
+            
+            return web.json_response({
+                'success': True,
+                'deleted_files': deleted_files
+            })
+            
+        except Exception as e:
+            logger.error(f"Error deleting model: {e}", exc_info=True)
+            return web.Response(text=str(e), status=500)
+
+    @staticmethod
+    async def handle_fetch_civitai(request: web.Request, scanner) -> web.Response:
+        """Handle CivitAI metadata fetch request
+        
+        Args:
+            request: The aiohttp request
+            scanner: The model scanner instance with cache management methods
+            
+        Returns:
+            web.Response: The HTTP response
+        """
+        try:
+            data = await request.json()
+            metadata_path = os.path.splitext(data['file_path'])[0] + '.metadata.json'
+            
+            # Check if model metadata exists
+            local_metadata = await ModelRouteUtils.load_local_metadata(metadata_path)
+            if not local_metadata or not local_metadata.get('sha256'):
+                return web.json_response({"success": False, "error": "No SHA256 hash found"}, status=400)
+
+            # Create a client for fetching from Civitai
+            client = CivitaiClient()
+            try:
+                # Fetch and update metadata
+                civitai_metadata = await client.get_model_by_hash(local_metadata["sha256"])
+                if not civitai_metadata:
+                    await ModelRouteUtils.handle_not_found_on_civitai(metadata_path, local_metadata)
+                    return web.json_response({"success": False, "error": "Not found on CivitAI"}, status=404)
+
+                await ModelRouteUtils.update_model_metadata(metadata_path, local_metadata, civitai_metadata, client)
+                
+                # Update the cache
+                await scanner.update_single_model_cache(data['file_path'], data['file_path'], local_metadata)
+                
+                return web.json_response({"success": True})
+            finally:
+                await client.close()
+
+        except Exception as e:
+            logger.error(f"Error fetching from CivitAI: {e}", exc_info=True)
+            return web.json_response({"success": False, "error": str(e)}, status=500)
+
+    @staticmethod
+    async def handle_replace_preview(request: web.Request, scanner) -> web.Response:
+        """Handle preview image replacement request
+        
+        Args:
+            request: The aiohttp request
+            scanner: The model scanner instance with methods to update cache
+            
+        Returns:
+            web.Response: The HTTP response
+        """
+        try:
+            reader = await request.multipart()
+            
+            # Read preview file data
+            field = await reader.next()
+            if field.name != 'preview_file':
+                raise ValueError("Expected 'preview_file' field")
+            content_type = field.headers.get('Content-Type', 'image/png')
+            preview_data = await field.read()
+            
+            # Read model path
+            field = await reader.next()
+            if field.name != 'model_path':
+                raise ValueError("Expected 'model_path' field")
+            model_path = (await field.read()).decode()
+            
+            # Save preview file
+            base_name = os.path.splitext(os.path.basename(model_path))[0]
+            folder = os.path.dirname(model_path)
+            
+            # Determine if content is video or image
+            if content_type.startswith('video/'):
+                # For videos, keep original format and use .mp4 extension
+                extension = '.mp4'
+                optimized_data = preview_data
+            else:
+                # For images, optimize and convert to WebP
+                optimized_data, _ = ExifUtils.optimize_image(
+                    image_data=preview_data,
+                    target_width=CARD_PREVIEW_WIDTH,
+                    format='webp',
+                    quality=85,
+                    preserve_metadata=True
+                )
+                extension = '.webp'  # Use .webp without .preview part
+            
+            preview_path = os.path.join(folder, base_name + extension).replace(os.sep, '/')
+            
+            with open(preview_path, 'wb') as f:
+                f.write(optimized_data)
+            
+            # Update preview path in metadata
+            metadata_path = os.path.splitext(model_path)[0] + '.metadata.json'
+            if os.path.exists(metadata_path):
+                try:
+                    with open(metadata_path, 'r', encoding='utf-8') as f:
+                        metadata = json.load(f)
+                    
+                    # Update preview_url directly in the metadata dict
+                    metadata['preview_url'] = preview_path
+                    
+                    with open(metadata_path, 'w', encoding='utf-8') as f:
+                        json.dump(metadata, f, indent=2, ensure_ascii=False)
+                except Exception as e:
+                    logger.error(f"Error updating metadata: {e}")
+            
+            # Update preview URL in scanner cache
+            if hasattr(scanner, 'update_preview_in_cache'):
+                await scanner.update_preview_in_cache(model_path, preview_path)
+            
+            return web.json_response({
+                "success": True,
+                "preview_url": config.get_preview_static_url(preview_path)
+            })
+            
+        except Exception as e:
+            logger.error(f"Error replacing preview: {e}", exc_info=True)
+            return web.Response(text=str(e), status=500)
+
+    @staticmethod
+    async def handle_download_model(request: web.Request, download_manager: DownloadManager, model_type="lora") -> web.Response:
+        """Handle model download request
+        
+        Args:
+            request: The aiohttp request
+            download_manager: Instance of DownloadManager
+            model_type: Type of model ('lora' or 'checkpoint')
+            
+        Returns:
+            web.Response: The HTTP response
+        """
+        try:
+            data = await request.json()
+            
+            # Create progress callback
+            async def progress_callback(progress):
+                from ..services.websocket_manager import ws_manager
+                await ws_manager.broadcast({
+                    'status': 'progress',
+                    'progress': progress
+                })
+            
+            # Check which identifier is provided
+            download_url = data.get('download_url')
+            model_hash = data.get('model_hash')
+            model_version_id = data.get('model_version_id')
+            
+            # Validate that at least one identifier is provided
+            if not any([download_url, model_hash, model_version_id]):
+                return web.Response(
+                    status=400, 
+                    text="Missing required parameter: Please provide either 'download_url', 'hash', or 'modelVersionId'"
+                )
+            
+            # Use the correct root directory based on model type
+            root_key = 'checkpoint_root' if model_type == 'checkpoint' else 'lora_root'
+            save_dir = data.get(root_key)
+            
+            result = await download_manager.download_from_civitai(
+                download_url=download_url,
+                model_hash=model_hash,
+                model_version_id=model_version_id,
+                save_dir=save_dir,
+                relative_path=data.get('relative_path', ''),
+                progress_callback=progress_callback,
+                model_type=model_type
+            )
+            
+            if not result.get('success', False):
+                error_message = result.get('error', 'Unknown error')
+                
+                # Return 401 for early access errors
+                if 'early access' in error_message.lower():
+                    logger.warning(f"Early access download failed: {error_message}")
+                    return web.Response(
+                        status=401,  # Use 401 status code to match Civitai's response
+                        text=f"Early Access Restriction: {error_message}"
+                    )
+                
+                return web.Response(status=500, text=error_message)
+            
+            return web.json_response(result)
+            
+        except Exception as e:
+            error_message = str(e)
+            
+            # Check if this might be an early access error
+            if '401' in error_message:
+                logger.warning(f"Early access error (401): {error_message}")
+                return web.Response(
+                    status=401,
+                    text="Early Access Restriction: This model requires purchase. Please buy early access on Civitai.com."
+                )
+            
+            logger.error(f"Error downloading {model_type}: {error_message}")
+            return web.Response(status=500, text=error_message)

py/utils/utils.py

@@ -0,0 +1,116 @@
+from difflib import SequenceMatcher
+import requests
+import tempfile
+import re
+from bs4 import BeautifulSoup
+
+def download_twitter_image(url):
+    """Download image from a URL containing twitter:image meta tag
+    
+    Args:
+        url (str): The URL to download image from
+        
+    Returns:
+        str: Path to downloaded temporary image file
+    """
+    try:
+        # Download page content
+        response = requests.get(url)
+        response.raise_for_status()
+        
+        # Parse HTML
+        soup = BeautifulSoup(response.text, 'html.parser')
+        
+        # Find twitter:image meta tag
+        meta_tag = soup.find('meta', attrs={'property': 'twitter:image'})
+        if not meta_tag:
+            return None
+            
+        image_url = meta_tag['content']
+        
+        # Download image
+        image_response = requests.get(image_url)
+        image_response.raise_for_status()
+        
+        # Save to temp file
+        with tempfile.NamedTemporaryFile(delete=False, suffix='.jpg') as temp_file:
+            temp_file.write(image_response.content)
+            return temp_file.name
+            
+    except Exception as e:
+        print(f"Error downloading twitter image: {e}")
+        return None
+
+def download_civitai_image(url):
+    """Download image from a URL containing avatar image with specific class and style attributes
+    
+    Args:
+        url (str): The URL to download image from
+        
+    Returns:
+        str: Path to downloaded temporary image file
+    """
+    try:
+        # Download page content
+        response = requests.get(url)
+        response.raise_for_status()
+        
+        # Parse HTML
+        soup = BeautifulSoup(response.text, 'html.parser')
+        
+        # Find image with specific class and style attributes
+        image = soup.select_one('img.EdgeImage_image__iH4_q.max-h-full.w-auto.max-w-full')
+        
+        if not image or 'src' not in image.attrs:
+            return None
+            
+        image_url = image['src']
+        
+        # Download image
+        image_response = requests.get(image_url)
+        image_response.raise_for_status()
+        
+        # Save to temp file
+        with tempfile.NamedTemporaryFile(delete=False, suffix='.jpg') as temp_file:
+            temp_file.write(image_response.content)
+            return temp_file.name
+            
+    except Exception as e:
+        print(f"Error downloading civitai avatar: {e}")
+        return None
+
+def fuzzy_match(text: str, pattern: str, threshold: float = 0.7) -> bool:
+        """
+        Check if text matches pattern using fuzzy matching.
+        Returns True if similarity ratio is above threshold.
+        """
+        if not pattern or not text:
+            return False
+        
+        # Convert both to lowercase for case-insensitive matching
+        text = text.lower()
+        pattern = pattern.lower()
+        
+        # Split pattern into words
+        search_words = pattern.split()
+        
+        # Check each word
+        for word in search_words:
+            # First check if word is a substring (faster)
+            if word in text:
+                continue
+            
+            # If not found as substring, try fuzzy matching
+            # Check if any part of the text matches this word
+            found_match = False
+            for text_part in text.split():
+                ratio = SequenceMatcher(None, text_part, word).ratio()
+                if ratio >= threshold:
+                    found_match = True
+                    break
+                
+            if not found_match:
+                return False
+        
+        # All words found either as substrings or fuzzy matches
+        return True

py/workflow/__init__.py

@@ -0,0 +1,3 @@
+"""
+ComfyUI workflow parsing module to extract generation parameters
+""" 

py/workflow/cli.py

@@ -0,0 +1,58 @@
+"""
+Command-line interface for the ComfyUI workflow parser
+"""
+import argparse
+import json
+import os
+import logging
+import sys
+from .parser import parse_workflow
+
+logging.basicConfig(
+    level=logging.INFO,
+    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
+    handlers=[logging.StreamHandler()]
+)
+logger = logging.getLogger(__name__)
+
+def main():
+    """Entry point for the CLI"""
+    parser = argparse.ArgumentParser(description='Parse ComfyUI workflow files')
+    parser.add_argument('input', help='Input workflow JSON file path')
+    parser.add_argument('-o', '--output', help='Output JSON file path')
+    parser.add_argument('-p', '--pretty', action='store_true', help='Pretty print JSON output')
+    parser.add_argument('--debug', action='store_true', help='Enable debug logging')
+    
+    args = parser.parse_args()
+    
+    # Set logging level
+    if args.debug:
+        logging.getLogger().setLevel(logging.DEBUG)
+    
+    # Validate input file
+    if not os.path.isfile(args.input):
+        logger.error(f"Input file not found: {args.input}")
+        sys.exit(1)
+    
+    # Parse workflow
+    try:
+        result = parse_workflow(args.input, args.output)
+        
+        # Print result to console if output file not specified
+        if not args.output:
+            if args.pretty:
+                print(json.dumps(result, indent=4))
+            else:
+                print(json.dumps(result))
+        else:
+            logger.info(f"Output saved to: {args.output}")
+            
+    except Exception as e:
+        logger.error(f"Error parsing workflow: {e}")
+        if args.debug:
+            import traceback
+            traceback.print_exc()
+        sys.exit(1)
+
+if __name__ == "__main__":
+    main() 

py/workflow/ext/__init__.py

@@ -0,0 +1,3 @@
+"""
+Extension directory for custom node mappers
+""" 

py/workflow/ext/comfyui_core.py

@@ -0,0 +1,285 @@
+"""
+ComfyUI Core nodes mappers extension for workflow parsing
+"""
+import logging
+from typing import Dict, Any, List
+
+logger = logging.getLogger(__name__)
+
+# =============================================================================
+# Transform Functions
+# =============================================================================
+
+def transform_random_noise(inputs: Dict) -> Dict:
+    """Transform function for RandomNoise node"""
+    return {"seed": str(inputs.get("noise_seed", ""))}
+
+def transform_ksampler_select(inputs: Dict) -> Dict:
+    """Transform function for KSamplerSelect node"""
+    return {"sampler": inputs.get("sampler_name", "")}
+
+def transform_basic_scheduler(inputs: Dict) -> Dict:
+    """Transform function for BasicScheduler node"""
+    result = {
+        "scheduler": inputs.get("scheduler", ""),
+        "denoise": str(inputs.get("denoise", "1.0"))
+    }
+    
+    # Get steps from inputs or steps input
+    if "steps" in inputs:
+        if isinstance(inputs["steps"], str):
+            result["steps"] = inputs["steps"]
+        elif isinstance(inputs["steps"], dict) and "value" in inputs["steps"]:
+            result["steps"] = str(inputs["steps"]["value"])
+        else:
+            result["steps"] = str(inputs["steps"])
+    
+    return result
+
+def transform_basic_guider(inputs: Dict) -> Dict:
+    """Transform function for BasicGuider node"""
+    result = {}
+    
+    # Process conditioning
+    if "conditioning" in inputs:
+        if isinstance(inputs["conditioning"], str):
+            result["prompt"] = inputs["conditioning"]
+        elif isinstance(inputs["conditioning"], dict):
+            result["conditioning"] = inputs["conditioning"]
+    
+    # Get model information if needed
+    if "model" in inputs and isinstance(inputs["model"], dict):
+        result["model"] = inputs["model"]
+    
+    return result
+
+def transform_model_sampling_flux(inputs: Dict) -> Dict:
+    """Transform function for ModelSamplingFlux - mostly a pass-through node"""
+    # This node is primarily used for routing, so we mostly pass through values
+    
+    return inputs["model"]
+
+def transform_sampler_custom_advanced(inputs: Dict) -> Dict:
+    """Transform function for SamplerCustomAdvanced node"""
+    result = {}
+    
+    # Extract seed from noise
+    if "noise" in inputs and isinstance(inputs["noise"], dict):
+        result["seed"] = str(inputs["noise"].get("seed", ""))
+    
+    # Extract sampler info
+    if "sampler" in inputs and isinstance(inputs["sampler"], dict):
+        sampler = inputs["sampler"].get("sampler", "")
+        if sampler:
+            result["sampler"] = sampler
+    
+    # Extract scheduler, steps, denoise from sigmas
+    if "sigmas" in inputs and isinstance(inputs["sigmas"], dict):
+        sigmas = inputs["sigmas"]
+        result["scheduler"] = sigmas.get("scheduler", "")
+        result["steps"] = str(sigmas.get("steps", ""))
+        result["denoise"] = str(sigmas.get("denoise", "1.0"))
+    
+    # Extract prompt and guidance from guider
+    if "guider" in inputs and isinstance(inputs["guider"], dict):
+        guider = inputs["guider"]
+        
+        # Get prompt from conditioning
+        if "conditioning" in guider and isinstance(guider["conditioning"], str):
+            result["prompt"] = guider["conditioning"]
+        elif "conditioning" in guider and isinstance(guider["conditioning"], dict):
+            result["guidance"] = guider["conditioning"].get("guidance", "")
+            result["prompt"] = guider["conditioning"].get("prompt", "")
+
+        if "model" in guider and isinstance(guider["model"], dict):
+            result["checkpoint"] = guider["model"].get("checkpoint", "")
+            result["loras"] = guider["model"].get("loras", "")
+            result["clip_skip"] = str(int(guider["model"].get("clip_skip", "-1")) * -1)
+    
+    # Extract dimensions from latent_image
+    if "latent_image" in inputs and isinstance(inputs["latent_image"], dict):
+        latent = inputs["latent_image"]
+        width = latent.get("width", 0)
+        height = latent.get("height", 0)
+        if width and height:
+            result["width"] = width
+            result["height"] = height
+            result["size"] = f"{width}x{height}"
+    
+    return result
+
+def transform_ksampler(inputs: Dict) -> Dict:
+    """Transform function for KSampler nodes"""
+    result = {
+        "seed": str(inputs.get("seed", "")),
+        "steps": str(inputs.get("steps", "")),
+        "cfg": str(inputs.get("cfg", "")),
+        "sampler": inputs.get("sampler_name", ""),
+        "scheduler": inputs.get("scheduler", ""),
+    }
+    
+    # Process positive prompt
+    if "positive" in inputs:
+        result["prompt"] = inputs["positive"]
+    
+    # Process negative prompt
+    if "negative" in inputs:
+        result["negative_prompt"] = inputs["negative"]
+    
+    # Get dimensions from latent image
+    if "latent_image" in inputs and isinstance(inputs["latent_image"], dict):
+        width = inputs["latent_image"].get("width", 0)
+        height = inputs["latent_image"].get("height", 0)
+        if width and height:
+            result["size"] = f"{width}x{height}"
+    
+    # Add clip_skip if present
+    if "clip_skip" in inputs:
+        result["clip_skip"] = str(inputs.get("clip_skip", ""))
+
+    # Add guidance if present
+    if "guidance" in inputs:
+        result["guidance"] = str(inputs.get("guidance", ""))
+
+    # Add model if present
+    if "model" in inputs:
+        result["checkpoint"] = inputs.get("model", {}).get("checkpoint", "")
+        result["loras"] = inputs.get("model", {}).get("loras", "")
+        result["clip_skip"] = str(inputs.get("model", {}).get("clip_skip", -1) * -1)
+        
+    return result
+
+def transform_empty_latent(inputs: Dict) -> Dict:
+    """Transform function for EmptyLatentImage nodes"""
+    width = inputs.get("width", 0)
+    height = inputs.get("height", 0)
+    return {"width": width, "height": height, "size": f"{width}x{height}"}
+
+def transform_clip_text(inputs: Dict) -> Any:
+    """Transform function for CLIPTextEncode nodes"""
+    return inputs.get("text", "")
+
+def transform_flux_guidance(inputs: Dict) -> Dict:
+    """Transform function for FluxGuidance nodes"""
+    result = {}
+    
+    if "guidance" in inputs:
+        result["guidance"] = inputs["guidance"]
+    
+    if "conditioning" in inputs:
+        conditioning = inputs["conditioning"]
+        if isinstance(conditioning, str):
+            result["prompt"] = conditioning
+        else:
+            result["prompt"] = "Unknown prompt"
+    
+    return result
+
+def transform_unet_loader(inputs: Dict) -> Dict:
+    """Transform function for UNETLoader node"""
+    unet_name = inputs.get("unet_name", "")
+    return {"checkpoint": unet_name} if unet_name else {}
+
+def transform_checkpoint_loader(inputs: Dict) -> Dict:
+    """Transform function for CheckpointLoaderSimple node"""
+    ckpt_name = inputs.get("ckpt_name", "")
+    return {"checkpoint": ckpt_name} if ckpt_name else {}
+
+def transform_latent_upscale_by(inputs: Dict) -> Dict:
+    """Transform function for LatentUpscaleBy node"""
+    result = {}
+    
+    width = inputs["samples"].get("width", 0) * inputs["scale_by"]
+    height = inputs["samples"].get("height", 0) * inputs["scale_by"]
+    result["width"] = width
+    result["height"] = height
+    result["size"] = f"{width}x{height}"
+    
+    return result
+
+def transform_clip_set_last_layer(inputs: Dict) -> Dict:
+    """Transform function for CLIPSetLastLayer node"""
+    result = {}
+    
+    if "stop_at_clip_layer" in inputs:
+        result["clip_skip"] = inputs["stop_at_clip_layer"]
+    
+    return result
+
+# =============================================================================
+# Node Mapper Definitions
+# =============================================================================
+
+# Define the mappers for ComfyUI core nodes not in main mapper
+NODE_MAPPERS_EXT = {
+    # KSamplers
+    "SamplerCustomAdvanced": {
+        "inputs_to_track": ["noise", "guider", "sampler", "sigmas", "latent_image"],
+        "transform_func": transform_sampler_custom_advanced
+    },
+    "KSampler": {
+        "inputs_to_track": [
+            "seed", "steps", "cfg", "sampler_name", "scheduler", 
+            "denoise", "positive", "negative", "latent_image",
+            "model", "clip_skip"
+        ],
+        "transform_func": transform_ksampler
+    },
+    # ComfyUI core nodes
+    "EmptyLatentImage": {
+        "inputs_to_track": ["width", "height", "batch_size"],
+        "transform_func": transform_empty_latent
+    },
+    "EmptySD3LatentImage": {
+        "inputs_to_track": ["width", "height", "batch_size"],
+        "transform_func": transform_empty_latent
+    },
+    "CLIPTextEncode": {
+        "inputs_to_track": ["text", "clip"],
+        "transform_func": transform_clip_text
+    },
+    "FluxGuidance": {
+        "inputs_to_track": ["guidance", "conditioning"],
+        "transform_func": transform_flux_guidance
+    },
+    "RandomNoise": {
+        "inputs_to_track": ["noise_seed"],
+        "transform_func": transform_random_noise
+    },
+    "KSamplerSelect": {
+        "inputs_to_track": ["sampler_name"],
+        "transform_func": transform_ksampler_select
+    },
+    "BasicScheduler": {
+        "inputs_to_track": ["scheduler", "steps", "denoise", "model"],
+        "transform_func": transform_basic_scheduler
+    },
+    "BasicGuider": {
+        "inputs_to_track": ["model", "conditioning"],
+        "transform_func": transform_basic_guider
+    },
+    "ModelSamplingFlux": {
+        "inputs_to_track": ["max_shift", "base_shift", "width", "height", "model"],
+        "transform_func": transform_model_sampling_flux
+    },
+    "UNETLoader": {
+        "inputs_to_track": ["unet_name"],
+        "transform_func": transform_unet_loader
+    },
+    "CheckpointLoaderSimple": {
+        "inputs_to_track": ["ckpt_name"],
+        "transform_func": transform_checkpoint_loader
+    },
+    "LatentUpscale": {
+        "inputs_to_track": ["width", "height"],
+        "transform_func": transform_empty_latent
+    },
+    "LatentUpscaleBy": {
+        "inputs_to_track": ["samples", "scale_by"],
+        "transform_func": transform_latent_upscale_by
+    },
+    "CLIPSetLastLayer": {
+        "inputs_to_track": ["clip", "stop_at_clip_layer"],
+        "transform_func": transform_clip_set_last_layer
+    }
+} 

py/workflow/ext/kjnodes.py

@@ -0,0 +1,74 @@
+"""
+KJNodes mappers extension for ComfyUI workflow parsing
+"""
+import logging
+import re
+from typing import Dict, Any
+
+logger = logging.getLogger(__name__)
+
+# =============================================================================
+# Transform Functions
+# =============================================================================
+
+def transform_join_strings(inputs: Dict) -> str:
+    """Transform function for JoinStrings nodes"""
+    string1 = inputs.get("string1", "")
+    string2 = inputs.get("string2", "")
+    delimiter = inputs.get("delimiter", "")
+    return f"{string1}{delimiter}{string2}"
+
+def transform_string_constant(inputs: Dict) -> str:
+    """Transform function for StringConstant nodes"""
+    return inputs.get("string", "")
+
+def transform_empty_latent_presets(inputs: Dict) -> Dict:
+    """Transform function for EmptyLatentImagePresets nodes"""
+    dimensions = inputs.get("dimensions", "")
+    invert = inputs.get("invert", False)
+    
+    # Extract width and height from dimensions string
+    # Expected format: "width x height (ratio)" or similar
+    width = 0
+    height = 0
+    
+    if dimensions:
+        # Try to extract dimensions using regex
+        match = re.search(r'(\d+)\s*x\s*(\d+)', dimensions)
+        if match:
+            width = int(match.group(1))
+            height = int(match.group(2))
+    
+    # If invert is True, swap width and height
+    if invert and width and height:
+        width, height = height, width
+    
+    return {"width": width, "height": height, "size": f"{width}x{height}"}
+
+def transform_int_constant(inputs: Dict) -> int:
+    """Transform function for INTConstant nodes"""
+    return inputs.get("value", 0)
+
+# =============================================================================
+# Node Mapper Definitions
+# =============================================================================
+
+# Define the mappers for KJNodes
+NODE_MAPPERS_EXT = {
+    "JoinStrings": {
+        "inputs_to_track": ["string1", "string2", "delimiter"],
+        "transform_func": transform_join_strings
+    },
+    "StringConstantMultiline": {
+        "inputs_to_track": ["string"],
+        "transform_func": transform_string_constant
+    },
+    "EmptyLatentImagePresets": {
+        "inputs_to_track": ["dimensions", "invert", "batch_size"],
+        "transform_func": transform_empty_latent_presets
+    },
+    "INTConstant": {
+        "inputs_to_track": ["value"],
+        "transform_func": transform_int_constant
+    }
+} 

py/workflow/main.py

@@ -0,0 +1,37 @@
+"""
+Main entry point for the workflow parser module
+"""
+import os
+import sys
+import logging
+from typing import Dict, Optional, Union
+
+# Add the parent directory to sys.path to enable imports
+SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
+ROOT_DIR = os.path.abspath(os.path.join(SCRIPT_DIR, '..', '..'))
+sys.path.insert(0, os.path.dirname(SCRIPT_DIR))
+
+from .parser import parse_workflow
+
+logger = logging.getLogger(__name__)
+
+def parse_comfyui_workflow(
+    workflow_path: str, 
+    output_path: Optional[str] = None
+) -> Dict:
+    """
+    Parse a ComfyUI workflow file and extract generation parameters
+    
+    Args:
+        workflow_path: Path to the workflow JSON file
+        output_path: Optional path to save the output JSON
+        
+    Returns:
+        Dictionary containing extracted parameters
+    """
+    return parse_workflow(workflow_path, output_path)
+
+if __name__ == "__main__":
+    # If run directly, use the CLI
+    from .cli import main
+    main() 

py/workflow/mappers.py

@@ -0,0 +1,282 @@
+"""
+Node mappers for ComfyUI workflow parsing
+"""
+import logging
+import os
+import importlib.util
+import inspect
+from typing import Dict, List, Any, Optional, Union, Type, Callable, Tuple
+
+logger = logging.getLogger(__name__)
+
+# Global mapper registry
+_MAPPER_REGISTRY: Dict[str, Dict] = {}
+
+# =============================================================================
+# Mapper Definition Functions
+# =============================================================================
+
+def create_mapper(
+    node_type: str,
+    inputs_to_track: List[str],
+    transform_func: Callable[[Dict], Any] = None
+) -> Dict:
+    """Create a mapper definition for a node type"""
+    mapper = {
+        "node_type": node_type,
+        "inputs_to_track": inputs_to_track,
+        "transform": transform_func or (lambda inputs: inputs)
+    }
+    return mapper
+
+def register_mapper(mapper: Dict) -> None:
+    """Register a node mapper in the global registry"""
+    _MAPPER_REGISTRY[mapper["node_type"]] = mapper
+    logger.debug(f"Registered mapper for node type: {mapper['node_type']}")
+
+def get_mapper(node_type: str) -> Optional[Dict]:
+    """Get a mapper for the specified node type"""
+    return _MAPPER_REGISTRY.get(node_type)
+
+def get_all_mappers() -> Dict[str, Dict]:
+    """Get all registered mappers"""
+    return _MAPPER_REGISTRY.copy()
+
+# =============================================================================
+# Node Processing Function
+# =============================================================================
+
+def process_node(node_id: str, node_data: Dict, workflow: Dict, parser: 'WorkflowParser') -> Any: # type: ignore
+    """Process a node using its mapper and extract relevant information"""
+    node_type = node_data.get("class_type")
+    mapper = get_mapper(node_type)
+    
+    if not mapper:
+        logger.warning(f"No mapper found for node type: {node_type}")
+        return None
+        
+    result = {}
+    
+    # Extract inputs based on the mapper's tracked inputs
+    for input_name in mapper["inputs_to_track"]:
+        if input_name in node_data.get("inputs", {}):
+            input_value = node_data["inputs"][input_name]
+            
+            # Check if input is a reference to another node's output
+            if isinstance(input_value, list) and len(input_value) == 2:
+                try:
+                    # Format is [node_id, output_slot]
+                    ref_node_id, output_slot = input_value
+                    # Convert node_id to string if it's an integer
+                    if isinstance(ref_node_id, int):
+                        ref_node_id = str(ref_node_id)
+                    
+                    # Recursively process the referenced node
+                    ref_value = parser.process_node(ref_node_id, workflow)
+                    
+                    if ref_value is not None:
+                        result[input_name] = ref_value
+                    else:
+                        # If we couldn't get a value from the reference, store the raw value
+                        result[input_name] = input_value
+                except Exception as e:
+                    logger.error(f"Error processing reference in node {node_id}, input {input_name}: {e}")
+                    result[input_name] = input_value
+            else:
+                # Direct value
+                result[input_name] = input_value
+    
+    # Apply the transform function
+    try:
+        return mapper["transform"](result)
+    except Exception as e:
+        logger.error(f"Error in transform function for node {node_id} of type {node_type}: {e}")
+        return result
+
+# =============================================================================
+# Transform Functions
+# =============================================================================
+
+
+
+def transform_lora_loader(inputs: Dict) -> Dict:
+    """Transform function for LoraLoader nodes"""
+    loras_data = inputs.get("loras", [])
+    lora_stack = inputs.get("lora_stack", {}).get("lora_stack", [])
+    
+    lora_texts = []
+    
+    # Process loras array
+    if isinstance(loras_data, dict) and "__value__" in loras_data:
+        loras_list = loras_data["__value__"]
+    elif isinstance(loras_data, list):
+        loras_list = loras_data
+    else:
+        loras_list = []
+        
+    # Process each active lora entry
+    for lora in loras_list:
+        if isinstance(lora, dict) and lora.get("active", False):
+            lora_name = lora.get("name", "")
+            strength = lora.get("strength", 1.0)
+            lora_texts.append(f"<lora:{lora_name}:{strength}>")
+    
+    # Process lora_stack if valid
+    if lora_stack and isinstance(lora_stack, list):
+        if not (len(lora_stack) == 2 and isinstance(lora_stack[0], (str, int)) and isinstance(lora_stack[1], int)):
+            for stack_entry in lora_stack:
+                lora_name = stack_entry[0]
+                strength = stack_entry[1]
+                lora_texts.append(f"<lora:{lora_name}:{strength}>")
+
+    result = {
+        "checkpoint": inputs.get("model", {}).get("checkpoint", ""),
+        "loras": " ".join(lora_texts)
+    }
+
+    if "clip" in inputs and isinstance(inputs["clip"], dict):
+        result["clip_skip"] = inputs["clip"].get("clip_skip", "-1")
+    
+    return result
+
+def transform_lora_stacker(inputs: Dict) -> Dict:
+    """Transform function for LoraStacker nodes"""
+    loras_data = inputs.get("loras", [])
+    result_stack = []
+    
+    # Handle existing stack entries
+    existing_stack = []
+    lora_stack_input = inputs.get("lora_stack", [])
+    
+    if isinstance(lora_stack_input, dict) and "lora_stack" in lora_stack_input:
+        existing_stack = lora_stack_input["lora_stack"]
+    elif isinstance(lora_stack_input, list):
+        if not (len(lora_stack_input) == 2 and isinstance(lora_stack_input[0], (str, int)) and 
+               isinstance(lora_stack_input[1], int)):
+            existing_stack = lora_stack_input
+    
+    # Add existing entries
+    if existing_stack:
+        result_stack.extend(existing_stack)
+    
+    # Process new loras
+    if isinstance(loras_data, dict) and "__value__" in loras_data:
+        loras_list = loras_data["__value__"]
+    elif isinstance(loras_data, list):
+        loras_list = loras_data
+    else:
+        loras_list = []
+        
+    for lora in loras_list:
+        if isinstance(lora, dict) and lora.get("active", False):
+            lora_name = lora.get("name", "")
+            strength = float(lora.get("strength", 1.0))
+            result_stack.append((lora_name, strength))
+    
+    return {"lora_stack": result_stack}
+
+def transform_trigger_word_toggle(inputs: Dict) -> str:
+    """Transform function for TriggerWordToggle nodes"""
+    toggle_data = inputs.get("toggle_trigger_words", [])
+
+    if isinstance(toggle_data, dict) and "__value__" in toggle_data:
+        toggle_words = toggle_data["__value__"]
+    elif isinstance(toggle_data, list):
+        toggle_words = toggle_data
+    else:
+        toggle_words = []
+    
+    # Filter active trigger words
+    active_words = []
+    for item in toggle_words:
+        if isinstance(item, dict) and item.get("active", False):
+            word = item.get("text", "")
+            if word and not word.startswith("__dummy"):
+                active_words.append(word)
+    
+    return ", ".join(active_words)
+
+# =============================================================================
+# Node Mapper Definitions
+# =============================================================================
+
+# Central definition of all supported node types and their configurations
+NODE_MAPPERS = {
+    
+    # LoraManager nodes
+    "Lora Loader (LoraManager)": {
+        "inputs_to_track": ["model", "clip", "loras", "lora_stack"],
+        "transform_func": transform_lora_loader
+    },
+    "Lora Stacker (LoraManager)": {
+        "inputs_to_track": ["loras", "lora_stack"],
+        "transform_func": transform_lora_stacker
+    },
+    "TriggerWord Toggle (LoraManager)": {
+        "inputs_to_track": ["toggle_trigger_words"],
+        "transform_func": transform_trigger_word_toggle
+    }   
+}
+
+def register_all_mappers() -> None:
+    """Register all mappers from the NODE_MAPPERS dictionary"""
+    for node_type, config in NODE_MAPPERS.items():
+        mapper = create_mapper(
+            node_type=node_type,
+            inputs_to_track=config["inputs_to_track"],
+            transform_func=config["transform_func"]
+        )
+        register_mapper(mapper)
+    logger.info(f"Registered {len(NODE_MAPPERS)} node mappers")
+
+# =============================================================================
+# Extension Loading
+# =============================================================================
+
+def load_extensions(ext_dir: str = None) -> None:
+    """
+    Load mapper extensions from the specified directory
+    
+    Extension files should define a NODE_MAPPERS_EXT dictionary containing mapper configurations.
+    These will be added to the global NODE_MAPPERS dictionary and registered automatically.
+    """
+    # Use default path if none provided
+    if ext_dir is None:
+        # Get the directory of this file
+        current_dir = os.path.dirname(os.path.abspath(__file__))
+        ext_dir = os.path.join(current_dir, 'ext')
+    
+    # Ensure the extension directory exists
+    if not os.path.exists(ext_dir):
+        os.makedirs(ext_dir, exist_ok=True)
+        logger.info(f"Created extension directory: {ext_dir}")
+        return
+    
+    # Load each Python file in the extension directory
+    for filename in os.listdir(ext_dir):
+        if filename.endswith('.py') and not filename.startswith('_'):
+            module_path = os.path.join(ext_dir, filename)
+            module_name = f"workflow.ext.{filename[:-3]}"  # Remove .py
+            
+            try:
+                # Load the module
+                spec = importlib.util.spec_from_file_location(module_name, module_path)
+                if spec and spec.loader:
+                    module = importlib.util.module_from_spec(spec)
+                    spec.loader.exec_module(module)
+                    
+                    # Check if the module defines NODE_MAPPERS_EXT
+                    if hasattr(module, 'NODE_MAPPERS_EXT'):
+                        # Add the extension mappers to the global NODE_MAPPERS dictionary
+                        NODE_MAPPERS.update(module.NODE_MAPPERS_EXT)
+                        logger.info(f"Added {len(module.NODE_MAPPERS_EXT)} mappers from extension: {filename}")
+                    else:
+                        logger.warning(f"Extension {filename} does not define NODE_MAPPERS_EXT dictionary")
+            except Exception as e:
+                logger.warning(f"Error loading extension {filename}: {e}")
+    
+    # Re-register all mappers after loading extensions
+    register_all_mappers()
+
+# Initialize the registry with default mappers
+# register_default_mappers() 

py/workflow/parser.py

@@ -0,0 +1,181 @@
+"""
+Main workflow parser implementation for ComfyUI
+"""
+import json
+import logging
+from typing import Dict, List, Any, Optional, Union, Set
+from .mappers import get_mapper, get_all_mappers, load_extensions, process_node
+from .utils import (
+    load_workflow, save_output, find_node_by_type,
+    trace_model_path
+)
+
+logger = logging.getLogger(__name__)
+
+class WorkflowParser:
+    """Parser for ComfyUI workflows"""
+    
+    def __init__(self):
+        """Initialize the parser with mappers"""
+        self.processed_nodes: Set[str] = set()  # Track processed nodes to avoid cycles
+        self.node_results_cache: Dict[str, Any] = {}  # Cache for processed node results
+        
+        # Load extensions
+        load_extensions()
+    
+    def process_node(self, node_id: str, workflow: Dict) -> Any:
+        """Process a single node and extract relevant information"""
+        # Return cached result if available
+        if node_id in self.node_results_cache:
+            return self.node_results_cache[node_id]
+            
+        # Check if we're in a cycle
+        if node_id in self.processed_nodes:
+            return None
+        
+        # Mark this node as being processed (to detect cycles)
+        self.processed_nodes.add(node_id)
+        
+        if node_id not in workflow:
+            self.processed_nodes.remove(node_id)
+            return None
+        
+        node_data = workflow[node_id]
+        node_type = node_data.get("class_type")
+        
+        result = None
+        if get_mapper(node_type):
+            try:
+                result = process_node(node_id, node_data, workflow, self)
+                # Cache the result
+                self.node_results_cache[node_id] = result
+            except Exception as e:
+                logger.error(f"Error processing node {node_id} of type {node_type}: {e}", exc_info=True)
+                # Return a partial result or None depending on how we want to handle errors
+                result = {}
+        
+        # Remove node from processed set to allow it to be processed again in a different context
+        self.processed_nodes.remove(node_id)
+        return result
+    
+    def find_primary_sampler_node(self, workflow: Dict) -> Optional[str]:
+        """
+        Find the primary sampler node in the workflow.
+        
+        Priority:
+        1. First try to find a SamplerCustomAdvanced node
+        2. If not found, look for KSampler nodes with denoise=1.0
+        3. If still not found, use the first KSampler node
+        
+        Args:
+            workflow: The workflow data as a dictionary
+            
+        Returns:
+            The node ID of the primary sampler node, or None if not found
+        """
+        # First check for SamplerCustomAdvanced nodes
+        sampler_advanced_nodes = []
+        ksampler_nodes = []
+        
+        # Scan workflow for sampler nodes
+        for node_id, node_data in workflow.items():
+            node_type = node_data.get("class_type")
+            
+            if node_type == "SamplerCustomAdvanced":
+                sampler_advanced_nodes.append(node_id)
+            elif node_type == "KSampler":
+                ksampler_nodes.append(node_id)
+        
+        # If we found SamplerCustomAdvanced nodes, return the first one
+        if sampler_advanced_nodes:
+            logger.debug(f"Found SamplerCustomAdvanced node: {sampler_advanced_nodes[0]}")
+            return sampler_advanced_nodes[0]
+        
+        # If we have KSampler nodes, look for one with denoise=1.0
+        if ksampler_nodes:
+            for node_id in ksampler_nodes:
+                node_data = workflow[node_id]
+                inputs = node_data.get("inputs", {})
+                denoise = inputs.get("denoise", 0)
+                
+                # Check if denoise is 1.0 (allowing for small floating point differences)
+                if abs(float(denoise) - 1.0) < 0.001:
+                    logger.debug(f"Found KSampler node with denoise=1.0: {node_id}")
+                    return node_id
+            
+            # If no KSampler with denoise=1.0 found, use the first one
+            logger.debug(f"No KSampler with denoise=1.0 found, using first KSampler: {ksampler_nodes[0]}")
+            return ksampler_nodes[0]
+        
+        # No sampler nodes found
+        logger.warning("No sampler nodes found in workflow")
+        return None
+    
+    def parse_workflow(self, workflow_data: Union[str, Dict], output_path: Optional[str] = None) -> Dict:
+        """
+        Parse the workflow and extract generation parameters
+        
+        Args:
+            workflow_data: The workflow data as a dictionary or a file path
+            output_path: Optional path to save the output JSON
+            
+        Returns:
+            Dictionary containing extracted parameters
+        """
+        # Load workflow from file if needed
+        if isinstance(workflow_data, str):
+            workflow = load_workflow(workflow_data)
+        else:
+            workflow = workflow_data
+            
+        # Reset the processed nodes tracker and cache
+        self.processed_nodes = set()
+        self.node_results_cache = {}
+        
+        # Find the primary sampler node
+        sampler_node_id = self.find_primary_sampler_node(workflow)
+        if not sampler_node_id:
+            logger.warning("No suitable sampler node found in workflow")
+            return {}
+        
+        # Process sampler node to extract parameters
+        sampler_result = self.process_node(sampler_node_id, workflow)
+        if not sampler_result:
+            return {}
+        
+        # Return the sampler result directly - it's already in the format we need
+        # This simplifies the structure and makes it easier to use in recipe_routes.py
+        
+        # Handle standard ComfyUI names vs our output format
+        if "cfg" in sampler_result:
+            sampler_result["cfg_scale"] = sampler_result.pop("cfg")
+            
+        # Add clip_skip = 1 to match reference output if not already present
+        if "clip_skip" not in sampler_result:
+            sampler_result["clip_skip"] = "1"
+        
+        # Ensure the prompt is a string and not a nested dictionary
+        if "prompt" in sampler_result and isinstance(sampler_result["prompt"], dict):
+            if "prompt" in sampler_result["prompt"]:
+                sampler_result["prompt"] = sampler_result["prompt"]["prompt"]
+        
+        # Save the result if requested
+        if output_path:
+            save_output(sampler_result, output_path)
+            
+        return sampler_result
+
+
+def parse_workflow(workflow_path: str, output_path: Optional[str] = None) -> Dict:
+    """
+    Parse a ComfyUI workflow file and extract generation parameters
+    
+    Args:
+        workflow_path: Path to the workflow JSON file
+        output_path: Optional path to save the output JSON
+        
+    Returns:
+        Dictionary containing extracted parameters
+    """
+    parser = WorkflowParser()
+    return parser.parse_workflow(workflow_path, output_path)

py/workflow/test.py

@@ -0,0 +1,63 @@
+"""
+Test script for the ComfyUI workflow parser
+"""
+import os
+import json
+import logging
+from .parser import parse_workflow
+
+logging.basicConfig(
+    level=logging.INFO,
+    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
+    handlers=[logging.StreamHandler()]
+)
+logger = logging.getLogger(__name__)
+
+# Configure paths
+SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
+ROOT_DIR = os.path.abspath(os.path.join(SCRIPT_DIR, '..', '..'))
+REFS_DIR = os.path.join(ROOT_DIR, 'refs')
+OUTPUT_DIR = os.path.join(ROOT_DIR, 'output')
+
+def test_parse_flux_workflow():
+    """Test parsing the flux example workflow"""
+    # Ensure output directory exists
+    os.makedirs(OUTPUT_DIR, exist_ok=True)
+    
+    # Define input and output paths
+    input_path = os.path.join(REFS_DIR, 'flux_prompt.json')
+    output_path = os.path.join(OUTPUT_DIR, 'parsed_flux_output.json')
+    
+    # Parse workflow
+    logger.info(f"Parsing workflow: {input_path}")
+    result = parse_workflow(input_path, output_path)
+    
+    # Print result summary
+    logger.info(f"Output saved to: {output_path}")
+    logger.info(f"Parsing completed. Result summary:")
+    logger.info(f"  LoRAs: {result.get('loras', '')}")
+    
+    gen_params = result.get('gen_params', {})
+    logger.info(f"  Prompt: {gen_params.get('prompt', '')[:50]}...")
+    logger.info(f"  Steps: {gen_params.get('steps', '')}")
+    logger.info(f"  Sampler: {gen_params.get('sampler', '')}")
+    logger.info(f"  Size: {gen_params.get('size', '')}")
+    
+    # Compare with reference output
+    ref_output_path = os.path.join(REFS_DIR, 'flux_output.json')
+    try:
+        with open(ref_output_path, 'r') as f:
+            ref_output = json.load(f)
+            
+        # Simple validation
+        loras_match = result.get('loras', '') == ref_output.get('loras', '')
+        prompt_match = gen_params.get('prompt', '') == ref_output.get('gen_params', {}).get('prompt', '')
+        
+        logger.info(f"Validation against reference:")
+        logger.info(f"  LoRAs match: {loras_match}")
+        logger.info(f"  Prompt match: {prompt_match}")
+    except Exception as e:
+        logger.warning(f"Failed to compare with reference output: {e}")
+
+if __name__ == "__main__":
+    test_parse_flux_workflow() 

py/workflow/utils.py

@@ -0,0 +1,120 @@
+"""
+Utility functions for ComfyUI workflow parsing
+"""
+import json
+import os
+import logging
+from typing import Dict, List, Any, Optional, Union, Set, Tuple
+
+logger = logging.getLogger(__name__)
+
+def load_workflow(workflow_path: str) -> Dict:
+    """Load a workflow from a JSON file"""
+    try:
+        with open(workflow_path, 'r', encoding='utf-8') as f:
+            return json.load(f)
+    except Exception as e:
+        logger.error(f"Error loading workflow from {workflow_path}: {e}")
+        raise
+
+def save_output(output: Dict, output_path: str) -> None:
+    """Save the parsed output to a JSON file"""
+    os.makedirs(os.path.dirname(os.path.abspath(output_path)), exist_ok=True)
+    try:
+        with open(output_path, 'w', encoding='utf-8') as f:
+            json.dump(output, f, indent=4)
+    except Exception as e:
+        logger.error(f"Error saving output to {output_path}: {e}")
+        raise
+
+def find_node_by_type(workflow: Dict, node_type: str) -> Optional[str]:
+    """Find a node of the specified type in the workflow"""
+    for node_id, node_data in workflow.items():
+        if node_data.get("class_type") == node_type:
+            return node_id
+    return None
+
+def find_nodes_by_type(workflow: Dict, node_type: str) -> List[str]:
+    """Find all nodes of the specified type in the workflow"""
+    return [node_id for node_id, node_data in workflow.items() 
+            if node_data.get("class_type") == node_type]
+
+def get_input_node_ids(workflow: Dict, node_id: str) -> Dict[str, Tuple[str, int]]:
+    """
+    Get the node IDs for all inputs of the given node
+    
+    Returns a dictionary mapping input names to (node_id, output_slot) tuples
+    """
+    result = {}
+    if node_id not in workflow:
+        return result
+        
+    node_data = workflow[node_id]
+    for input_name, input_value in node_data.get("inputs", {}).items():
+        # Check if this input is connected to another node
+        if isinstance(input_value, list) and len(input_value) == 2:
+            # Input is connected to another node's output
+            # Format: [node_id, output_slot]
+            ref_node_id, output_slot = input_value
+            result[input_name] = (str(ref_node_id), output_slot)
+            
+    return result
+
+def trace_model_path(workflow: Dict, start_node_id: str) -> List[str]:
+    """
+    Trace the model path backward from KSampler to find all LoRA nodes
+    
+    Args:
+        workflow: The workflow data
+        start_node_id: The starting node ID (usually KSampler)
+        
+    Returns:
+        List of node IDs in the model path
+    """
+    model_path_nodes = []
+    
+    # Get the model input from the start node
+    if start_node_id not in workflow:
+        return model_path_nodes
+    
+    # Track visited nodes to avoid cycles
+    visited = set()
+    
+    # Stack for depth-first search
+    stack = []
+    
+    # Get model input reference if available
+    start_node = workflow[start_node_id]
+    if "inputs" in start_node and "model" in start_node["inputs"] and isinstance(start_node["inputs"]["model"], list):
+        model_ref = start_node["inputs"]["model"]
+        stack.append(str(model_ref[0]))
+    
+    # Perform depth-first search
+    while stack:
+        node_id = stack.pop()
+        
+        # Skip if already visited
+        if node_id in visited:
+            continue
+        
+        # Mark as visited
+        visited.add(node_id)
+        
+        # Skip if node doesn't exist
+        if node_id not in workflow:
+            continue
+        
+        node = workflow[node_id]
+        node_type = node.get("class_type", "")
+        
+        # Add current node to result list if it's a LoRA node
+        if "Lora" in node_type:
+            model_path_nodes.append(node_id)
+        
+        # Add all input nodes that have a "model" or "lora_stack" output to the stack
+        if "inputs" in node:
+            for input_name, input_value in node["inputs"].items():
+                if input_name in ["model", "lora_stack"] and isinstance(input_value, list) and len(input_value) == 2:
+                    stack.append(str(input_value[0]))
+    
+    return model_path_nodes 

pyproject.toml

@@ -1,13 +1,18 @@
 [project]
 name = "comfyui-lora-manager"
 description = "LoRA Manager for ComfyUI - Access it at http://localhost:8188/loras for managing LoRA models with previews and metadata integration."
-version = "0.7.35-beta"
+version = "0.8.6"
 license = {file = "LICENSE"}
 dependencies = [
     "aiohttp",
     "jinja2", 
     "safetensors",
-    "watchdog"
+    "watchdog",
+    "beautifulsoup4",
+    "piexif",
+    "Pillow",
+    "olefile", # for getting rid of warning message
+    "requests"
 ]
 
 [project.urls]

refs/civitai_api_model_by_versionId.json

@@ -0,0 +1,100 @@
+{
+    "id": 1387174,
+    "modelId": 1231067,
+    "name": "v1.0",
+    "createdAt": "2025-02-08T11:15:47.197Z",
+    "updatedAt": "2025-02-08T11:29:04.526Z",
+    "status": "Published",
+    "publishedAt": "2025-02-08T11:29:04.487Z",
+    "trainedWords": [
+        "ppstorybook"
+    ],
+    "trainingStatus": null,
+    "trainingDetails": null,
+    "baseModel": "Flux.1 D",
+    "baseModelType": null,
+    "earlyAccessEndsAt": null,
+    "earlyAccessConfig": null,
+    "description": null,
+    "uploadType": "Created",
+    "usageControl": "Download",
+    "air": "urn:air:flux1:lora:civitai:1231067@1387174",
+    "stats": {
+        "downloadCount": 1436,
+        "ratingCount": 0,
+        "rating": 0,
+        "thumbsUpCount": 316
+    },
+    "model": {
+        "name": "Vivid Impressions Storybook Style",
+        "type": "LORA",
+        "nsfw": false,
+        "poi": false
+    },
+    "files": [
+        {
+            "id": 1289799,
+            "sizeKB": 18829.1484375,
+            "name": "pp-storybook_rank2_bf16.safetensors",
+            "type": "Model",
+            "pickleScanResult": "Success",
+            "pickleScanMessage": "No Pickle imports",
+            "virusScanResult": "Success",
+            "virusScanMessage": null,
+            "scannedAt": "2025-02-08T11:21:04.247Z",
+            "metadata": {
+                "format": "SafeTensor",
+                "size": null,
+                "fp": null
+            },
+            "hashes": {
+                "AutoV1": "F414C813",
+                "AutoV2": "9753338AB6",
+                "SHA256": "9753338AB693CA82BF89ED77A5D1912879E40051463EC6E330FB9866CE798668",
+                "CRC32": "A65AE7B3",
+                "BLAKE3": "A5F8AB95AC2486345E4ACCAE541FF19D97ED53EFB0A7CC9226636975A0437591",
+                "AutoV3": "34A22376739D"
+            },
+            "primary": true,
+            "downloadUrl": "https://civitai.com/api/download/models/1387174"
+        }
+    ],
+    "images": [
+        {
+            "url": "https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/42b875cf-c62b-41fa-a349-383b7f074351/width=832/56547310.jpeg",
+            "nsfwLevel": 1,
+            "width": 832,
+            "height": 1216,
+            "hash": "U5IiO6s-4Vn+0~EO^5xa00VsL#IU_O?E7yWC",
+            "type": "image",
+            "metadata": {
+                "hash": "U5IiO6s-4Vn+0~EO^5xa00VsL#IU_O?E7yWC",
+                "size": 1361590,
+                "width": 832,
+                "height": 1216
+            },
+            "meta": {
+                "Size": "832x1216",
+                "seed": 1116375220995209,
+                "Model": "flux_dev_fp8",
+                "steps": 23,
+                "hashes": {
+                    "model": ""
+                },
+                "prompt": "ppstorybook,A dreamy bunny hopping across a rainbow bridge, with fluffy clouds surrounding it and tiny birds flying alongside, rendered in a magical, soft-focus style with pastel hues and glowing accents.",
+                "Version": "ComfyUI",
+                "sampler": "DPM++ 2M",
+                "cfgScale": 3.5,
+                "clipSkip": 1,
+                "resources": [],
+                "Model hash": ""
+            },
+            "availability": "Public",
+            "hasMeta": true,
+            "hasPositivePrompt": true,
+            "onSite": false,
+            "remixOfId": null
+        }
+    ],
+    "downloadUrl": "https://civitai.com/api/download/models/1387174"
+}

refs/civitai_comfy_metadata.json

@@ -0,0 +1,153 @@
+{
+    "resource-stack": {
+      "class_type": "CheckpointLoaderSimple",
+      "inputs": { "ckpt_name": "urn:air:sdxl:checkpoint:civitai:827184@1410435" }
+    },
+    "resource-stack-1": {
+      "class_type": "LoraLoader",
+      "inputs": {
+        "lora_name": "urn:air:sdxl:lora:civitai:1107767@1253442",
+        "strength_model": 1,
+        "strength_clip": 1,
+        "model": ["resource-stack", 0],
+        "clip": ["resource-stack", 1]
+      }
+    },
+    "resource-stack-2": {
+      "class_type": "LoraLoader",
+      "inputs": {
+        "lora_name": "urn:air:sdxl:lora:civitai:1342708@1516344",
+        "strength_model": 1,
+        "strength_clip": 1,
+        "model": ["resource-stack-1", 0],
+        "clip": ["resource-stack-1", 1]
+      }
+    },
+    "resource-stack-3": {
+      "class_type": "LoraLoader",
+      "inputs": {
+        "lora_name": "urn:air:sdxl:lora:civitai:122359@135867",
+        "strength_model": 1.55,
+        "strength_clip": 1,
+        "model": ["resource-stack-2", 0],
+        "clip": ["resource-stack-2", 1]
+      }
+    },
+    "6": {
+      "class_type": "smZ CLIPTextEncode",
+      "inputs": {
+        "text": "masterpiece, best quality, amazing quality, detailed setting, detailed background, 1girl, yunyun (konosuba), nude, red eyes, hair ornament, braid, hair between eyes,low twintails, pink ribbon, bow, hair bow, pussy, frilled skirt, layered skirt, belt, pink thighhighs, (pussy juice), large insertion, vaginal tugging, pussy grip, detailed skin, detailed soles, stretched pussy, feet in stockings, ass, nipples, medium breasts, french kiss, anus, shocked, nervous, penis awe,  BREAK Professor\u0027s office, college student, pornographic, 1boy, close eyes, (musscular male, detailed large cock), vaginal sex, college office setting, ass grab, fucking, riding, cowgirl, erotic, side view, deep fucking",
+        "parser": "comfy",
+        "text_g": "",
+        "text_l": "",
+        "ascore": 2.5,
+        "width": 0,
+        "height": 0,
+        "crop_w": 0,
+        "crop_h": 0,
+        "target_width": 0,
+        "target_height": 0,
+        "smZ_steps": 1,
+        "mean_normalization": true,
+        "multi_conditioning": true,
+        "use_old_emphasis_implementation": false,
+        "with_SDXL": false,
+        "clip": ["resource-stack-3", 1]
+      },
+      "_meta": { "title": "Positive" }
+    },
+    "7": {
+      "class_type": "smZ CLIPTextEncode",
+      "inputs": {
+        "text": "bad quality,worst quality,worst detail,sketch,censor",
+        "parser": "comfy",
+        "text_g": "",
+        "text_l": "",
+        "ascore": 2.5,
+        "width": 0,
+        "height": 0,
+        "crop_w": 0,
+        "crop_h": 0,
+        "target_width": 0,
+        "target_height": 0,
+        "smZ_steps": 1,
+        "mean_normalization": true,
+        "multi_conditioning": true,
+        "use_old_emphasis_implementation": false,
+        "with_SDXL": false,
+        "clip": ["resource-stack-3", 1]
+      },
+      "_meta": { "title": "Negative" }
+    },
+    "20": {
+      "class_type": "UpscaleModelLoader",
+      "inputs": { "model_name": "urn:air:other:upscaler:civitai:147759@164821" },
+      "_meta": { "title": "Load Upscale Model" }
+    },
+    "17": {
+      "class_type": "LoadImage",
+      "inputs": {
+        "image": "https://orchestration.civitai.com/v2/consumer/blobs/5KZ6358TW8CNEGPZKD08NVDB30",
+        "upload": "image"
+      },
+      "_meta": { "title": "Image Load" }
+    },
+    "19": {
+      "class_type": "ImageUpscaleWithModel",
+      "inputs": { "upscale_model": ["20", 0], "image": ["17", 0] },
+      "_meta": { "title": "Upscale Image (using Model)" }
+    },
+    "23": {
+      "class_type": "ImageScale",
+      "inputs": {
+        "upscale_method": "nearest-exact",
+        "crop": "disabled",
+        "width": 1280,
+        "height": 1856,
+        "image": ["19", 0]
+      },
+      "_meta": { "title": "Upscale Image" }
+    },
+    "21": {
+      "class_type": "VAEEncode",
+      "inputs": { "pixels": ["23", 0], "vae": ["resource-stack", 2] },
+      "_meta": { "title": "VAE Encode" }
+    },
+    "11": {
+      "class_type": "KSampler",
+      "inputs": {
+        "sampler_name": "euler_ancestral",
+        "scheduler": "normal",
+        "seed": 2088370631,
+        "steps": 47,
+        "cfg": 6.5,
+        "denoise": 0.3,
+        "model": ["resource-stack-3", 0],
+        "positive": ["6", 0],
+        "negative": ["7", 0],
+        "latent_image": ["21", 0]
+      },
+      "_meta": { "title": "KSampler" }
+    },
+    "13": {
+      "class_type": "VAEDecode",
+      "inputs": { "samples": ["11", 0], "vae": ["resource-stack", 2] },
+      "_meta": { "title": "VAE Decode" }
+    },
+    "12": {
+      "class_type": "SaveImage",
+      "inputs": { "filename_prefix": "ComfyUI", "images": ["13", 0] },
+      "_meta": { "title": "Save Image" }
+    },
+    "extra": {
+      "airs": [
+        "urn:air:other:upscaler:civitai:147759@164821",
+        "urn:air:sdxl:checkpoint:civitai:827184@1410435",
+        "urn:air:sdxl:lora:civitai:1107767@1253442",
+        "urn:air:sdxl:lora:civitai:1342708@1516344",
+        "urn:air:sdxl:lora:civitai:122359@135867"
+      ]
+    },
+    "extraMetadata": "{\u0022prompt\u0022:\u0022masterpiece, best quality, amazing quality, detailed setting, detailed background, 1girl, yunyun (konosuba), nude, red eyes, hair ornament, braid, hair between eyes,low twintails, pink ribbon, bow, hair bow, pussy, frilled skirt, layered skirt, belt, pink thighhighs, (pussy juice), large insertion, vaginal tugging, pussy grip, detailed skin, detailed soles, stretched pussy, feet in stockings, ass, nipples, medium breasts, french kiss, anus, shocked, nervous, penis awe,  BREAK Professor\u0027s office, college student, pornographic, 1boy, close eyes, (musscular male, detailed large cock), vaginal sex, college office setting, ass grab, fucking, riding, cowgirl, erotic, side view, deep fucking\u0022,\u0022negativePrompt\u0022:\u0022bad quality,worst quality,worst detail,sketch,censor\u0022,\u0022steps\u0022:47,\u0022cfgScale\u0022:6.5,\u0022sampler\u0022:\u0022euler_ancestral\u0022,\u0022workflowId\u0022:\u0022img2img-hires\u0022,\u0022resources\u0022:[{\u0022modelVersionId\u0022:1410435,\u0022strength\u0022:1},{\u0022modelVersionId\u0022:1410435,\u0022strength\u0022:1},{\u0022modelVersionId\u0022:1253442,\u0022strength\u0022:1},{\u0022modelVersionId\u0022:1516344,\u0022strength\u0022:1},{\u0022modelVersionId\u0022:135867,\u0022strength\u0022:1.55}],\u0022remixOfId\u0022:32140259}"
+  }
+  

refs/flux_output.json

@@ -0,0 +1,15 @@
+{
+    "loras": "<lora:pp-enchanted-whimsy:0.9> <lora:ral-frctlgmtry_flux:1> <lora:pp-storybook_rank2_bf16:0.8>",
+    "gen_params": {
+        "prompt": "in the style of ppWhimsy, ral-frctlgmtry, ppstorybook,Stylized geek cat artist with glasses and a paintbrush, smiling at the viewer while holding a sign that reads 'Stay tuned!', solid white background",
+        "negative_prompt": "",
+        "steps": "25",
+        "sampler": "dpmpp_2m",
+        "scheduler": "beta",
+        "cfg": "1",
+        "seed": "48",
+        "guidance": 3.5,
+        "size": "896x1152",
+        "clip_skip": "2"
+    }
+}

refs/flux_prompt.json

@@ -0,0 +1,314 @@
+{
+    "6": {
+        "inputs": {
+            "text": [
+                "46",
+                0
+            ],
+            "clip": [
+                "58",
+                1
+            ]
+        },
+        "class_type": "CLIPTextEncode",
+        "_meta": {
+            "title": "CLIP Text Encode (Positive Prompt)"
+        }
+    },
+    "8": {
+        "inputs": {
+            "samples": [
+                "31",
+                0
+            ],
+            "vae": [
+                "39",
+                0
+            ]
+        },
+        "class_type": "VAEDecode",
+        "_meta": {
+            "title": "VAE Decode"
+        }
+    },
+    "27": {
+        "inputs": {
+            "width": 896,
+            "height": 1152,
+            "batch_size": 1
+        },
+        "class_type": "EmptySD3LatentImage",
+        "_meta": {
+            "title": "EmptySD3LatentImage"
+        }
+    },
+    "31": {
+        "inputs": {
+            "seed": 44,
+            "steps": 25,
+            "cfg": 1,
+            "sampler_name": "dpmpp_2m",
+            "scheduler": "beta",
+            "denoise": 1,
+            "model": [
+                "58",
+                0
+            ],
+            "positive": [
+                "35",
+                0
+            ],
+            "negative": [
+                "33",
+                0
+            ],
+            "latent_image": [
+                "27",
+                0
+            ]
+        },
+        "class_type": "KSampler",
+        "_meta": {
+            "title": "KSampler"
+        }
+    },
+    "33": {
+        "inputs": {
+            "text": "",
+            "clip": [
+                "58",
+                1
+            ]
+        },
+        "class_type": "CLIPTextEncode",
+        "_meta": {
+            "title": "CLIP Text Encode (Negative Prompt)"
+        }
+    },
+    "35": {
+        "inputs": {
+            "guidance": 3.5,
+            "conditioning": [
+                "6",
+                0
+            ]
+        },
+        "class_type": "FluxGuidance",
+        "_meta": {
+            "title": "FluxGuidance"
+        }
+    },
+    "37": {
+        "inputs": {
+            "unet_name": "flux\\flux1-dev-fp8-e4m3fn.safetensors",
+            "weight_dtype": "fp8_e4m3fn_fast"
+        },
+        "class_type": "UNETLoader",
+        "_meta": {
+            "title": "Load Diffusion Model"
+        }
+    },
+    "38": {
+        "inputs": {
+            "clip_name1": "t5xxl_fp8_e4m3fn.safetensors",
+            "clip_name2": "clip_l.safetensors",
+            "type": "flux",
+            "device": "default"
+        },
+        "class_type": "DualCLIPLoader",
+        "_meta": {
+            "title": "DualCLIPLoader"
+        }
+    },
+    "39": {
+        "inputs": {
+            "vae_name": "flux1\\ae.safetensors"
+        },
+        "class_type": "VAELoader",
+        "_meta": {
+            "title": "Load VAE"
+        }
+    },
+    "46": {
+        "inputs": {
+            "string1": [
+                "59",
+                0
+            ],
+            "string2": [
+                "51",
+                0
+            ],
+            "delimiter": ","
+        },
+        "class_type": "JoinStrings",
+        "_meta": {
+            "title": "Join Strings"
+        }
+    },
+    "50": {
+        "inputs": {
+            "images": [
+                "8",
+                0
+            ]
+        },
+        "class_type": "PreviewImage",
+        "_meta": {
+            "title": "Preview Image"
+        }
+    },
+    "51": {
+        "inputs": {
+            "string": "Stylized geek cat artist with glasses and a paintbrush, smiling at the viewer while holding a sign that reads 'Stay tuned!', solid white background",
+            "strip_newlines": true
+        },
+        "class_type": "StringConstantMultiline",
+        "_meta": {
+            "title": "positive"
+        }
+    },
+    "58": {
+        "inputs": {
+            "text": "<lora:pp-enchanted-whimsy:0.9><lora:ral-frctlgmtry_flux:1><lora:pp-storybook_rank2_bf16:0.8>",
+            "loras": [
+                {
+                    "name": "pp-enchanted-whimsy",
+                    "strength": "0.90",
+                    "active": false
+                },
+                {
+                    "name": "ral-frctlgmtry_flux",
+                    "strength": "0.85",
+                    "active": false
+                },
+                {
+                    "name": "pp-storybook_rank2_bf16",
+                    "strength": 0.8,
+                    "active": true
+                },
+                {
+                    "name": "__dummy_item1__",
+                    "strength": 0,
+                    "active": false,
+                    "_isDummy": true
+                },
+                {
+                    "name": "__dummy_item2__",
+                    "strength": 0,
+                    "active": false,
+                    "_isDummy": true
+                }
+            ],
+            "model": [
+                "37",
+                0
+            ],
+            "clip": [
+                "38",
+                0
+            ]
+        },
+        "class_type": "Lora Loader (LoraManager)",
+        "_meta": {
+            "title": "Lora Loader (LoraManager)"
+        }
+    },
+    "59": {
+        "inputs": {
+            "group_mode": "",
+            "toggle_trigger_words": [
+                {
+                    "text": "ppstorybook",
+                    "active": false
+                },
+                {
+                    "text": "__dummy_item__",
+                    "active": false,
+                    "_isDummy": true
+                },
+                {
+                    "text": "__dummy_item__",
+                    "active": false,
+                    "_isDummy": true
+                }
+            ],
+            "orinalMessage": "ppstorybook",
+            "trigger_words": [
+                "58",
+                2
+            ]
+        },
+        "class_type": "TriggerWord Toggle (LoraManager)",
+        "_meta": {
+            "title": "TriggerWord Toggle (LoraManager)"
+        }
+    },
+    "61": {
+        "inputs": {
+            "add_noise": "enable",
+            "noise_seed": 1111423448930884,
+            "steps": 20,
+            "cfg": 8,
+            "sampler_name": "euler",
+            "scheduler": "normal",
+            "start_at_step": 0,
+            "end_at_step": 10000,
+            "return_with_leftover_noise": "disable"
+        },
+        "class_type": "KSamplerAdvanced",
+        "_meta": {
+            "title": "KSampler (Advanced)"
+        }
+    },
+    "62": {
+        "inputs": {
+            "sigmas": [
+                "63",
+                0
+            ]
+        },
+        "class_type": "SamplerCustomAdvanced",
+        "_meta": {
+            "title": "SamplerCustomAdvanced"
+        }
+    },
+    "63": {
+        "inputs": {
+            "scheduler": "normal",
+            "steps": 20,
+            "denoise": 1
+        },
+        "class_type": "BasicScheduler",
+        "_meta": {
+            "title": "BasicScheduler"
+        }
+    },
+    "64": {
+        "inputs": {
+            "seed": 1089899258710474,
+            "steps": 20,
+            "cfg": 8,
+            "sampler_name": "euler",
+            "scheduler": "normal",
+            "denoise": 1
+        },
+        "class_type": "KSampler",
+        "_meta": {
+            "title": "KSampler"
+        }
+    },
+    "65": {
+        "inputs": {
+            "text": ",Stylized geek cat artist with glasses and a paintbrush, smiling at the viewer while holding a sign that reads 'Stay tuned!', solid white background",
+            "anything": [
+                "46",
+                0
+            ]
+        },
+        "class_type": "easy showAnything",
+        "_meta": {
+            "title": "Show Any"
+        }
+    }
+}

refs/jpeg_civitai_exif_userComment_example

@@ -0,0 +1,18 @@
+a dynamic and dramatic digital artwork featuring a stylized anthropomorphic white tiger with striking yellow eyes. The tiger is depicted in a powerful stance, wielding a katana with one hand raised above its head. Its fur is detailed with black stripes, and its mane flows wildly, blending with the stormy background. The scene is set amidst swirling dark clouds and flashes of lightning, enhancing the sense of movement and energy. The composition is vertical, with the tiger positioned centrally, creating a sense of depth and intensity. The color palette is dominated by shades of blue, gray, and white, with bright highlights from the lightning. The overall style is reminiscent of fantasy or manga art, with a focus on dynamic action and dramatic lighting.
+Negative prompt:
+Steps: 30, Sampler: Undefined, CFG scale: 3.5, Seed: 90300501, Size: 832x1216, Clip skip: 2, Created Date: 2025-03-05T13:51:18.1770234Z, Civitai resources: [{"type":"checkpoint","modelVersionId":691639,"modelName":"FLUX","modelVersionName":"Dev"},{"type":"lora","weight":0.4,"modelVersionId":1202162,"modelName":"Velvet\u0027s Mythic Fantasy Styles | Flux \u002B Pony \u002B illustrious","modelVersionName":"Flux Gothic Lines"},{"type":"lora","weight":0.8,"modelVersionId":1470588,"modelName":"Velvet\u0027s Mythic Fantasy Styles | Flux \u002B Pony \u002B illustrious","modelVersionName":"Flux Retro"},{"type":"lora","weight":0.75,"modelVersionId":746484,"modelName":"Elden Ring - Yoshitaka Amano","modelVersionName":"V1"},{"type":"lora","weight":0.2,"modelVersionId":914935,"modelName":"Ink-style","modelVersionName":"ink-dynamic"},{"type":"lora","weight":0.2,"modelVersionId":1189379,"modelName":"Painterly Fantasy by ChronoKnight - [FLUX \u0026 IL]","modelVersionName":"FLUX"},{"type":"lora","weight":0.2,"modelVersionId":757030,"modelName":"Mezzotint Artstyle for Flux - by Ethanar","modelVersionName":"V1"}], Civitai metadata: {}
+
+masterpiece, best quality, good quality, very aesthetic, absurdres, newest, 8K, depth of field, focused subject,
+dynamic angle, dutch angle, from below, epic half body portrait, gritty, wabi sabi, looking at viewer, woman is a geisha, parted lips,
+holographic skin, holofoil glitter, faint, glowing, ethereal, neon hair, glowing hair, otherworldly glow, she is dangerous
+<lora:ck-shadow-circuit-IL:0.78>, <lora:ck-nc-cyberpunk-IL-000011:0.4>, <lora:ck-neon-retrowave-IL:0.2>, <lora:ck-yoneyama-mai-IL-000014:0.4>
+Negative prompt: score_6, score_5, score_4, bad quality, worst quality, worst detail, sketch, censorship, furry, window, headphones,
+Steps: 30, Sampler: Euler a, Schedule type: Simple, CFG scale: 7, Seed: 1405717592, Size: 832x1216, Model hash: 1ad6ca7f70, Model: waiNSFWIllustrious_v100, Denoising strength: 0.35, Hires CFG Scale: 5, Hires upscale: 1.3, Hires steps: 20, Hires upscaler: 4x-AnimeSharp, Lora hashes: "ck-shadow-circuit-IL: 88e247aa8c3d, ck-nc-cyberpunk-IL-000011: 935e6755554c, ck-neon-retrowave-IL: edafb9df7da1, ck-yoneyama-mai-IL-000014: 1b9305692a2e", Version: f2.0.1v1.10.1-1.10.1, Diffusion in Low Bits: Automatic (fp16 LoRA)
+
+Masterpiece, best quality, high quality, newest, highres, 8K, HDR, absurdres, 1girl, solo, futuristic warrior, sleek exosuit with glowing energy cores, long braided hair flowing behind, gripping a high-tech bow with an energy arrow drawn, standing on a floating platform overlooking a massive space station, planets and nebulae in the distance, soft glow from distant stars, cinematic depth, foreshortening, dynamic pose, dramatic sci-fi lighting.
+Negative prompt: worst quality, normal quality, anatomical nonsense, bad anatomy,interlocked fingers, extra fingers,watermark,simple background, loli,
+Steps: 20, Sampler: euler_ancestral_karras, CFG scale: 8.0, Seed: 691121152183439, Model: il\waiNSFWIllustrious_v110.safetensors, Model hash: c3688ee04c, Lora_0 Model name: iLLMythAn1m3Style.safetensors, Lora_0 Model hash: ba7a040786, Lora_0 Strength model: 1.0, Lora_0 Strength clip: 1.0, Hashes: {"model": "c3688ee04c", "lora:iLLMythAn1m3Style": "ba7a040786"}
+
+Immerse yourself in the enchanting journey, where harmonious transmutation of Bauhaus art unites photographic precision and contemporary illustration, capturing an enthralling blend between vivid abstract nature and urban landscapes. Let your eyes be captivated by a kaleidoscope of rich, deep reds and yellows, entwined with intriguing shades that beckon a somber atmosphere. As your spirit ventures along this haunting path, witness the mysterious, high-angle perspective dominated by scattered clouds – granting you a mesmerizing glimpse into the ever-transforming realm of metamorphosing environments. ,<lora:flux/fav/ck-charcoal-drawing-000014.safetensors:1.0:1.0>   
+Negative prompt:
+Steps: 20, Sampler: Euler, CFG scale: 3.5, Seed: 885491426361006, Size: 832x1216, Model hash: 4610115bb0, Model: flux_dev, Hashes: {"LORA:flux/fav/ck-charcoal-drawing-000014.safetensors": "34d36c17c1", "model": "4610115bb0"}, Version: ComfyUI

refs/meta_format.txt

@@ -0,0 +1,3 @@
+In this ethereal masterpiece, metallic sculptures juxtapose effortlessly against a subtle backdrop of misty neutral hues. Exquisite curvatures and geometric shapes converge harmoniously, creating an illuminating realm of polished metallic surfaces. Shimmering copper, gleaming silver, and lustrous gold hues dance in perfect balance, highlighting the intricate play of light and shadow cast upon these celestial forms.  A halo of diffused radiance envelops each piece, enhancing their textured depths and metallic brilliance while allowing delicate details to emerge from obscurity. The composition conveys a serene yet mesmerizing atmosphere, as if suspended in a dreamlike limbo between reality and fantasy. The tantalizing interplay of colors within this transcendent realm creates a profound sense of depth and grandeur that invites the viewer into an enchanting voyage through abstract metallic beauty.  This captivating artwork evokes emotions of boundless curiosity and reverence reminiscent of the timeless works by artists such as Giorgio de Chirico or Paul Klee, while asserting a unique, modern artistic sensibility. With every observation, a new nuance unfolds, as if a never-ending story waiting to be discovered through the lens of metallic artistry.
+Negative prompt:
+Steps: 25, Sampler: dpmpp_2m_sgm_uniform, Seed: 471889513588087, Model: Fluxmania V5P.safetensors, Model hash: 8ae0583b06, VAE: ae.sft, VAE hash: afc8e28272, Lora_0 Model name: ArtVador I.safetensors, Lora_0 Model hash: 08f7133a58, Lora_0 Strength model: 0.65, Lora_0 Strength clip: 0.65, Lora_1 Model name: Kaoru Yamada.safetensors, Lora_1 Model hash: d4893f7202, Lora_1 Strength model: 0.75, Lora_1 Strength clip: 0.75, Hashes: {"model": "8ae0583b06", "vae": "afc8e28272", "lora:ArtVador I": "08f7133a58", "lora:Kaoru Yamada": "d4893f7202"}

refs/output.json

@@ -0,0 +1,11 @@
+{
+    "loras": "<lora:ck-neon-retrowave-IL-000012:0.8> <lora:aorunIllstrious:1> <lora:ck-shadow-circuit-IL-000012:0.78> <lora:MoriiMee_Gothic_Niji_Style_Illustrious_r1:0.45> <lora:ck-nc-cyberpunk-IL-000011:0.4>",
+    "prompt": "in the style of ck-rw, aorun, scales, makeup, bare shoulders, pointy ears, dress, claws, in the style of cksc, artist:moriimee, in the style of cknc, masterpiece, best quality, good quality, very aesthetic, absurdres, newest, 8K, depth of field, focused subject, close up, stylized, in gold and neon shades, wabi sabi, 1girl, rainbow angel wings, looking at viewer, dynamic angle, from below, from side, relaxing",
+    "negative_prompt": "bad quality, worst quality, worst detail, sketch ,signature, watermark, patreon logo, nsfw",
+    "steps": "20",
+    "sampler": "euler_ancestral",
+    "cfg_scale": "8",
+    "seed": "241",
+    "size": "832x1216",
+    "clip_skip": "2"
+}

refs/prompt.json

@@ -0,0 +1,401 @@
+{
+    "6": {
+        "inputs": {
+            "text": [
+                "301",
+                0
+            ],
+            "clip": [
+                "299",
+                1
+            ]
+        },
+        "class_type": "CLIPTextEncode",
+        "_meta": {
+            "title": "CLIP Text Encode (Prompt)"
+        }
+    },
+    "8": {
+        "inputs": {
+            "samples": [
+                "13",
+                1
+            ],
+            "vae": [
+                "10",
+                0
+            ]
+        },
+        "class_type": "VAEDecode",
+        "_meta": {
+            "title": "VAE Decode"
+        }
+    },
+    "10": {
+        "inputs": {
+            "vae_name": "flux1\\ae.safetensors"
+        },
+        "class_type": "VAELoader",
+        "_meta": {
+            "title": "Load VAE"
+        }
+    },
+    "11": {
+        "inputs": {
+            "clip_name1": "t5xxl_fp8_e4m3fn.safetensors",
+            "clip_name2": "ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors",
+            "type": "flux",
+            "device": "default"
+        },
+        "class_type": "DualCLIPLoader",
+        "_meta": {
+            "title": "DualCLIPLoader"
+        }
+    },
+    "13": {
+        "inputs": {
+            "noise": [
+                "147",
+                0
+            ],
+            "guider": [
+                "22",
+                0
+            ],
+            "sampler": [
+                "16",
+                0
+            ],
+            "sigmas": [
+                "17",
+                0
+            ],
+            "latent_image": [
+                "48",
+                0
+            ]
+        },
+        "class_type": "SamplerCustomAdvanced",
+        "_meta": {
+            "title": "SamplerCustomAdvanced"
+        }
+    },
+    "16": {
+        "inputs": {
+            "sampler_name": "dpmpp_2m"
+        },
+        "class_type": "KSamplerSelect",
+        "_meta": {
+            "title": "KSamplerSelect"
+        }
+    },
+    "17": {
+        "inputs": {
+            "scheduler": "beta",
+            "steps": [
+                "246",
+                0
+            ],
+            "denoise": 1,
+            "model": [
+                "28",
+                0
+            ]
+        },
+        "class_type": "BasicScheduler",
+        "_meta": {
+            "title": "BasicScheduler"
+        }
+    },
+    "22": {
+        "inputs": {
+            "model": [
+                "28",
+                0
+            ],
+            "conditioning": [
+                "29",
+                0
+            ]
+        },
+        "class_type": "BasicGuider",
+        "_meta": {
+            "title": "BasicGuider"
+        }
+    },
+    "28": {
+        "inputs": {
+            "max_shift": 1.1500000000000001,
+            "base_shift": 0.5,
+            "width": [
+                "48",
+                1
+            ],
+            "height": [
+                "48",
+                2
+            ],
+            "model": [
+                "299",
+                0
+            ]
+        },
+        "class_type": "ModelSamplingFlux",
+        "_meta": {
+            "title": "ModelSamplingFlux"
+        }
+    },
+    "29": {
+        "inputs": {
+            "guidance": 3.5,
+            "conditioning": [
+                "6",
+                0
+            ]
+        },
+        "class_type": "FluxGuidance",
+        "_meta": {
+            "title": "FluxGuidance"
+        }
+    },
+    "48": {
+        "inputs": {
+            "resolution": "832x1216 (0.68)",
+            "batch_size": 1,
+            "width_override": 0,
+            "height_override": 0
+        },
+        "class_type": "SDXLEmptyLatentSizePicker+",
+        "_meta": {
+            "title": "🔧 SDXL Empty Latent Size Picker"
+        }
+    },
+    "65": {
+        "inputs": {
+            "unet_name": "flux\\flux1-dev-fp8-e4m3fn.safetensors",
+            "weight_dtype": "fp8_e4m3fn_fast"
+        },
+        "class_type": "UNETLoader",
+        "_meta": {
+            "title": "Load Diffusion Model"
+        }
+    },
+    "147": {
+        "inputs": {
+            "noise_seed": 651532572596956
+        },
+        "class_type": "RandomNoise",
+        "_meta": {
+            "title": "RandomNoise"
+        }
+    },
+    "148": {
+        "inputs": {
+            "wildcard_text": "__some-prompts__",
+            "populated_text": "A surreal digital artwork showcases a forward-thinking inventor captivated by his intricate mechanical creation through a large magnifying glass. Viewed from an unconventional perspective, the scene reveals an eccentric assembly of gears, springs, and brass instruments within his workshop. Soft, ethereal light radiates from the invention, casting enigmatic shadows on the walls as time appears to bend around its metallic form, invoking a sense of curiosity, wonder, and exhilaration in discovery.",
+            "mode": "fixed",
+            "seed": 553084268162351,
+            "Select to add Wildcard": "Select the Wildcard to add to the text"
+        },
+        "class_type": "ImpactWildcardProcessor",
+        "_meta": {
+            "title": "ImpactWildcardProcessor"
+        }
+    },
+    "151": {
+        "inputs": {
+            "text": "A hyper-realistic close-up portrait of a young woman with shoulder-length black hair styled in edgy, futuristic layers, adorned with glowing tips. She wears mecha eyewear with a neon green visor that transitions into iridescent shades of teal and gold. The frame is sleek, with angular edges and fine mechanical detailing. Her expression is fierce and confident, with flawless skin highlighted by the neon reflections. She wears a high-tech bodysuit with integrated LED lines and metallic panels. The background depicts a hazy rendition of The Great Wave off Kanagawa by Hokusai, its powerful waves blending seamlessly with the neon tones, amplifying her intense, defiant aura."
+        },
+        "class_type": "Text Multiline",
+        "_meta": {
+            "title": "Text Multiline"
+        }
+    },
+    "191": {
+        "inputs": {
+            "text": "A cinematic, oil painting masterpiece captures the essence of impressionistic surrealism, inspired by Claude Monet. A mysterious woman in a flowing crimson dress stands at the edge of a tranquil lake, where lily pads shimmer under an ethereal, golden twilight. The water’s surface reflects a dreamlike sky, its swirling hues of violet and sapphire melting together like liquid light. The thick, expressive brushstrokes lend depth to the scene, evoking a sense of nostalgia and quiet longing, as if the world itself is caught between reality and a fleeting dream.  \nA mesmerizing oil painting masterpiece inspired by Salvador Dalí, blending surrealism with post-impressionist texture. A lone violinist plays atop a melting clock tower, his form distorted by the passage of time. The sky is a cascade of swirling, liquid oranges and deep blues, where floating staircases spiral endlessly into the horizon. The impasto technique gives depth and movement to the surreal elements, making time itself feel fluid, as if the world is dissolving into a dream.  \nA stunning impressionistic oil painting evokes the spirit of Edvard Munch, capturing a solitary figure standing on a rain-soaked street, illuminated by the glow of flickering gas lamps. The swirling, chaotic strokes of deep blues and fiery reds reflect the turbulence of emotion, while the blurred reflections in the wet cobblestone suggest a merging of past and present. The faceless figure, draped in a dark overcoat, seems lost in thought, embodying the ephemeral nature of memory and time.  \nA breathtaking oil painting masterpiece, inspired by Gustav Klimt, presents a celestial ballroom where faceless dancers swirl in an eternal waltz beneath a gilded, star-speckled sky. Their golden garments shimmer with intricate patterns, blending into the opulent mosaic floor that seems to stretch into infinity. The dreamlike composition, rich in warm amber and deep sapphire hues, captures an otherworldly elegance, as if the dancers are suspended in a moment that transcends time.  \nA visionary oil painting inspired by Marc Chagall depicts a dreamlike cityscape where gravity ceases to exist. A couple floats above a crimson-tinted town, their forms dissolving into the swirling strokes of a vast, cerulean sky. The buildings below twist and bend in rhythmic motion, their windows glowing like tiny stars. The thick, textured brushwork conveys a sense of weightlessness and wonder, as if love itself has defied the laws of the universe.  \nAn impressionistic oil painting in the style of J.M.W. Turner, depicting a ghostly ship sailing through a sea of swirling golden mist. The waves crash and dissolve into abstract, fiery strokes of orange and deep indigo, blurring the line between ocean and sky. The ship appears almost ethereal, as if drifting between worlds, lost in the ever-changing tides of memory and myth. The dynamic brushstrokes capture the relentless power of nature and the fleeting essence of time.  \nA captivating oil painting masterpiece, infused with surrealist impressionism, portrays a grand library where books float midair, their pages unraveling into ribbons of light. The towering shelves twist into the heavens, vanishing into an infinite, starry void. A lone scholar, illuminated by the glow of a suspended lantern, reaches for a book that seems to pulse with life. The scene pulses with mystery, where the impasto textures bring depth to the interplay between knowledge and dreams.  \nA luminous impressionistic oil painting captures the melancholic beauty of an abandoned carnival, its faded carousel horses frozen mid-gallop beneath a sky of swirling lavender and gold. The wind carries fragments of forgotten laughter through the empty fairground, where scattered ticket stubs and crumbling banners whisper tales of joy long past. The thick, textured brushstrokes blend nostalgia with an eerie dreamlike quality, as if the carnival exists only in the echoes of memory.  \nA surreal oil painting in the spirit of René Magritte, featuring a towering lighthouse that emits not light, but cascading waterfalls from its peak. The swirling sky, painted in deep midnight blues, is punctuated by glowing, crescent moons that defy gravity. A lone figure stands at the water’s edge, gazing up in quiet contemplation, as if caught between wonder and the unknown. The painting’s rich textures and luminous colors create an enigmatic, dreamlike landscape.  \nA striking impressionistic oil painting, reminiscent of Van Gogh, portrays a lone traveler on a winding cobblestone path, their silhouette bathed in the golden glow of lantern-lit cherry blossoms. The petals swirl through the night air like glowing embers, blending with the deep, rhythmic strokes of a star-filled indigo sky. The scene captures a feeling of wistful solitude, as if the traveler is walking not only through the city, but through the fleeting nature of time itself."
+        },
+        "class_type": "Text Multiline",
+        "_meta": {
+            "title": "Text Multiline"
+        }
+    },
+    "203": {
+        "inputs": {
+            "string1": [
+                "289",
+                0
+            ],
+            "string2": [
+                "293",
+                0
+            ],
+            "delimiter": ", "
+        },
+        "class_type": "JoinStrings",
+        "_meta": {
+            "title": "Join Strings"
+        }
+    },
+    "208": {
+        "inputs": {
+            "file_path": "",
+            "dictionary_name": "[filename]",
+            "label": "TextBatch",
+            "mode": "automatic",
+            "index": 0,
+            "multiline_text": [
+                "191",
+                0
+            ]
+        },
+        "class_type": "Text Load Line From File",
+        "_meta": {
+            "title": "Text Load Line From File"
+        }
+    },
+    "226": {
+        "inputs": {
+            "images": [
+                "8",
+                0
+            ]
+        },
+        "class_type": "PreviewImage",
+        "_meta": {
+            "title": "Preview Image"
+        }
+    },
+    "246": {
+        "inputs": {
+            "value": 25
+        },
+        "class_type": "INTConstant",
+        "_meta": {
+            "title": "Steps"
+        }
+    },
+    "289": {
+        "inputs": {
+            "group_mode": true,
+            "toggle_trigger_words": [
+                {
+                    "text": "bo-exposure",
+                    "active": true
+                },
+                {
+                    "text": "__dummy_item__",
+                    "active": false,
+                    "_isDummy": true
+                },
+                {
+                    "text": "__dummy_item__",
+                    "active": false,
+                    "_isDummy": true
+                }
+            ],
+            "orinalMessage": "bo-exposure",
+            "trigger_words": [
+                "299",
+                2
+            ]
+        },
+        "class_type": "TriggerWord Toggle (LoraManager)",
+        "_meta": {
+            "title": "TriggerWord Toggle (LoraManager)"
+        }
+    },
+    "293": {
+        "inputs": {
+            "input": 1,
+            "text1": [
+                "208",
+                0
+            ],
+            "text2": [
+                "151",
+                0
+            ]
+        },
+        "class_type": "easy textSwitch",
+        "_meta": {
+            "title": "Text Switch"
+        }
+    },
+    "297": {
+        "inputs": {
+            "text": ""
+        },
+        "class_type": "Lora Stacker (LoraManager)",
+        "_meta": {
+            "title": "Lora Stacker (LoraManager)"
+        }
+    },
+    "298": {
+        "inputs": {
+            "anything": [
+                "297",
+                0
+            ]
+        },
+        "class_type": "easy showAnything",
+        "_meta": {
+            "title": "Show Any"
+        }
+    },
+    "299": {
+        "inputs": {
+            "text": "<lora:boFLUX Double Exposure Magic v2:0.8> <lora:FluxDFaeTasticDetails:0.65>",
+            "loras": [
+                {
+                    "name": "boFLUX Double Exposure Magic v2",
+                    "strength": 0.8,
+                    "active": true
+                },
+                {
+                    "name": "FluxDFaeTasticDetails",
+                    "strength": 0.65,
+                    "active": true
+                },
+                {
+                    "name": "__dummy_item1__",
+                    "strength": 0,
+                    "active": false,
+                    "_isDummy": true
+                },
+                {
+                    "name": "__dummy_item2__",
+                    "strength": 0,
+                    "active": false,
+                    "_isDummy": true
+                }
+            ],
+            "model": [
+                "65",
+                0
+            ],
+            "clip": [
+                "11",
+                0
+            ],
+            "lora_stack": [
+                "297",
+                0
+            ]
+        },
+        "class_type": "Lora Loader (LoraManager)",
+        "_meta": {
+            "title": "Lora Loader (LoraManager)"
+        }
+    },
+    "301": {
+        "inputs": {
+            "string": "A hyper-realistic close-up portrait of a young woman with shoulder-length black hair styled in edgy, futuristic layers, adorned with glowing tips. She wears mecha eyewear with a neon green visor that transitions into iridescent shades of teal and gold. The frame is sleek, with angular edges and fine mechanical detailing. Her expression is fierce and confident, with flawless skin highlighted by the neon reflections. She wears a high-tech bodysuit with integrated LED lines and metallic panels. The background depicts a hazy rendition of The Great Wave off Kanagawa by Hokusai, its powerful waves blending seamlessly with the neon tones, amplifying her intense, defiant aura.",
+            "strip_newlines": true
+        },
+        "class_type": "StringConstantMultiline",
+        "_meta": {
+            "title": "String Constant Multiline"
+        }
+    }
+}

refs/recipe.json

@@ -0,0 +1,82 @@
+{
+    "id": "0448c06d-de1b-46ab-975c-c5aa60d90dbc",
+    "file_path": "D:/Workspace/ComfyUI/models/loras/recipes/0448c06d-de1b-46ab-975c-c5aa60d90dbc.jpg",
+    "title": "a mysterious, steampunk-inspired character standing in a dramatic pose",
+    "modified": 1741837612.3931093,
+    "created_date": 1741492786.5581934,
+    "base_model": "Flux.1 D",
+    "loras": [
+        {
+            "file_name": "ChronoDivinitiesFlux_r1",
+            "hash": "ddbc5abd00db46ad464f5e3ca85f8f7121bc14b594d6785f441d9b002fffe66a",
+            "strength": 0.8,
+            "modelVersionId": 1438879,
+            "modelName": "Chrono Divinities - By HailoKnight",
+            "modelVersionName": "Flux"
+        },
+        {
+            "file_name": "flux.1_lora_flyway_ink-dynamic",
+            "hash": "4b4f3b469a0d5d3a04a46886abfa33daa37a905db070ccfbd10b345c6fb00eff",
+            "strength": 0.2,
+            "modelVersionId": 914935,
+            "modelName": "Ink-style",
+            "modelVersionName": "ink-dynamic"
+        },
+        {
+            "file_name": "ck-painterly-fantasy-000017",
+            "hash": "48c67064e2936aec342580a2a729d91d75eb818e45ecf993b9650cc66c94c420",
+            "strength": 0.2,
+            "modelVersionId": 1189379,
+            "modelName": "Painterly Fantasy by ChronoKnight - [FLUX & IL]",
+            "modelVersionName": "FLUX"
+        },
+        {
+            "file_name": "RetroAnimeFluxV1",
+            "hash": "8f43c31b6c3238ac44195c970d511d759c5893bddd00f59f42b8fe51e8e76fa0",
+            "strength": 0.8,
+            "modelVersionId": 806265,
+            "modelName": "Retro Anime Flux - Style",
+            "modelVersionName": "v1.0"
+        },
+        {
+            "file_name": "Mezzotint_Artstyle_for_Flux_-_by_Ethanar",
+            "hash": "e6961502769123bf23a66c5c5298d76264fd6b9610f018319a0ccb091bfc308e",
+            "strength": 0.2,
+            "modelVersionId": 757030,
+            "modelName": "Mezzotint Artstyle for Flux - by Ethanar",
+            "modelVersionName": "V1"
+        },
+        {
+            "file_name": "FluxMythG0thicL1nes",
+            "hash": "ecb03595de62bd6183a0dd2b38bea35669fd4d509f4bbae5aa0572cfb7ef4279",
+            "strength": 0.4,
+            "modelVersionId": 1202162,
+            "modelName": "Velvet's Mythic Fantasy Styles | Flux + Pony + illustrious",
+            "modelVersionName": "Flux Gothic Lines"
+        },
+        {
+            "file_name": "Elden_Ring_-_Yoshitaka_Amano",
+            "hash": "c660c4c55320be7206cb6a917c59d8da3953cc07169fe10bda833a54ec0024f9",
+            "strength": 0.75,
+            "modelVersionId": 746484,
+            "modelName": "Elden Ring - Yoshitaka Amano",
+            "modelVersionName": "V1"
+        }
+    ],
+    "gen_params": {
+        "prompt": "a mysterious, steampunk-inspired character standing in a dramatic pose. The character is dressed in a long, intricately detailed dark coat with ornate patterns, a wide-brimmed hat, and leather boots. The face is partially obscured by the hat's shadow, adding to the enigmatic aura. The background showcases a large, antique clock with Roman numerals, surrounded by dynamic lightning and ethereal white birds, enhancing the fantastical atmosphere. The color palette is dominated by dark tones with striking contrasts of white and blue lightning, creating a sense of tension and energy. The overall composition is vertical, with the character centrally positioned, exuding a sense of power and mystery. hkchrono",
+        "negative_prompt": "",
+        "checkpoint": {
+            "type": "checkpoint",
+            "modelVersionId": 691639,
+            "modelName": "FLUX",
+            "modelVersionName": "Dev"
+        },
+        "steps": "30",
+        "sampler": "Undefined",
+        "cfg_scale": "3.5",
+        "seed": "1472903449",
+        "size": "832x1216",
+        "clip_skip": "2"
+    }
+}

refs/test_output.txt

@@ -0,0 +1,294 @@
+Loading workflow from D:\Workspace\ComfyUI\custom_nodes\ComfyUI-Lora-Manager\refs\prompt.json
+Expected output from D:\Workspace\ComfyUI\custom_nodes\ComfyUI-Lora-Manager\refs\output.json
+
+Expected output:
+{
+  "loras": "<lora:ck-neon-retrowave-IL-000012:0.8> <lora:aorunIllstrious:1> <lora:ck-shadow-circuit-IL-000012:0.78> <lora:MoriiMee_Gothic_Niji_Style_Illustrious_r1:0.45> <lora:ck-nc-cyberpunk-IL-000011:0.4>",
+  "gen_params": {
+    "prompt": "in the style of ck-rw, aorun, scales, makeup, bare shoulders, pointy ears, dress, claws, in the style of cksc, artist:moriimee, in the style of cknc, masterpiece, best quality, good quality, very aesthetic, absurdres, newest, 8K, depth of field, focused subject, close up, stylized, in gold and neon shades, wabi sabi, 1girl, rainbow angel wings, looking at viewer, dynamic angle, from below, from side, relaxing",
+    "negative_prompt": "bad quality, worst quality, worst detail, sketch ,signature, watermark, patreon logo, nsfw",
+    "steps": "20",
+    "sampler": "euler_ancestral",
+    "cfg_scale": "8",
+    "seed": "241",
+    "size": "832x1216",
+    "clip_skip": "2"
+  }
+}
+
+Sampler node:
+{
+  "inputs": {
+    "seed": 241,
+    "steps": 20,
+    "cfg": 8,
+    "sampler_name": "euler_ancestral",
+    "scheduler": "karras",
+    "denoise": 1,
+    "model": [
+      "56",
+      0
+    ],
+    "positive": [
+      "6",
+      0
+    ],
+    "negative": [
+      "7",
+      0
+    ],
+    "latent_image": [
+      "5",
+      0
+    ]
+  },
+  "class_type": "KSampler",
+  "_meta": {
+    "title": "KSampler"
+  }
+}
+
+Extracted parameters:
+seed: 241
+steps: 20
+cfg_scale: 8
+
+Positive node (6):
+{
+  "inputs": {
+    "text": [
+      "22",
+      0
+    ],
+    "clip": [
+      "56",
+      1
+    ]
+  },
+  "class_type": "CLIPTextEncode",
+  "_meta": {
+    "title": "CLIP Text Encode (Prompt)"
+  }
+}
+
+Text node (22):
+{
+  "inputs": {
+    "string1": [
+      "55",
+      0
+    ],
+    "string2": [
+      "21",
+      0
+    ],
+    "delimiter": ", "
+  },
+  "class_type": "JoinStrings",
+  "_meta": {
+    "title": "Join Strings"
+  }
+}
+
+String1 node (55):
+{
+  "inputs": {
+    "group_mode": true,
+    "toggle_trigger_words": [
+      {
+        "text": "in the style of ck-rw",
+        "active": true
+      },
+      {
+        "text": "aorun, scales, makeup, bare shoulders, pointy ears",
+        "active": true
+      },
+      {
+        "text": "dress",
+        "active": true
+      },
+      {
+        "text": "claws",
+        "active": true
+      },
+      {
+        "text": "in the style of cksc",
+        "active": true
+      },
+      {
+        "text": "artist:moriimee",
+        "active": true
+      },
+      {
+        "text": "in the style of cknc",
+        "active": true
+      },
+      {
+        "text": "__dummy_item__",
+        "active": false,
+        "_isDummy": true
+      },
+      {
+        "text": "__dummy_item__",
+        "active": false,
+        "_isDummy": true
+      }
+    ],
+    "orinalMessage": "in the style of ck-rw,, aorun, scales, makeup, bare shoulders, pointy ears,, dress,, claws,, in the style of cksc,, artist:moriimee,, in the style of cknc",
+    "trigger_words": [
+      "56",
+      2
+    ]
+  },
+  "class_type": "TriggerWord Toggle (LoraManager)",
+  "_meta": {
+    "title": "TriggerWord Toggle (LoraManager)"
+  }
+}
+
+String2 node (21):
+{
+  "inputs": {
+    "string": "masterpiece, best quality, good quality, very aesthetic, absurdres, newest, 8K, depth of field, focused subject, close up, stylized, in gold and neon shades, wabi sabi, 1girl, rainbow angel wings, looking at viewer, dynamic angle, from below, from side, relaxing",
+    "strip_newlines": false
+  },
+  "class_type": "StringConstantMultiline",
+  "_meta": {
+    "title": "positive"
+  }
+}
+
+Negative node (7):
+{
+  "inputs": {
+    "text": "bad quality, worst quality, worst detail, sketch ,signature, watermark, patreon logo, nsfw",
+    "clip": [
+      "56",
+      1
+    ]
+  },
+  "class_type": "CLIPTextEncode",
+  "_meta": {
+    "title": "CLIP Text Encode (Prompt)"
+  }
+}
+
+LoRA nodes (3):
+
+LoRA node 56:
+{
+  "inputs": {
+    "text": "<lora:ck-shadow-circuit-IL-000012:0.78> <lora:MoriiMee_Gothic_Niji_Style_Illustrious_r1:0.45> <lora:ck-nc-cyberpunk-IL-000011:0.4>",
+    "loras": [
+      {
+        "name": "ck-shadow-circuit-IL-000012",
+        "strength": 0.78,
+        "active": true
+      },
+      {
+        "name": "MoriiMee_Gothic_Niji_Style_Illustrious_r1",
+        "strength": 0.45,
+        "active": true
+      },
+      {
+        "name": "ck-nc-cyberpunk-IL-000011",
+        "strength": 0.4,
+        "active": true
+      },
+      {
+        "name": "__dummy_item1__",
+        "strength": 0,
+        "active": false,
+        "_isDummy": true
+      },
+      {
+        "name": "__dummy_item2__",
+        "strength": 0,
+        "active": false,
+        "_isDummy": true
+      }
+    ],
+    "model": [
+      "4",
+      0
+    ],
+    "clip": [
+      "4",
+      1
+    ],
+    "lora_stack": [
+      "57",
+      0
+    ]
+  },
+  "class_type": "Lora Loader (LoraManager)",
+  "_meta": {
+    "title": "Lora Loader (LoraManager)"
+  }
+}
+
+LoRA node 57:
+{
+  "inputs": {
+    "text": "<lora:aorunIllstrious:1>",
+    "loras": [
+      {
+        "name": "aorunIllstrious",
+        "strength": "0.90",
+        "active": true
+      },
+      {
+        "name": "__dummy_item1__",
+        "strength": 0,
+        "active": false,
+        "_isDummy": true
+      },
+      {
+        "name": "__dummy_item2__",
+        "strength": 0,
+        "active": false,
+        "_isDummy": true
+      }
+    ],
+    "lora_stack": [
+      "59",
+      0
+    ]
+  },
+  "class_type": "Lora Stacker (LoraManager)",
+  "_meta": {
+    "title": "Lora Stacker (LoraManager)"
+  }
+}
+
+LoRA node 59:
+{
+  "inputs": {
+    "text": "<lora:ck-neon-retrowave-IL-000012:0.8>",
+    "loras": [
+      {
+        "name": "ck-neon-retrowave-IL-000012",
+        "strength": 0.8,
+        "active": true
+      },
+      {
+        "name": "__dummy_item1__",
+        "strength": 0,
+        "active": false,
+        "_isDummy": true
+      },
+      {
+        "name": "__dummy_item2__",
+        "strength": 0,
+        "active": false,
+        "_isDummy": true
+      }
+    ]
+  },
+  "class_type": "Lora Stacker (LoraManager)",
+  "_meta": {
+    "title": "Lora Stacker (LoraManager)"
+  }
+}
+
+Test completed.

requirements.txt

@@ -1,4 +1,9 @@
 aiohttp
 jinja2
 safetensors
-watchdog
+watchdog
+beautifulsoup4
+piexif
+Pillow
+olefile
+requests

static/css/base.css

@@ -1,6 +1,8 @@
-/* 强制显示滚动条，防止页面跳动 */
-html {
-    overflow-y: scroll;
+html, body {
+    margin: 0;
+    padding: 0;
+    height: 100%;
+    overflow: hidden; /* Disable default scrolling */
 }
 
 /* 针对Firefox */
@@ -16,6 +18,7 @@ html {
 
 ::-webkit-scrollbar-track {
     background: transparent;
+    margin-top: 0;
 }
 
 ::-webkit-scrollbar-thumb {
@@ -35,6 +38,7 @@ html {
     --lora-border: oklch(90% 0.02 256 / 0.15);
     --lora-text: oklch(95% 0.02 256);
     --lora-error: oklch(75% 0.32 29);
+    --lora-warning: oklch(75% 0.25 80); /* Add warning color for deleted LoRAs */
     
     /* Spacing Scale */
     --space-1: calc(8px * 1);
@@ -43,6 +47,7 @@ html {
     
     /* Z-index Scale */
     --z-base: 10;
+    --z-header: 100;
     --z-modal: 1000;
     --z-overlay: 2000;
 
@@ -64,11 +69,14 @@ html {
     --lora-surface: oklch(25% 0.02 256 / 0.98);
     --lora-border: oklch(90% 0.02 256 / 0.15);
     --lora-text: oklch(98% 0.02 256);
+    --lora-warning: oklch(75% 0.25 80); /* Add warning color for dark theme too */
 }
 
 body {
-    margin: 0;
     font-family: 'Segoe UI', sans-serif;
     background: var(--bg-color);
     color: var(--text-color);
+    display: flex;
+    flex-direction: column;
+    padding-top: 0; /* Remove the padding-top */
 }

static/css/components/bulk.css

@@ -262,6 +262,83 @@
     background: var(--lora-accent);
 }
 
+/* NSFW Level Selector */
+.nsfw-level-selector {
+    position: fixed;
+    top: 50%;
+    left: 50%;
+    transform: translate(-50%, -50%);
+    background: var(--card-bg);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-base);
+    padding: 16px;
+    box-shadow: 0 4px 20px rgba(0, 0, 0, 0.2);
+    z-index: var(--z-modal);
+    width: 300px;
+    display: none;
+}
+
+.nsfw-level-header {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    margin-bottom: 16px;
+}
+
+.nsfw-level-header h3 {
+    margin: 0;
+    font-size: 16px;
+    font-weight: 500;
+}
+
+.close-nsfw-selector {
+    background: transparent;
+    border: none;
+    color: var(--text-color);
+    cursor: pointer;
+    padding: 4px;
+    border-radius: var(--border-radius-xs);
+}
+
+.close-nsfw-selector:hover {
+    background: var(--border-color);
+}
+
+.current-level {
+    margin-bottom: 12px;
+    padding: 8px;
+    background: var(--bg-color);
+    border-radius: var(--border-radius-xs);
+    border: 1px solid var(--border-color);
+}
+
+.nsfw-level-options {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px;
+}
+
+.nsfw-level-btn {
+    flex: 1 0 calc(33% - 8px);
+    padding: 8px;
+    border-radius: var(--border-radius-xs);
+    background: var(--bg-color);
+    border: 1px solid var(--border-color);
+    color: var(--text-color);
+    cursor: pointer;
+    transition: all 0.2s ease;
+}
+
+.nsfw-level-btn:hover {
+    background: var(--lora-border);
+}
+
+.nsfw-level-btn.active {
+    background: var(--lora-accent);
+    color: white;
+    border-color: var(--lora-accent);
+}
+
 /* Mobile optimizations */
 @media (max-width: 768px) {
     .selected-thumbnails-strip {

static/css/components/card.css

@@ -20,6 +20,10 @@
     aspect-ratio: 896/1152;
     max-width: 260px; /* Adjusted from 320px to fit 5 cards */
     margin: 0 auto;
+    cursor: pointer; /* Added from recipe-card */
+    display: flex; /* Added from recipe-card */
+    flex-direction: column; /* Added from recipe-card */
+    overflow: hidden; /* Add overflow hidden to contain children */
 }
 
 .lora-card:hover {
@@ -47,9 +51,11 @@
 .card-preview {
     position: relative;
     width: 100%;
-    height: 100%;
+    height: 100%; /* This should work with aspect-ratio on parent */
     border-radius: var(--border-radius-base);
     overflow: hidden;
+    flex-shrink: 0; /* Prevent shrinking */
+    min-height: 0; /* Fix for potential flexbox sizing issue in Firefox */
 }
 
 .card-preview img,
@@ -57,6 +63,97 @@
     width: 100%;
     height: 100%;
     object-fit: cover;
+    object-position: center top; /* Align the top of the image with the top of the container */
+}
+
+/* NSFW Content Blur */
+.card-preview.blurred img,
+.card-preview.blurred video {
+    filter: blur(25px);
+}
+
+.nsfw-overlay {
+    position: absolute;
+    top: 0;
+    left: 0;
+    right: 0;
+    bottom: 0;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    z-index: 2;
+    pointer-events: none;
+}
+
+.nsfw-warning {
+    text-align: center;
+    color: white;
+    background: rgba(0, 0, 0, 0.6);
+    padding: var(--space-2);
+    border-radius: var(--border-radius-base);
+    backdrop-filter: blur(4px);
+    max-width: 80%;
+    pointer-events: auto;
+}
+
+.nsfw-warning p {
+    margin: 0 0 var(--space-1);
+    font-weight: bold;
+    font-size: 1.1em;
+    text-shadow: 1px 1px 1px rgba(0, 0, 0, 0.5);
+}
+
+.toggle-blur-btn {
+    position: absolute;
+    left: var(--space-1);
+    top: var(--space-1);
+    background: rgba(0, 0, 0, 0.5);
+    border: none;
+    border-radius: 50%;
+    width: 24px;
+    height: 24px;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    color: white;
+    cursor: pointer;
+    z-index: 3;
+    transition: background-color 0.2s, transform 0.2s;
+}
+
+.toggle-blur-btn:hover {
+    background: rgba(0, 0, 0, 0.7);
+    transform: scale(1.1);
+}
+
+.toggle-blur-btn i {
+    font-size: 0.9em;
+}
+
+.show-content-btn {
+    background: var(--lora-accent);
+    color: white;
+    border: none;
+    border-radius: var(--border-radius-xs);
+    padding: 4px var(--space-1);
+    cursor: pointer;
+    font-size: 0.9em;
+    transition: background-color 0.2s, transform 0.2s;
+}
+
+.show-content-btn:hover {
+    background: oklch(58% 0.28 256);
+    transform: scale(1.05);
+}
+
+/* Adjust base model label positioning when toggle button is present */
+.base-model-label.with-toggle {
+    margin-left: 28px; /* Make room for the toggle button */
+}
+
+/* Ensure card actions remain clickable */
+.card-header .card-actions {
+    z-index: 3;
 }
 
 .card-footer {
@@ -183,4 +280,55 @@
     border-radius: var(--border-radius-xs);
     backdrop-filter: blur(2px);
     font-size: 0.85em;
+}
+
+/* Recipe specific elements - migrated from recipe-card.css */
+.recipe-indicator {
+    position: absolute;
+    top: 6px;
+    left: 8px;
+    width: 24px;
+    height: 24px;
+    background: var(--lora-primary);
+    border-radius: 50%;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    color: white;
+    font-weight: bold;
+    z-index: 2;
+}
+
+.base-model-wrapper {
+    display: flex;
+    align-items: center;
+    gap: 8px;
+    margin-left: 32px; /* For accommodating the recipe indicator */
+}
+
+.lora-count {
+    display: flex;
+    align-items: center;
+    gap: 4px;
+    background: rgba(255, 255, 255, 0.2);
+    padding: 2px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.85em;
+    position: relative;
+}
+
+.lora-count.ready {
+    background: rgba(46, 204, 113, 0.3);
+}
+
+.lora-count.missing {
+    background: rgba(231, 76, 60, 0.3);
+}
+
+.placeholder-message {
+    grid-column: 1 / -1;
+    text-align: center;
+    padding: 2rem;
+    background: var(--lora-surface-alt);
+    border-radius: var(--border-radius-base);
 }

static/css/components/download-modal.css

@@ -23,12 +23,6 @@
     color: var(--text-color);
 }
 
-.error-message {
-    color: var(--lora-error);
-    font-size: 0.9em;
-    margin-top: 4px;
-}
-
 /* Version List Styles */
 .version-list {
     max-height: 400px;
@@ -104,6 +98,7 @@
 .version-info {
     display: flex;
     flex-wrap: wrap;
+    flex-direction: row !important;
     gap: 8px;
     align-items: center;
     font-size: 0.9em;
@@ -130,50 +125,6 @@
     gap: 4px;
 }
 
-/* Local Version Badge */
-.local-badge {
-    display: inline-flex;
-    align-items: center;
-    background: var(--lora-accent);
-    color: var(--lora-text);
-    padding: 4px 8px;
-    border-radius: var(--border-radius-xs);
-    font-size: 0.8em;
-    font-weight: 500;
-    white-space: nowrap;
-    flex-shrink: 0;
-    position: relative;
-}
-
-.local-badge i {
-    margin-right: 4px;
-    font-size: 0.9em;
-}
-
-.local-path {
-    display: none;
-    position: absolute;
-    top: 100%;
-    right: 0;
-    background: var(--card-bg);
-    border: 1px solid var(--border-color);
-    border-radius: var(--border-radius-xs);
-    padding: var(--space-1);
-    margin-top: 4px;
-    font-size: 0.9em;
-    color: var(--text-color);
-    white-space: normal;
-    word-break: break-all;
-    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
-    z-index: 1;
-    min-width: 200px;
-    max-width: 300px;
-}
-
-.local-badge:hover .local-path {
-    display: block;
-}
-
 /* Folder Browser Styles */
 .folder-browser {
     border: 1px solid var(--border-color);
@@ -251,47 +202,4 @@
 .version-item.exists-locally {
     background: oklch(var(--lora-accent) / 0.05);
     border-left: 4px solid var(--lora-accent);
-}
-
-.local-badge {
-    display: inline-flex;
-    align-items: center;
-    background: var(--lora-accent);
-    color: var(--lora-text);
-    padding: 4px 8px;
-    border-radius: var(--border-radius-xs);
-    font-size: 0.8em;
-    font-weight: 500;
-    white-space: nowrap;
-    flex-shrink: 0;
-    position: relative;
-}
-
-.local-badge i {
-    margin-right: 4px;
-    font-size: 0.9em;
-}
-
-.local-path {
-    display: none;
-    position: absolute;
-    top: 100%;
-    right: 0;
-    background: var(--card-bg);
-    border: 1px solid var(--border-color);
-    border-radius: var(--border-radius-xs);
-    padding: var(--space-1);
-    margin-top: 4px;
-    font-size: 0.9em;
-    color: var(--text-color);
-    white-space: normal;
-    word-break: break-all;
-    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
-    z-index: 1;
-    min-width: 200px;
-    max-width: 300px;
-}
-
-.local-badge:hover .local-path {
-    display: block;
-} 
+}

static/css/components/filter-indicator.css

@@ -0,0 +1,84 @@
+/* Filter indicator styles */
+.control-group .filter-active {
+    display: flex;
+    align-items: center;
+    gap: 6px;
+    background: var(--lora-accent);
+    color: white;
+    border-radius: var(--border-radius-xs);
+    padding: 4px 10px;
+    transition: all 0.2s ease;
+    border: 1px solid var(--lora-accent);
+    cursor: pointer;
+    box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1);
+    font-size: 0.85em;
+}
+
+.control-group .filter-active:hover {
+    opacity: 0.92;
+    transform: translateY(-1px);
+    box-shadow: 0 3px 5px rgba(0, 0, 0, 0.15);
+}
+
+.control-group .filter-active:active {
+    transform: translateY(0);
+    box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1);
+}
+
+.control-group .filter-active i.fa-filter {
+    font-size: 0.9em;
+    margin-right: 2px;
+    opacity: 0.9;
+}
+
+.control-group .filter-active i.clear-filter {
+    transition: transform 0.2s ease, background-color 0.2s ease;
+    cursor: pointer;
+    margin-left: 4px;
+    border-radius: 50%;
+    font-size: 0.85em;
+    width: 16px;
+    height: 16px;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+}
+
+.control-group .filter-active i.clear-filter:hover {
+    transform: scale(1.2);
+    background-color: rgba(255, 255, 255, 0.2);
+}
+
+.control-group .filter-active .lora-name {
+    font-weight: 500;
+    max-width: 150px;
+    overflow: hidden;
+    text-overflow: ellipsis;
+    white-space: nowrap;
+}
+
+/* Animation for filter indicator */
+@keyframes filterPulse {
+    0% { transform: scale(1); box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1); }
+    50% { transform: scale(1.03); box-shadow: 0 3px 8px rgba(0, 0, 0, 0.15); }
+    100% { transform: scale(1); box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1); }
+}
+
+.filter-active.animate {
+    animation: filterPulse 0.6s ease;
+}
+
+/* Make responsive */
+@media (max-width: 576px) {
+    .control-group .filter-active {
+        padding: 6px 10px;
+    }
+    
+    .control-group .filter-active .lora-name {
+        max-width: 100px;
+    }
+    
+    .control-group .filter-active:hover {
+        transform: none; /* Disable hover effects on mobile */
+    }
+}

static/css/components/header.css

@@ -0,0 +1,177 @@
+.app-header {
+  background: var(--card-bg);
+  border-bottom: 1px solid var(--border-color);
+  position: fixed;
+  top: 0;
+  z-index: var(--z-header);
+  height: 48px; /* Reduced height */
+  width: 100%;
+  box-shadow: 0 1px 3px rgba(0,0,0,0.05);
+}
+
+.header-container {
+  max-width: 1400px;
+  margin: 0 auto;
+  padding: 0 15px;
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  height: 100%;
+}
+
+/* Logo and title styling */
+.header-branding {
+  display: flex;
+  align-items: center;
+  flex-shrink: 0;
+}
+
+.logo-link {
+  display: flex;
+  align-items: center;
+  text-decoration: none;
+  color: var(--text-color);
+  gap: 8px;
+}
+
+.app-logo {
+  width: 24px;
+  height: 24px;
+}
+
+.app-title {
+  font-size: 1rem;
+  font-weight: 600;
+  margin: 0;
+}
+
+/* Navigation styling */
+.main-nav {
+  display: flex;
+  gap: 0.5rem;
+  flex-shrink: 0;
+  margin-right: 1rem;
+}
+
+.nav-item {
+  padding: 0.25rem 0.75rem;
+  border-radius: var(--border-radius-xs);
+  color: var(--text-color);
+  text-decoration: none;
+  display: flex;
+  align-items: center;
+  gap: 0.5rem;
+  transition: all 0.2s ease;
+  font-size: 0.9rem;
+}
+
+.nav-item:hover {
+  background-color: var(--lora-surface-hover, oklch(95% 0.02 256));
+}
+
+.nav-item.active {
+  background-color: var(--lora-accent);
+  color: white;
+}
+
+/* Header search */
+.header-search {
+  flex: 1;
+  max-width: 400px;
+  margin: 0 1rem;
+}
+
+/* Header controls (formerly corner controls) */
+.header-controls {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  flex-shrink: 0;
+}
+
+.header-controls > div {
+  width: 32px;
+  height: 32px;
+  border-radius: 50%;
+  background: var(--card-bg);
+  border: 1px solid var(--border-color);
+  color: var(--text-color);
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  cursor: pointer;
+  transition: all 0.2s ease;
+  position: relative;
+}
+
+.header-controls > div:hover {
+  background: var(--lora-accent);
+  color: white;
+  transform: translateY(-2px);
+}
+
+.theme-toggle {
+  position: relative; /* Ensure relative positioning for the container */
+}
+
+.theme-toggle .light-icon,
+.theme-toggle .dark-icon {
+  position: absolute;
+  top: 50%;
+  left: 50%;
+  transform: translate(-50%, -50%); /* Center perfectly */
+  opacity: 0;
+  transition: opacity 0.3s ease;
+}
+
+.theme-toggle .dark-icon {
+  opacity: 1;
+}
+
+[data-theme="light"] .theme-toggle .light-icon {
+  opacity: 1;
+}
+
+[data-theme="light"] .theme-toggle .dark-icon {
+  opacity: 0;
+}
+
+/* Mobile adjustments */
+@media (max-width: 768px) {
+  .app-title {
+    display: none; /* Hide text title on mobile */
+  }
+  
+  .header-controls {
+    gap: 4px;
+  }
+  
+  .header-controls > div {
+    width: 28px;
+    height: 28px;
+  }
+  
+  .header-search {
+    max-width: none;
+    margin: 0 0.5rem;
+  }
+  
+  .main-nav {
+    margin-right: 0.5rem;
+  }
+}
+
+/* For very small screens */
+@media (max-width: 600px) {
+  .header-container {
+    padding: 0 8px;
+  }
+  
+  .main-nav {
+    display: none; /* Hide navigation on very small screens */
+  }
+  
+  .header-search {
+    flex: 1;
+  }
+}

static/css/components/import-modal.css

@@ -0,0 +1,735 @@
+/* Import Modal Styles */
+.import-step {
+    margin: var(--space-2) 0;
+    transition: none !important; /* Disable any transitions that might affect display */
+}
+
+/* Import Mode Toggle */
+.import-mode-toggle {
+    display: flex;
+    margin-bottom: var(--space-3);
+    border-radius: var(--border-radius-sm);
+    overflow: hidden;
+    border: 1px solid var(--border-color);
+}
+
+.toggle-btn {
+    flex: 1;
+    padding: 10px 16px;
+    background: var(--bg-color);
+    color: var(--text-color);
+    border: none;
+    cursor: pointer;
+    font-weight: 500;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    gap: 8px;
+    transition: background-color 0.2s, color 0.2s;
+}
+
+.toggle-btn:first-child {
+    border-right: 1px solid var(--border-color);
+}
+
+.toggle-btn.active {
+    background: var(--lora-accent);
+    color: var(--lora-text);
+}
+
+.toggle-btn:hover:not(.active) {
+    background: var(--lora-surface);
+}
+
+.import-section {
+    margin-bottom: var(--space-3);
+}
+
+/* File Input Styles */
+.file-input-wrapper {
+    position: relative;
+    margin-bottom: var(--space-1);
+}
+
+.file-input-wrapper input[type="file"] {
+    position: absolute;
+    width: 100%;
+    height: 100%;
+    opacity: 0;
+    cursor: pointer;
+    z-index: 2;
+}
+
+.file-input-button {
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    gap: 8px;
+    padding: 10px 16px;
+    background: var(--lora-accent);
+    color: var(--lora-text);
+    border-radius: var(--border-radius-xs);
+    font-weight: 500;
+    cursor: pointer;
+    transition: background-color 0.2s;
+}
+
+.file-input-button:hover {
+    background: oklch(from var(--lora-accent) l c h / 0.9);
+}
+
+.file-input-wrapper:hover .file-input-button {
+    background: oklch(from var(--lora-accent) l c h / 0.9);
+}
+
+/* Recipe Details Layout */
+.recipe-details-layout {
+    display: grid;
+    grid-template-columns: 200px 1fr;
+    gap: var(--space-3);
+    margin-bottom: var(--space-3);
+}
+
+.recipe-image-container {
+    width: 100%;
+    height: 200px;
+    border-radius: var(--border-radius-sm);
+    overflow: hidden;
+    background: var(--lora-surface);
+    border: 1px solid var(--border-color);
+}
+
+.recipe-image {
+    width: 100%;
+    height: 100%;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+}
+
+.recipe-image img {
+    max-width: 100%;
+    max-height: 100%;
+    object-fit: contain;
+}
+
+.recipe-form-container {
+    display: flex;
+    flex-direction: column;
+    gap: var(--space-2);
+}
+
+/* Tags Input Styles */
+.tag-input-container {
+    display: flex;
+    gap: 8px;
+    margin-bottom: var(--space-1);
+}
+
+.tag-input-container input {
+    flex: 1;
+    padding: 8px;
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    background: var(--bg-color);
+    color: var(--text-color);
+}
+
+.tags-container {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px;
+    margin-top: var(--space-1);
+    min-height: 32px;
+}
+
+.recipe-tag {
+    display: inline-flex;
+    align-items: center;
+    gap: 6px;
+    padding: 4px 10px;
+    background: var(--lora-surface);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    font-size: 0.9em;
+}
+
+.recipe-tag i {
+    cursor: pointer;
+    opacity: 0.7;
+    transition: opacity 0.2s;
+}
+
+.recipe-tag i:hover {
+    opacity: 1;
+    color: var(--lora-error);
+}
+
+.empty-tags {
+    color: var(--text-color);
+    opacity: 0.6;
+    font-size: 0.9em;
+    font-style: italic;
+}
+
+/* LoRAs List Styles */
+.loras-list {
+    max-height: 300px;
+    overflow-y: auto;
+    margin: var(--space-2) 0;
+    display: flex;
+    flex-direction: column;
+    gap: 12px;
+    padding: 1px;
+}
+
+.lora-item {
+    display: flex;
+    gap: var(--space-2);
+    padding: var(--space-2);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-sm);
+    background: var(--bg-color);
+    margin: 1px;
+}
+
+.lora-item.exists-locally {
+    background: oklch(var(--lora-accent) / 0.05);
+    border-left: 4px solid var(--lora-accent);
+}
+
+.lora-item.missing-locally {
+    border-left: 4px solid var(--lora-error);
+}
+
+.lora-item.is-deleted {
+    background: oklch(var(--lora-warning) / 0.05);
+    border-left: 4px solid var(--lora-warning);
+}
+
+.lora-item.is-early-access {
+    background: rgba(0, 184, 122, 0.05);
+    border-left: 4px solid #00B87A;
+}
+
+.lora-item.missing-locally {
+    border-left: 4px solid var(--lora-error);
+}
+
+.lora-thumbnail {
+    width: 80px;
+    height: 80px;
+    flex-shrink: 0;
+    border-radius: var(--border-radius-xs);
+    overflow: hidden;
+    background: var(--bg-color);
+}
+
+.lora-thumbnail img {
+    width: 100%;
+    height: 100%;
+    object-fit: cover;
+}
+
+.lora-content {
+    display: flex;
+    flex-direction: column;
+    gap: 8px;
+    flex: 1;
+    min-width: 0;
+}
+
+.lora-header {
+    display: flex;
+    align-items: flex-start;
+    justify-content: space-between;
+    gap: var(--space-2);
+}
+
+.lora-content h3 {
+    margin: 0;
+    font-size: 1.1em;
+    color: var(--text-color);
+    flex: 1;
+}
+
+.lora-info {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px;
+    align-items: center;
+    font-size: 0.9em;
+}
+
+.lora-info .base-model {
+    background: oklch(var(--lora-accent) / 0.1);
+    color: var(--lora-accent);
+    padding: 2px 8px;
+    border-radius: var(--border-radius-xs);
+}
+
+.lora-version {
+    font-size: 0.9em;
+    color: var(--text-color);
+    opacity: 0.7;
+}
+
+.weight-badge {
+    background: var(--lora-surface);
+    padding: 2px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.85em;
+}
+
+/* Missing LoRAs List */
+.missing-loras-list {
+    max-height: 200px;
+    overflow-y: auto;
+    margin: var(--space-2) 0;
+    display: flex;
+    flex-direction: column;
+    gap: 8px;
+    padding: var(--space-1);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-sm);
+    background: var(--lora-surface);
+}
+
+.missing-lora-item {
+    display: flex;
+    gap: var(--space-2);
+    padding: var(--space-1);
+    border-bottom: 1px solid var(--border-color);
+}
+
+.missing-lora-item:last-child {
+    border-bottom: none;
+}
+
+.missing-lora-item.is-early-access {
+    background: rgba(0, 184, 122, 0.05);
+    border-left: 3px solid #00B87A;
+    padding-left: 10px;
+}
+
+.missing-badge {
+    display: inline-flex;
+    align-items: center;
+    background: var(--lora-error);
+    color: white;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.8em;
+    font-weight: 500;
+    white-space: nowrap;
+    flex-shrink: 0;
+}
+
+.missing-badge i {
+    margin-right: 4px;
+    font-size: 0.9em;
+}
+
+.lora-count-info {
+    font-size: 0.85em;
+    opacity: 0.8;
+    font-weight: normal;
+    margin-left: 8px;
+}
+
+/* Location Selection Styles */
+.location-selection {
+    margin: var(--space-2) 0;
+    padding: var(--space-2);
+    background: var(--lora-surface);
+    border-radius: var(--border-radius-sm);
+}
+
+/* Reuse folder browser and path preview styles from download-modal.css */
+.folder-browser {
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    padding: var(--space-1);
+    max-height: 200px;
+    overflow-y: auto;
+}
+
+.folder-item {
+    padding: 8px;
+    cursor: pointer;
+    border-radius: var(--border-radius-xs);
+    transition: background-color 0.2s;
+}
+
+.folder-item:hover {
+    background: var(--lora-surface);
+}
+
+.folder-item.selected {
+    background: oklch(var(--lora-accent) / 0.1);
+    border: 1px solid var(--lora-accent);
+}
+
+.path-preview {
+    margin-bottom: var(--space-3);
+    padding: var(--space-2);
+    background: var(--bg-color);
+    border-radius: var(--border-radius-sm);
+    border: 1px dashed var(--border-color);
+}
+
+.path-preview label {
+    display: block;
+    margin-bottom: 8px;
+    color: var(--text-color);
+    font-size: 0.9em;
+    opacity: 0.8;
+}
+
+.path-display {
+    padding: var(--space-1);
+    color: var(--text-color);
+    font-family: monospace;
+    font-size: 0.9em;
+    line-height: 1.4;
+    white-space: pre-wrap;
+    word-break: break-all;
+    opacity: 0.85;
+    background: var(--lora-surface);
+    border-radius: var(--border-radius-xs);
+}
+
+/* Input Group Styles */
+.input-group {
+    margin-bottom: var(--space-2);
+}
+
+.input-with-button {
+    display: flex;
+    gap: 8px;
+}
+
+.input-with-button input {
+    flex: 1;
+    min-width: 0;
+}
+
+.input-with-button button {
+    flex-shrink: 0;
+    white-space: nowrap;
+    padding: 8px 16px;
+    background: var(--lora-accent);
+    color: var(--lora-text);
+    border: none;
+    border-radius: var(--border-radius-xs);
+    cursor: pointer;
+    transition: background-color 0.2s;
+}
+
+.input-with-button button:hover {
+    background: oklch(from var(--lora-accent) l c h / 0.9);
+}
+
+.input-group label {
+    display: block;
+    margin-bottom: 8px;
+    color: var(--text-color);
+}
+
+.input-group input,
+.input-group select {
+    width: 100%;
+    padding: 8px;
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    background: var(--bg-color);
+    color: var(--text-color);
+}
+
+/* Dark theme adjustments */
+[data-theme="dark"] .lora-item {
+    background: var(--lora-surface);
+}
+
+[data-theme="dark"] .recipe-tag {
+    background: var(--card-bg);
+}
+
+/* Responsive adjustments */
+@media (max-width: 768px) {
+    .recipe-details-layout {
+        grid-template-columns: 1fr;
+    }
+    
+    .recipe-image-container {
+        height: 150px;
+    }
+}
+
+/* Size badge for LoRA items */
+.size-badge {
+    background: var(--lora-surface);
+    padding: 2px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.85em;
+    color: var(--text-color);
+    opacity: 0.8;
+}
+
+/* Improved Missing LoRAs summary section */
+.missing-loras-summary {
+    margin-bottom: var(--space-3);
+    padding: var(--space-2);
+    background: var(--bg-color);
+    border-radius: var(--border-radius-sm);
+    border: 1px solid var(--border-color);
+}
+
+.summary-header {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    margin-bottom: 0;
+}
+
+.summary-header h3 {
+    margin: 0;
+    font-size: 1.1em;
+    color: var(--text-color);
+    display: flex;
+    align-items: center;
+    gap: var(--space-1);
+}
+
+.lora-count-badge {
+    font-size: 0.9em;
+    font-weight: normal;
+    opacity: 0.7;
+}
+
+.total-size-badge {
+    font-size: 0.85em;
+    font-weight: normal;
+    background: var(--lora-surface);
+    padding: 2px 8px;
+    border-radius: var(--border-radius-xs);
+    margin-left: var(--space-1);
+}
+
+.toggle-list-btn {
+    background: none;
+    border: none;
+    cursor: pointer;
+    color: var(--text-color);
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+}
+
+.toggle-list-btn:hover {
+    background: var(--lora-surface);
+}
+
+.missing-loras-list {
+    max-height: 200px;
+    overflow-y: auto;
+    transition: max-height 0.3s ease, margin-top 0.3s ease, padding-top 0.3s ease;
+    margin-top: 0;
+    padding-top: 0;
+}
+
+.missing-loras-list.collapsed {
+    max-height: 0;
+    overflow: hidden;
+    padding-top: 0;
+}
+
+.missing-loras-list:not(.collapsed) {
+    margin-top: var(--space-1);
+    padding-top: var(--space-1);
+    border-top: 1px solid var(--border-color);
+}
+
+.missing-lora-item {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    padding: 8px;
+    border-bottom: 1px solid var(--border-color);
+}
+
+.missing-lora-item:last-child {
+    border-bottom: none;
+}
+
+.missing-lora-info {
+    display: flex;
+    flex-direction: column;
+    gap: 4px;
+}
+
+.missing-lora-name {
+    font-weight: 500;
+}
+
+.lora-base-model {
+    font-size: 0.85em;
+    color: var(--lora-accent);
+    background: oklch(var(--lora-accent) / 0.1);
+    padding: 2px 6px;
+    border-radius: var(--border-radius-xs);
+    display: inline-block;
+}
+
+.missing-lora-size {
+    font-size: 0.9em;
+    color: var(--text-color);
+    opacity: 0.8;
+}
+
+/* Recipe name input select-all behavior */
+#recipeName:focus {
+    outline: 2px solid var(--lora-accent);
+}
+
+/* Prevent layout shift with scrollbar */
+.modal-content {
+    overflow-y: scroll; /* Always show scrollbar */
+    scrollbar-gutter: stable; /* Reserve space for scrollbar */
+}
+
+/* For browsers that don't support scrollbar-gutter */
+@supports not (scrollbar-gutter: stable) {
+    .modal-content {
+        padding-right: calc(var(--space-2) + var(--scrollbar-width)); /* Add extra padding for scrollbar */
+    }
+}
+
+/* Deleted LoRA styles - Fix layout issues */
+.lora-item.is-deleted {
+    background: oklch(var(--lora-warning) / 0.05);
+    border-left: 4px solid var(--lora-warning);
+}
+
+.deleted-badge {
+    display: inline-flex;
+    align-items: center;
+    background: var(--lora-warning);
+    color: white;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.8em;
+    font-weight: 500;
+    white-space: nowrap;
+    flex-shrink: 0;
+}
+
+.deleted-badge i {
+    margin-right: 4px;
+    font-size: 0.9em;
+}
+
+.exclude-lora-checkbox {
+    display: none;
+}
+
+/* Deleted LoRAs warning - redesigned to not interfere with modal buttons */
+.deleted-loras-warning {
+    display: flex;
+    align-items: flex-start;
+    gap: 12px;
+    padding: 12px 16px;
+    background: oklch(var(--lora-warning) / 0.1);
+    border: 1px solid var(--lora-warning);
+    border-radius: var(--border-radius-sm);
+    color: var(--text-color);
+    margin-bottom: var(--space-2);
+}
+
+.warning-icon {
+    color: var(--lora-warning);
+    font-size: 1.2em;
+    padding-top: 2px;
+}
+
+.warning-content {
+    flex: 1;
+}
+
+.warning-title {
+    font-weight: 600;
+    margin-bottom: 4px;
+}
+
+.warning-text {
+    font-size: 0.9em;
+    line-height: 1.4;
+}
+
+/* Remove the old warning-message styles that were causing layout issues */
+.warning-message {
+    display: none; /* Hide the old style */
+}
+
+/* Update deleted badge to be more prominent */
+.deleted-badge {
+    display: inline-flex;
+    align-items: center;
+    background: var(--lora-warning);
+    color: white;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.8em;
+    font-weight: 500;
+    white-space: nowrap;
+    flex-shrink: 0;
+}
+
+.deleted-badge i {
+    margin-right: 4px;
+    font-size: 0.9em;
+}
+
+/* Error message styling */
+.error-message {
+    color: var(--lora-error);
+    font-size: 0.9em;
+    margin-top: 8px;
+    min-height: 20px; /* Ensure there's always space for the error message */
+    font-weight: 500;
+}
+
+.early-access-warning {
+    display: flex;
+    align-items: flex-start;
+    gap: 12px;
+    padding: 12px 16px;
+    background: rgba(0, 184, 122, 0.1);
+    border: 1px solid #00B87A;
+    border-radius: var(--border-radius-sm);
+    color: var(--text-color);
+    margin-bottom: var(--space-2);
+}
+
+/* Add special styling for early access badge in the missing loras list */
+.missing-lora-item .early-access-badge {
+    padding: 2px 6px;
+    font-size: 0.75em;
+    margin-top: 4px;
+    display: inline-flex;
+}
+
+/* Specific styling for the early access warning container in import modal */
+.early-access-warning .warning-icon {
+    color: #00B87A;
+    font-size: 1.2em;
+}
+
+.early-access-warning .warning-title {
+    font-weight: 600;
+    margin-bottom: 4px;
+}
+
+.early-access-warning .warning-text {
+    font-size: 0.9em;
+    line-height: 1.4;
+}

static/css/components/initialization.css

@@ -0,0 +1,359 @@
+/* Initialization Component Styles */
+
+.initialization-container {
+  width: 100%;
+  height: 100%;
+  padding: var(--space-3);
+  background: var(--lora-surface);
+  animation: fadeIn 0.3s ease-in-out;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+}
+
+.initialization-content {
+  max-width: 800px;
+  width: 100%;
+}
+
+/* Override loading.css width for initialization component */
+.initialization-container .loading-content {
+  width: 100%;
+  max-width: 100%;
+  background: transparent;
+  backdrop-filter: none;
+  border: none;
+  padding: 0;
+}
+
+.initialization-header {
+  text-align: center;
+  margin-bottom: var(--space-3);
+}
+
+.initialization-header h2 {
+  font-size: 1.8rem;
+  margin-bottom: var(--space-1);
+  color: var(--text-color);
+}
+
+.init-subtitle {
+  color: var(--text-color);
+  opacity: 0.8;
+  font-size: 1rem;
+}
+
+/* Progress Bar Styles specific to initialization */
+.initialization-progress {
+  margin-bottom: var(--space-3);
+}
+
+/* Renamed container class */
+.init-progress-container {
+  width: 100%; /* Use full width within its container */
+  height: 8px; /* Match height from previous .progress-bar-container */
+  background-color: var(--lora-border); /* Consistent background */
+  border-radius: 4px;
+  overflow: hidden;
+  margin: 0 auto var(--space-1); /* Center horizontally, add bottom margin */
+}
+
+/* Renamed progress bar class */
+.init-progress-bar {
+  height: 100%;
+  /* Use a gradient consistent with the theme accent */
+  background: linear-gradient(90deg, var(--lora-accent) 0%, color-mix(in oklch, var(--lora-accent) 80%, transparent) 100%);
+  border-radius: 4px; /* Match container radius */
+  transition: width 0.3s ease;
+  width: 0%; /* Start at 0% */
+}
+
+/* Remove the old .progress-bar rule specific to initialization to avoid conflicts */
+/* .progress-bar { ... } */
+
+/* Progress Details */
+.progress-details {
+  display: flex;
+  justify-content: space-between;
+  font-size: 0.9rem;
+  color: var(--text-color);
+  margin-top: var(--space-1);
+  padding: 0 2px;
+}
+
+#remainingTime {
+  font-style: italic;
+  color: var(--text-color);
+  opacity: 0.8;
+}
+
+/* Stages Styles */
+.initialization-stages {
+  margin-bottom: var(--space-3);
+}
+
+.stage-item {
+  display: flex;
+  align-items: flex-start;
+  padding: var(--space-2);
+  border-radius: var(--border-radius-xs);
+  margin-bottom: var(--space-1);
+  transition: background-color 0.2s ease;
+  border: 1px solid transparent;
+}
+
+.stage-item.active {
+  background-color: rgba(var(--lora-accent), 0.1);
+  border-color: var(--lora-accent);
+}
+
+.stage-item.completed {
+  background-color: rgba(0, 150, 0, 0.05);
+  border-color: rgba(0, 150, 0, 0.2);
+}
+
+.stage-icon {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  width: 40px;
+  height: 40px;
+  background: var(--lora-border);
+  border-radius: 50%;
+  margin-right: var(--space-2);
+}
+
+.stage-item.active .stage-icon {
+  background: var(--lora-accent);
+  color: white;
+}
+
+.stage-item.completed .stage-icon {
+  background: rgb(0, 150, 0);
+  color: white;
+}
+
+.stage-content {
+  flex: 1;
+}
+
+.stage-content h4 {
+  margin: 0 0 5px 0;
+  font-size: 1rem;
+  color: var(--text-color);
+}
+
+.stage-details {
+  font-size: 0.85rem;
+  color: var(--text-color);
+  opacity: 0.8;
+}
+
+.stage-status {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  width: 24px;
+  height: 24px;
+}
+
+.stage-status.pending {
+  color: var(--text-color);
+  opacity: 0.5;
+}
+
+.stage-status.in-progress {
+  color: var(--lora-accent);
+}
+
+.stage-status.completed {
+  color: rgb(0, 150, 0);
+}
+
+/* Tips Container */
+.tips-container {
+  margin-top: var(--space-3);
+  background: rgba(var(--lora-accent), 0.05);
+  border-radius: var(--border-radius-base);
+  padding: var(--space-2);
+  box-shadow: 0 2px 8px rgba(0, 0, 0, 0.05);
+}
+
+.tips-header {
+  display: flex;
+  align-items: center;
+  margin-bottom: var(--space-2);
+  padding-bottom: var(--space-1);
+  border-bottom: 1px solid var(--lora-border);
+}
+
+.tips-header i {
+  margin-right: 10px;
+  color: var(--lora-accent);
+  font-size: 1.2rem;
+}
+
+.tips-header h3 {
+  font-size: 1.2rem;
+  margin: 0;
+  color: var(--text-color);
+}
+
+/* Tip Carousel with Images */
+.tips-content {
+  position: relative;
+}
+
+.tip-carousel {
+  position: relative;
+  height: 160px;
+  overflow: hidden;
+}
+
+.tip-item {
+  position: absolute;
+  width: 100%;
+  height: 100%;
+  display: flex;
+  opacity: 0;
+  transition: opacity 0.5s ease;
+  padding: 0;
+  border-radius: var(--border-radius-sm);
+  overflow: hidden;
+}
+
+.tip-item.active {
+  opacity: 1;
+}
+
+.tip-image {
+  width: 40%;
+  overflow: hidden;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  background-color: var(--lora-border);
+}
+
+.tip-image img {
+  width: 100%;
+  height: 100%;
+  object-fit: cover;
+}
+
+.tip-text {
+  width: 60%;
+  padding: var(--space-2);
+  display: flex;
+  flex-direction: column;
+  justify-content: center;
+}
+
+.tip-text h4 {
+  margin: 0 0 var(--space-1) 0;
+  font-size: 1.1rem;
+  color: var(--text-color);
+}
+
+.tip-text p {
+  margin: 0;
+  line-height: 1.5;
+  font-size: 0.9rem;
+  color: var(--text-color);
+}
+
+.tip-navigation {
+  display: flex;
+  justify-content: center;
+  margin-top: var(--space-2);
+}
+
+.tip-dot {
+  width: 10px;
+  height: 10px;
+  border-radius: 50%;
+  background-color: var(--lora-border);
+  margin: 0 5px;
+  cursor: pointer;
+  transition: background-color 0.2s ease, transform 0.2s ease;
+}
+
+.tip-dot:hover {
+  transform: scale(1.2);
+}
+
+.tip-dot.active {
+  background-color: var(--lora-accent);
+}
+
+/* Animation */
+@keyframes fadeIn {
+  from {
+    opacity: 0;
+    transform: translateY(10px);
+  }
+  to {
+    opacity: 1;
+    transform: translateY(0);
+  }
+}
+
+/* Different stage status animations */
+@keyframes pulse {
+  0% {
+    transform: scale(1);
+  }
+  50% {
+    transform: scale(1.2);
+  }
+  100% {
+    transform: scale(1);
+  }
+}
+
+.stage-item.active .stage-icon i {
+  animation: pulse 1s infinite;
+}
+
+/* Responsive Adjustments */
+@media (max-width: 768px) {
+  .initialization-container {
+    padding: var(--space-2);
+  }
+  
+  .stage-item {
+    padding: var(--space-1);
+  }
+  
+  .stage-icon {
+    width: 32px;
+    height: 32px;
+    min-width: 32px;
+  }
+
+  .tip-item {
+    flex-direction: column;
+    height: 220px;
+  }
+  
+  .tip-image, .tip-text {
+    width: 100%;
+  }
+  
+  .tip-image {
+    height: 120px;
+  }
+  
+  .tip-carousel {
+    height: 220px;
+  }
+}
+
+@media (prefers-reduced-motion: reduce) {
+  .initialization-container,
+  .tip-item,
+  .tip-dot {
+    transition: none;
+    animation: none;
+  }
+}

static/css/components/loading.css

@@ -56,6 +56,53 @@
     transition: width 200ms ease-out;
 }
 
+/* Enhanced progress display */
+.progress-details-container {
+    margin-top: var(--space-3);
+    width: 100%;
+    text-align: left;
+}
+
+.overall-progress-label {
+    font-size: 0.9rem;
+    margin-bottom: var(--space-1);
+    color: var(--text-color);
+}
+
+.current-item-progress {
+    margin-top: var(--space-2);
+}
+
+.current-item-label {
+    font-size: 0.9rem;
+    margin-bottom: var(--space-1);
+    color: var(--text-color);
+    white-space: nowrap;
+    overflow: hidden;
+    text-overflow: ellipsis;
+}
+
+.current-item-bar-container {
+    height: 8px;
+    background-color: var(--lora-border);
+    border-radius: 4px;
+    overflow: hidden;
+    margin-bottom: var(--space-1);
+}
+
+.current-item-bar {
+    height: 100%;
+    background-color: var(--lora-accent);
+    transition: width 200ms ease-out;
+    width: 0%;
+}
+
+.current-item-percent {
+    font-size: 0.8rem;
+    color: var(--text-color-secondary, var(--text-color));
+    opacity: 0.7;
+}
+
 @keyframes spin {
     0% { transform: rotate(0deg); }
     100% { transform: rotate(360deg); }
@@ -63,7 +110,8 @@
 
 @media (prefers-reduced-motion: reduce) {
     .lora-card,
-    .progress-bar {
+    .progress-bar,
+    .current-item-bar {
         transition: none;
     }
 }

static/css/components/modal.css

@@ -2,13 +2,13 @@
 .modal {
     display: none;
     position: fixed;
-    top: 0;
+    top: 48px; /* Start below the header */
     left: 0;
     width: 100%;
-    height: 100%;
+    height: calc(100% - 48px); /* Adjust height to exclude header */
     background: rgba(0, 0, 0, 0.2); /* 调整为更淡的半透明黑色 */
     z-index: var(--z-modal);
-    overflow: hidden; /* 改为 hidden，防止双滚动条 */
+    overflow: auto; /* Change from hidden to auto to allow scrolling */
 }
 
 /* 当模态窗口打开时，禁止body滚动 */
@@ -23,8 +23,8 @@ body.modal-open {
     position: relative;
     max-width: 800px;
     height: auto;
-    max-height: 90vh;
-    margin: 2rem auto;
+    max-height: calc(90vh - 48px); /* Adjust to account for header height */
+    margin: 1rem auto; /* Keep reduced top margin */
     background: var(--lora-surface);
     border-radius: var(--border-radius-base);
     padding: var(--space-3);
@@ -196,7 +196,7 @@ body.modal-open {
 }
 
 .settings-modal {
-    max-width: 500px;
+    max-width: 650px; /* Further increased from 600px for more space */
 }
 
 /* Settings Links */
@@ -266,14 +266,22 @@ body.modal-open {
     }
 }
 
+/* API key input specific styles */
 .api-key-input {
+    width: 100%; /* Take full width of parent */
     position: relative;
     display: flex;
     align-items: center;
 }
 
 .api-key-input input {
-    padding-right: 40px;
+    width: 100%;
+    padding: 6px 40px 6px 10px; /* Add left padding */
+    height: 32px;
+    border-radius: var(--border-radius-xs);
+    border: 1px solid var(--border-color);
+    background-color: var(--lora-surface);
+    color: var(--text-color);
 }
 
 .api-key-input .toggle-visibility {
@@ -294,8 +302,10 @@ body.modal-open {
 .input-help {
     font-size: 0.85em;
     color: var(--text-color);
-    opacity: 0.8;
-    margin-top: 4px;
+    opacity: 0.7;
+    margin-top: 8px; /* Space between control and help */
+    line-height: 1.4;
+    width: 100%; /* Full width */
 }
 
 /* 统一各个 section 的样式 */
@@ -323,4 +333,244 @@ body.modal-open {
 [data-theme="dark"] .path-preview {
     background: rgba(255, 255, 255, 0.03);
     border: 1px solid var(--lora-border);
+}
+
+/* Settings Styles */
+.settings-section {
+    margin-top: var(--space-3);
+    border-top: 1px solid var(--lora-border);
+    padding-top: var(--space-2);
+}
+
+.settings-section h3 {
+    font-size: 1.1em;
+    margin-bottom: var(--space-2);
+    color: var(--text-color);
+    opacity: 0.9;
+}
+
+.setting-item {
+    display: flex;
+    flex-direction: column; /* Changed to column for help text placement */
+    margin-bottom: var(--space-3); /* Increased to provide more spacing between items */
+    padding: var(--space-1);
+    border-radius: var(--border-radius-xs);
+}
+
+.setting-item:hover {
+    background: rgba(0, 0, 0, 0.02);
+}
+
+[data-theme="dark"] .setting-item:hover {
+    background: rgba(255, 255, 255, 0.05);
+}
+
+/* Control row with label and input together */
+.setting-row {
+    display: flex;
+    flex-direction: row;
+    justify-content: space-between;
+    align-items: center;
+    width: 100%;
+}
+
+.setting-info {
+    margin-bottom: 0;
+    width: 35%; /* Increased from 30% to prevent wrapping */
+    flex-shrink: 0; /* Prevent shrinking */
+}
+
+.setting-info label {
+    display: block;
+    font-weight: 500;
+    margin-bottom: 0;
+    white-space: nowrap; /* Prevent label wrapping */
+}
+
+.setting-control {
+    width: 60%; /* Decreased slightly from 65% */
+    margin-bottom: 0;
+    display: flex;
+    justify-content: flex-end; /* Right-align all controls */
+}
+
+/* Select Control Styles */
+.select-control {
+    width: 100%;
+    display: flex;
+    justify-content: flex-end;
+}
+
+.select-control select {
+    width: 100%;
+    max-width: 100%; /* Increased from 200px */
+    padding: 6px 10px;
+    border-radius: var(--border-radius-xs);
+    border: 1px solid var(--border-color);
+    background-color: var(--lora-surface);
+    color: var(--text-color);
+    font-size: 0.95em;
+    height: 32px;
+}
+
+/* Fix dark theme select dropdown text color */
+[data-theme="dark"] .select-control select {
+    background-color: rgba(30, 30, 30, 0.9);
+    color: var(--text-color);
+}
+
+[data-theme="dark"] .select-control select option {
+    background-color: #2d2d2d;
+    color: var(--text-color);
+}
+
+.select-control select:focus {
+    border-color: var(--lora-accent);
+    outline: none;
+}
+
+/* Toggle Switch */
+.toggle-switch {
+    position: relative;
+    display: inline-block;
+    width: 50px;
+    height: 24px;
+    cursor: pointer;
+    margin-left: auto; /* Push to right side */
+}
+
+.toggle-switch input {
+    opacity: 0;
+    width: 0;
+    height: 0;
+}
+
+.toggle-slider {
+    position: absolute;
+    top: 0;
+    left: 0;
+    right: 0;
+    bottom: 0;
+    background-color: var(--border-color);
+    transition: .3s;
+    border-radius: 24px;
+}
+
+.toggle-slider:before {
+    position: absolute;
+    content: "";
+    height: 18px;
+    width: 18px;
+    left: 3px;
+    bottom: 3px;
+    background-color: white;
+    transition: .3s;
+    border-radius: 50%;
+}
+
+input:checked + .toggle-slider {
+    background-color: var(--lora-accent);
+}
+
+input:checked + .toggle-slider:before {
+    transform: translateX(26px);
+}
+
+.toggle-label {
+    margin-left: 60px;
+    line-height: 24px;
+}
+
+/* Add small animation for the toggle */
+.toggle-slider:active:before {
+    width: 22px;
+}
+
+/* Blur effect for NSFW content */
+.nsfw-blur {
+    filter: blur(12px);
+    transition: filter 0.3s ease;
+}
+
+.nsfw-blur:hover {
+    filter: blur(8px);
+}
+
+/* Add styles for delete preview image */
+.delete-preview {
+    max-width: 150px;
+    margin: 0 auto var(--space-2);
+    overflow: hidden;
+}
+
+.delete-preview img {
+    width: 100%;
+    height: auto;
+    max-height: 150px;
+    object-fit: contain;
+    border-radius: var(--border-radius-sm);
+}
+
+.delete-info {
+    text-align: center;
+}
+
+.delete-info h3 {
+    margin-bottom: var(--space-1);
+    word-break: break-word;
+}
+
+.delete-info p {
+    margin: var(--space-1) 0;
+    font-size: 0.9em;
+    opacity: 0.8;
+}
+
+.delete-note {
+    font-size: 0.85em;
+    color: var(--text-color);
+    opacity: 0.7;
+    font-style: italic;
+    margin-top: var(--space-1);
+    text-align: center;
+}
+
+/* Add styles for markdown elements in changelog */
+.changelog-item ul {
+    padding-left: 20px;
+    margin-top: 8px;
+}
+
+.changelog-item li {
+    margin-bottom: 6px;
+    line-height: 1.4;
+}
+
+.changelog-item strong {
+    font-weight: 600;
+}
+
+.changelog-item em {
+    font-style: italic;
+}
+
+.changelog-item code {
+    background: rgba(0, 0, 0, 0.05);
+    padding: 2px 4px;
+    border-radius: 3px;
+    font-family: monospace;
+    font-size: 0.9em;
+}
+
+[data-theme="dark"] .changelog-item code {
+    background: rgba(255, 255, 255, 0.1);
+}
+
+.changelog-item a {
+    color: var(--lora-accent);
+    text-decoration: none;
+}
+
+.changelog-item a:hover {
+    text-decoration: underline;
 }

static/css/components/recipe-modal.css

@@ -0,0 +1,862 @@
+.recipe-modal-header {
+    display: flex;
+    flex-direction: column;
+    justify-content: flex-start;
+    align-items: flex-start;
+    border-bottom: 1px solid var(--lora-border);
+    padding-bottom: 10px;
+    margin-bottom: 10px;
+}
+
+.recipe-modal-header h2 {
+    font-size: 1.4em; /* Reduced from default h2 size */
+    line-height: 1.3;
+    margin: 0;
+    max-height: 2.6em; /* Limit to 2 lines */
+    overflow: hidden;
+    text-overflow: ellipsis;
+    display: -webkit-box;
+    -webkit-line-clamp: 2;
+    -webkit-box-orient: vertical;
+    width: calc(100% - 20px);
+}
+
+/* Editable content styles */
+.editable-content {
+    position: relative;
+    width: 100%;
+    display: flex;
+    align-items: center;
+    justify-content: space-between;
+}
+
+.editable-content.hide {
+    display: none;
+}
+
+.editable-content .content-text {
+    flex: 1;
+    min-width: 0;
+    overflow: hidden;
+    text-overflow: ellipsis;
+}
+
+.edit-icon {
+    background: none;
+    border: none;
+    color: var(--text-color);
+    opacity: 0;
+    cursor: pointer;
+    padding: 4px 8px;
+    margin-left: 8px;
+    border-radius: var(--border-radius-xs);
+    transition: all 0.2s;
+    flex-shrink: 0;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+}
+
+.editable-content:hover .edit-icon {
+    opacity: 0.6;
+}
+
+.edit-icon:hover {
+    opacity: 1 !important;
+    background: var(--lora-surface);
+}
+
+/* Content editor styles */
+.content-editor {
+    display: none;
+    width: 100%;
+    padding: 4px 0;
+}
+
+.content-editor.active {
+    display: flex;
+    align-items: center;
+    gap: 8px;
+}
+
+.content-editor input {
+    flex: 1;
+    background: var(--bg-color);
+    border: 1px solid var(--lora-border);
+    border-radius: var(--border-radius-xs);
+    padding: 6px 8px;
+    font-size: 1em;
+    color: var(--text-color);
+    min-width: 0;
+}
+
+.content-editor.tags-editor input {
+    font-size: 0.9em;
+}
+
+/* 删除不再需要的按钮样式 */
+.editor-actions {
+    display: none;
+}
+
+/* Special styling for tags content */
+.tags-content {
+    display: flex;
+    align-items: center;
+    flex-wrap: nowrap;
+    gap: 8px;
+}
+
+.tags-display {
+    display: flex;
+    flex-wrap: nowrap;
+    gap: 6px;
+    align-items: center;
+    flex: 1;
+    min-width: 0;
+    overflow: hidden;
+}
+
+.no-tags {
+    font-size: 0.85em;
+    color: var(--text-color);
+    opacity: 0.6;
+    font-style: italic;
+}
+
+/* Recipe Tags styles */
+.recipe-tags-container {
+    position: relative;
+    margin-top: 6px;
+    margin-bottom: 10px;
+}
+
+.recipe-tags-compact {
+    display: flex;
+    flex-wrap: nowrap;
+    gap: 6px;
+    align-items: center;
+}
+
+.recipe-tag-compact {
+    background: rgba(0, 0, 0, 0.03);
+    border: 1px solid rgba(0, 0, 0, 0.1);
+    border-radius: var(--border-radius-xs);
+    padding: 2px 8px;
+    font-size: 0.75em;
+    color: var(--text-color);
+    white-space: nowrap;
+}
+
+[data-theme="dark"] .recipe-tag-compact {
+    background: rgba(255, 255, 255, 0.03);
+    border: 1px solid var(--lora-border);
+}
+
+.recipe-tag-more {
+    background: var(--lora-accent);
+    color: var(--lora-text);
+    border-radius: var(--border-radius-xs);
+    padding: 2px 8px;
+    font-size: 0.75em;
+    cursor: pointer;
+    white-space: nowrap;
+    font-weight: 500;
+}
+
+.recipe-tags-tooltip {
+    position: absolute;
+    top: calc(100% + 8px);
+    left: 0;
+    background: var(--card-bg);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-sm);
+    box-shadow: 0 3px 8px rgba(0, 0, 0, 0.15);
+    padding: 10px 14px;
+    max-width: 400px;
+    z-index: 10;
+    opacity: 0;
+    visibility: hidden;
+    transform: translateY(-4px);
+    transition: all 0.2s ease;
+    pointer-events: none;
+}
+
+.recipe-tags-tooltip.visible {
+    opacity: 1;
+    visibility: visible;
+    transform: translateY(0);
+    pointer-events: auto;
+}
+
+.tooltip-content {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 6px;
+    max-height: 200px;
+    overflow-y: auto;
+}
+
+.tooltip-tag {
+    background: rgba(0, 0, 0, 0.03);
+    border: 1px solid rgba(0, 0, 0, 0.1);
+    border-radius: var(--border-radius-xs);
+    padding: 3px 8px;
+    font-size: 0.75em;
+    color: var(--text-color);
+}
+
+[data-theme="dark"] .tooltip-tag {
+    background: rgba(255, 255, 255, 0.03);
+    border: 1px solid var(--lora-border);
+}
+
+/* Top Section: Preview and Gen Params */
+.recipe-top-section {
+    display: grid;
+    grid-template-columns: 280px 1fr;
+    gap: var(--space-2);
+    flex-shrink: 0;
+    margin-bottom: var(--space-2);
+}
+
+/* Recipe Preview */
+.recipe-preview-container {
+    width: 100%;
+    height: 360px;
+    border-radius: var(--border-radius-sm);
+    overflow: hidden;
+    background: var(--lora-surface);
+    border: 1px solid var(--border-color);
+    display: flex;
+    align-items: center;
+    justify-content: center;
+}
+
+.recipe-preview-container img,
+.recipe-preview-container video {
+    max-width: 100%;
+    max-height: 100%;
+    object-fit: contain;
+}
+
+.recipe-preview-media {
+    max-width: 100%;
+    max-height: 100%;
+    object-fit: contain;
+}
+
+/* Generation Parameters */
+.recipe-gen-params {
+    height: 360px;
+    display: flex;
+    flex-direction: column;
+}
+
+.recipe-gen-params h3 {
+    margin-top: 0;
+    margin-bottom: var(--space-2);
+    font-size: 1.2em;
+    color: var(--text-color);
+    padding-bottom: var(--space-1);
+    border-bottom: 1px solid var(--border-color);
+    flex-shrink: 0;
+}
+
+.gen-params-container {
+    display: flex;
+    flex-direction: column;
+    gap: var(--space-2);
+    overflow-y: auto;
+    flex: 1;
+}
+
+.param-group {
+    display: flex;
+    flex-direction: column;
+    gap: 8px;
+}
+
+.param-header {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+}
+
+.param-header label {
+    font-weight: 500;
+    color: var(--text-color);
+}
+
+.copy-btn {
+    background: none;
+    border: none;
+    color: var(--text-color);
+    opacity: 0.6;
+    cursor: pointer;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    transition: all 0.2s;
+}
+
+.copy-btn:hover {
+    opacity: 1;
+    background: var(--lora-surface);
+}
+
+.param-content {
+    background: var(--lora-surface);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    padding: var(--space-2);
+    color: var(--text-color);
+    font-size: 0.9em;
+    line-height: 1.5;
+    max-height: 150px;
+    overflow-y: auto;
+    white-space: pre-wrap;
+    word-break: break-word;
+}
+
+/* Other Parameters */
+.other-params {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px;
+    margin-top: var(--space-1);
+}
+
+.param-tag {
+    background: var(--lora-surface);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    padding: 4px 8px;
+    font-size: 0.85em;
+    color: var(--text-color);
+    display: flex;
+    align-items: center;
+    gap: 6px;
+}
+
+.param-tag .param-name {
+    font-weight: 500;
+    opacity: 0.8;
+}
+
+/* Bottom Section: Resources */
+.recipe-bottom-section {
+    max-height: 320px;
+    display: flex;
+    flex-direction: column;
+    border-top: 1px solid var(--border-color);
+    padding-top: var(--space-2);
+}
+
+.recipe-section-header {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    margin-bottom: var(--space-2);
+    padding-bottom: var(--space-1);
+    border-bottom: 1px solid var(--border-color);
+    flex-shrink: 0;
+}
+
+.recipe-section-header h3 {
+    margin: 0;
+    font-size: 1.2em;
+    color: var(--text-color);
+    display: flex;
+    align-items: center;
+    gap: 8px;
+}
+
+.recipe-status {
+    display: inline-flex;
+    align-items: center;
+    font-size: 0.85em;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    margin-left: var(--space-1);
+}
+
+.recipe-status.ready {
+    background: oklch(var(--lora-accent) / 0.1);
+    color: var(--lora-accent);
+}
+
+.recipe-status.missing {
+    background: oklch(var(--lora-error) / 0.1);
+    color: var(--lora-error);
+}
+
+.recipe-status i {
+    margin-right: 4px;
+}
+
+.recipe-section-actions {
+    display: flex;
+    align-items: center;
+    gap: var(--space-1);
+}
+
+/* View LoRAs button */
+.view-loras-btn {
+    background: none;
+    border: none;
+    color: var(--text-color);
+    opacity: 0.7;
+    cursor: pointer;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    transition: all 0.2s;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+}
+
+.view-loras-btn:hover {
+    opacity: 1;
+    background: var(--lora-surface);
+    color: var(--lora-accent);
+}
+
+#recipeLorasCount {
+    font-size: 0.9em;
+    color: var(--text-color);
+    opacity: 0.8;
+    display: flex;
+    align-items: center;
+    gap: 6px;
+}
+
+#recipeLorasCount i {
+    font-size: 1em;
+}
+
+/* LoRAs List */
+.recipe-loras-list {
+    display: flex;
+    flex-direction: column;
+    gap: 10px;
+    overflow-y: auto;
+    flex: 1;
+    padding-top: 4px; /* Add padding to prevent first item from being cut off when hovered */
+}
+
+.recipe-lora-item {
+    display: flex;
+    gap: var(--space-2);
+    padding: 10px var(--space-2);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-sm);
+    background: var(--bg-color);
+    /* Add will-change to create a new stacking context and force hardware acceleration */
+    will-change: transform;
+    /* Create a new containing block for absolutely positioned descendants */
+    transform: translateZ(0);
+    cursor: pointer; /* Make it clear the item is clickable */
+    transition: transform 0.2s ease, box-shadow 0.2s ease, border-color 0.2s ease;
+}
+
+.recipe-lora-item:hover {
+    transform: translateY(-1px);
+    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.08);
+    border-color: var(--lora-accent);
+}
+
+.recipe-lora-item.exists-locally {
+    background: oklch(var(--lora-accent) / 0.05);
+    border-left: 4px solid var(--lora-accent);
+}
+
+.recipe-lora-item.missing-locally {
+    border-left: 4px solid var(--lora-error);
+}
+
+.recipe-lora-item.is-deleted {
+    background: rgba(127, 127, 127, 0.05);
+    border-left: 4px solid #777;
+    opacity: 0.8;
+}
+
+.recipe-lora-thumbnail {
+    width: 46px;
+    height: 46px;
+    flex-shrink: 0;
+    border-radius: var(--border-radius-xs);
+    overflow: hidden;
+    background: var(--bg-color);
+    display: flex;
+    align-items: center;
+    justify-content: center;
+}
+
+.recipe-lora-thumbnail img,
+.recipe-lora-thumbnail video {
+    width: 100%;
+    height: 100%;
+    object-fit: cover;
+}
+
+.thumbnail-video {
+    width: 100%;
+    height: 100%;
+    object-fit: cover;
+}
+
+.recipe-lora-content {
+    display: flex;
+    flex-direction: column;
+    gap: 3px;
+    flex: 1;
+    min-width: 0;
+}
+
+.recipe-lora-header {
+    display: flex;
+    align-items: flex-start;
+    justify-content: space-between;
+    gap: var(--space-2);
+    position: relative;
+    min-height: 28px;
+    /* Ensure badges don't move during scroll in Chrome */
+    transform: translateZ(0);
+}
+
+.recipe-lora-content h4 {
+    margin: 0;
+    font-size: 1em;
+    color: var(--text-color);
+    flex: 1;
+    max-width: calc(100% - 120px); /* Make room for the badge */
+    overflow: hidden;
+    text-overflow: ellipsis;
+    display: -webkit-box;
+    -webkit-line-clamp: 2; /* Limit to 2 lines */
+    -webkit-box-orient: vertical;
+    line-height: 1.3;
+}
+
+.recipe-lora-info {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px;
+    align-items: center;
+    font-size: 0.85em;
+    margin-top: 4px;
+    padding-right: 4px;
+}
+
+.recipe-lora-info .base-model {
+    background: oklch(var(--lora-accent) / 0.1);
+    color: var(--lora-accent);
+    padding: 2px 8px;
+    border-radius: var(--border-radius-xs);
+}
+
+.recipe-lora-version {
+    font-size: 0.85em;
+    color: var(--text-color);
+    opacity: 0.7;
+}
+
+.recipe-lora-weight {
+    background: var(--lora-surface);
+    padding: 2px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.85em;
+    color: var(--lora-accent);
+}
+
+.local-badge, 
+.missing-badge {
+    position: absolute;
+    right: 0;
+    top: 0;
+    /* Force hardware acceleration for Chrome */
+    transform: translateZ(0);
+    backface-visibility: hidden;
+}
+
+/* Specific styles for recipe modal badges - update z-index */
+.recipe-lora-header .local-badge,
+.recipe-lora-header .missing-badge {
+    z-index: 2; /* Ensure the badge is above other elements */
+    backface-visibility: hidden;
+}
+
+/* Ensure local-path tooltip is properly positioned and won't move during scroll */
+.recipe-lora-header .local-badge .local-path {
+    z-index: 3;
+    top: calc(100% + 4px); /* Position tooltip below the badge */
+    right: -4px; /* Align with the badge */
+    max-width: 250px;
+    /* Force hardware acceleration for Chrome */
+    transform: translateZ(0);
+}
+
+.missing-badge {
+    display: inline-flex;
+    align-items: center;
+    background: var(--lora-error);
+    color: white;
+    padding: 3px 6px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.75em;
+    font-weight: 500;
+    white-space: nowrap;
+    flex-shrink: 0;
+}
+
+.missing-badge i {
+    margin-right: 4px;
+    font-size: 0.9em;
+}
+
+/* Deleted badge with reconnect functionality */
+.deleted-badge {
+    display: inline-flex;
+    align-items: center;
+    background: #777;
+    color: white;
+    padding: 3px 6px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.75em;
+    font-weight: 500;
+    white-space: nowrap;
+    flex-shrink: 0;
+}
+
+.deleted-badge i {
+    margin-right: 4px;
+    font-size: 0.9em;
+}
+
+/* Add reconnect functionality styles */
+.deleted-badge.reconnectable {
+    position: relative;
+    cursor: pointer;
+    transition: background-color 0.2s ease;
+}
+
+.deleted-badge.reconnectable:hover {
+    background-color: var(--lora-accent);
+}
+
+.deleted-badge .reconnect-tooltip {
+    position: absolute;
+    display: none;
+    background-color: var(--card-bg);
+    color: var(--text-color);
+    padding: 8px 12px;
+    border-radius: var(--border-radius-xs);
+    border: 1px solid var(--border-color);
+    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
+    z-index: var(--z-overlay);
+    width: max-content;
+    max-width: 200px;
+    font-size: 0.85rem;
+    font-weight: normal;
+    top: calc(100% + 5px);
+    left: 0;
+    margin-left: -100px;
+}
+
+.deleted-badge.reconnectable:hover .reconnect-tooltip {
+    display: block;
+}
+
+/* LoRA reconnect container */
+.lora-reconnect-container {
+    display: none;
+    flex-direction: column;
+    background: var(--lora-surface);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    padding: 12px;
+    margin-top: 10px;
+    gap: 10px;
+}
+
+.lora-reconnect-container.active {
+    display: flex;
+}
+
+.reconnect-instructions {
+    display: flex;
+    flex-direction: column;
+    gap: 5px;
+}
+
+.reconnect-instructions p {
+    margin: 0;
+    font-size: 0.95em;
+    font-weight: 500;
+    color: var(--text-color);
+}
+
+.reconnect-instructions small {
+    color: var(--text-color);
+    opacity: 0.7;
+    font-size: 0.85em;
+}
+
+.reconnect-instructions code {
+    background: rgba(0, 0, 0, 0.1);
+    padding: 2px 4px;
+    border-radius: 3px;
+    font-family: monospace;
+    font-size: 0.9em;
+}
+
+[data-theme="dark"] .reconnect-instructions code {
+    background: rgba(255, 255, 255, 0.1);
+}
+
+.reconnect-form {
+    display: flex;
+    flex-direction: column;
+    gap: 10px;
+}
+
+.reconnect-input {
+    width: calc(100% - 20px);
+    padding: 8px 10px;
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    background: var(--bg-color);
+    color: var(--text-color);
+    font-size: 0.9em;
+}
+
+.reconnect-actions {
+    display: flex;
+    justify-content: flex-end;
+    gap: 8px;
+}
+
+.reconnect-cancel-btn, 
+.reconnect-confirm-btn {
+    padding: 6px 12px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.85em;
+    cursor: pointer;
+    border: none;
+    transition: all 0.2s;
+}
+
+.reconnect-cancel-btn {
+    background: var(--bg-color);
+    color: var(--text-color);
+    border: 1px solid var(--border-color);
+}
+
+.reconnect-confirm-btn {
+    background: var(--lora-accent);
+    color: white;
+}
+
+.reconnect-cancel-btn:hover {
+    background: var(--lora-surface);
+}
+
+.reconnect-confirm-btn:hover {
+    background: color-mix(in oklch, var(--lora-accent), black 10%);
+}
+
+/* Recipe status partial state */
+.recipe-status.partial {
+    background: rgba(127, 127, 127, 0.1);
+    color: #777;
+}
+
+/* 标题输入框特定的样式 */
+.title-input {
+    font-size: 1.2em !important; /* 调整为更合适的大小 */
+    line-height: 1.2;
+    font-weight: 500;
+}
+
+/* Responsive adjustments */
+@media (max-width: 768px) {
+    .recipe-top-section {
+        grid-template-columns: 1fr;
+    }
+    
+    .recipe-preview-container {
+        height: 200px;
+    }
+    
+    .recipe-gen-params {
+        height: auto;
+        max-height: 300px;
+    }
+}
+
+.badge-container {
+    position: relative;
+    display: flex;
+    align-items: center;
+    justify-content: flex-end;
+    flex-shrink: 0;
+    min-width: 110px;
+    z-index: 2;
+}
+
+/* Update the local-badge and missing-badge to be positioned within the badge-container */
+.badge-container .local-badge, 
+.badge-container .missing-badge,
+.badge-container .deleted-badge {
+    position: static; /* Override absolute positioning */
+    transform: none; /* Remove the transform */
+}
+
+/* Ensure the tooltip is still properly positioned */
+.badge-container .local-badge .local-path {
+    position: fixed; /* Keep as fixed for Chrome */
+    z-index: 100;
+}
+
+/* Add styles for missing LoRAs download feature */
+.recipe-status.missing {
+    position: relative;
+    cursor: pointer;
+    transition: background-color 0.2s ease;
+}
+
+.recipe-status.missing:hover {
+    background-color: rgba(var(--lora-warning-rgb, 255, 165, 0), 0.2);
+}
+
+.recipe-status.missing .missing-tooltip {
+    position: absolute;
+    display: none;
+    background-color: var(--card-bg);
+    color: var(--text-color);
+    padding: 8px 12px;
+    border-radius: var(--border-radius-xs);
+    border: 1px solid var(--border-color);
+    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
+    z-index: var(--z-overlay);
+    width: max-content;
+    max-width: 200px;
+    font-size: 0.85rem;
+    font-weight: normal;
+    margin-left: -100px;
+    margin-top: -65px;
+}
+
+.recipe-status.missing:hover .missing-tooltip {
+    display: block;
+}
+
+.recipe-status.clickable {
+    cursor: pointer;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+}
+
+.recipe-status.clickable:hover {
+    background-color: rgba(var(--lora-warning-rgb, 255, 165, 0), 0.2);
+}

static/css/components/search-filter.css

@@ -1,9 +1,7 @@
 /* Search Container Styles */
 .search-container {
     position: relative;
-    width: 250px;
-    margin-left: auto;
-    flex-shrink: 0; /* 防止搜索框被压缩 */
+    width: 100%;
     display: flex;
     align-items: center;
     gap: 4px;
@@ -12,14 +10,14 @@
 /* 调整搜索框样式以匹配其他控件 */
 .search-container input {
     width: 100%;
-    padding: 6px 75px 6px 12px; /* Increased right padding to accommodate both buttons */
-    border: 1px solid oklch(65% 0.02 256); /* 更深的边框颜色，提高对比度 */
+    padding: 6px 35px 6px 12px; /* Reduced right padding */
+    border: 1px solid oklch(65% 0.02 256);
     border-radius: var(--border-radius-sm);
     background: var(--lora-surface);
     color: var(--text-color);
     font-size: 0.9em;
     height: 32px;
-    box-sizing: border-box; /* 确保padding不会增加总宽度 */
+    box-sizing: border-box;
 }
 
 .search-container input:focus {
@@ -34,7 +32,7 @@
     transform: translateY(-50%);
     color: oklch(var(--text-color) / 0.5);
     pointer-events: none;
-    line-height: 1; /* 防止图标影响容器高度 */
+    line-height: 1;
 }
 
 /* 修改清空按钮样式 */
@@ -47,8 +45,8 @@
     cursor: pointer;
     border: none;
     background: none;
-    padding: 4px 8px; /* 增加点击区域 */
-    display: none; /* 默认隐藏 */
+    padding: 4px 8px;
+    display: none;
     line-height: 1;
     transition: color 0.2s ease;
 }
@@ -144,19 +142,19 @@
 
 /* Filter Panel Styles */
 .filter-panel {
-    position: absolute;
-    top: 140px; /* Adjust to be closer to the filter button */
+    position: fixed;
     right: 20px;
-    width: 300px;
+    top: 50px; /* Position below header */
+    width: 320px;
     background-color: var(--card-bg);
     border: 1px solid var(--border-color);
     border-radius: var(--border-radius-base);
     box-shadow: 0 4px 10px rgba(0, 0, 0, 0.1);
-    z-index: var(--z-overlay); /* Increase z-index to be above cards */
+    z-index: var(--z-overlay);
     padding: 16px;
     transition: transform 0.3s ease, opacity 0.3s ease;
     transform-origin: top right;
-    max-height: calc(100vh - 160px);
+    max-height: calc(100vh - 70px); /* Adjusted for header height */
     overflow-y: auto;
 }
 
@@ -237,6 +235,44 @@
     border-color: var(--lora-accent);
 }
 
+/* Tag filter styles */
+.tag-filter {
+    display: flex;
+    align-items: center;
+    justify-content: space-between;
+    min-width: 60px;
+}
+
+.tag-count {
+    background: rgba(0, 0, 0, 0.1);
+    padding: 1px 6px;
+    border-radius: 10px;
+    font-size: 0.8em;
+    margin-left: 4px;
+}
+
+[data-theme="dark"] .tag-count {
+    background: rgba(255, 255, 255, 0.1);
+}
+
+.tag-filter.active .tag-count {
+    background: rgba(255, 255, 255, 0.3);
+    color: white;
+}
+
+.tags-loading, .tags-error, .no-tags {
+    width: 100%;
+    padding: 8px;
+    text-align: center;
+    font-size: 0.9em;
+    color: var(--text-color);
+    opacity: 0.7;
+}
+
+.tags-error {
+    color: var(--lora-error);
+}
+
 /* Filter actions */
 .filter-actions {
     display: flex;
@@ -274,6 +310,211 @@
         width: calc(100% - 40px);
         left: 20px;
         right: 20px;
-        top: 140px;
+        top: 160px; /* Adjusted for mobile layout */
     }
-} 
+}
+
+/* Search Options Toggle */
+.search-options-toggle {
+    background: var(--lora-surface);
+    border: 1px solid oklch(65% 0.02 256);
+    border-radius: var(--border-radius-sm);
+    color: var(--text-color);
+    width: 32px;
+    height: 32px;
+    padding: 0;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    cursor: pointer;
+    transition: all 0.2s ease;
+    flex-shrink: 0;
+}
+
+.search-options-toggle:hover {
+    background-color: var(--lora-surface-hover, oklch(95% 0.02 256));
+    color: var(--lora-accent);
+    border-color: var(--lora-accent);
+}
+
+.search-options-toggle.active {
+    background-color: oklch(95% 0.05 256);
+    color: var(--lora-accent);
+    border-color: var(--lora-accent);
+}
+
+.search-options-toggle i {
+    font-size: 0.9em;
+}
+
+/* Search Options Panel */
+.search-options-panel {
+    position: fixed;
+    right: 20px;
+    top: 50px; /* Position below header */
+    width: 280px;
+    background-color: var(--card-bg);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-base);
+    box-shadow: 0 4px 10px rgba(0, 0, 0, 0.1);
+    z-index: var(--z-overlay);
+    padding: 16px;
+    transition: transform 0.3s ease, opacity 0.3s ease;
+    transform-origin: top right;
+    display: block; /* Ensure it's block by default */
+}
+
+.search-options-panel.hidden {
+    opacity: 0;
+    transform: scale(0.95);
+    pointer-events: none;
+}
+
+.options-header {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    margin-bottom: 16px;
+}
+
+.options-header h3 {
+    margin: 0;
+    font-size: 16px;
+    color: var(--text-color);
+}
+
+.close-options-btn {
+    background: none;
+    border: none;
+    color: var(--text-color);
+    cursor: pointer;
+    font-size: 16px;
+    padding: 4px;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+}
+
+.close-options-btn:hover {
+    color: var(--lora-accent);
+}
+
+.options-section {
+    margin-bottom: 16px;
+}
+
+.options-section h4 {
+    margin: 0 0 8px 0;
+    font-size: 14px;
+    color: var(--text-color);
+    opacity: 0.8;
+}
+
+.search-option-tags {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 8px; /* Increased gap for better spacing */
+}
+
+.search-option-tag {
+    padding: 6px 8px; /* Adjusted padding for better text display */
+    border-radius: var(--border-radius-sm);
+    background-color: var(--lora-surface);
+    border: 1px solid var(--border-color);
+    color: var(--text-color);
+    font-size: 13px; /* Slightly smaller font size */
+    cursor: pointer;
+    transition: all 0.2s ease;
+    user-select: none;
+    flex: 1;
+    text-align: center;
+    white-space: nowrap; /* Prevent text wrapping */
+    min-width: 80px; /* Ensure minimum width for each tag */
+    display: inline-flex; /* Better control over layout */
+    justify-content: center;
+    align-items: center;
+}
+
+.search-option-tag:hover {
+    background-color: var(--lora-surface-hover);
+}
+
+.search-option-tag.active {
+    background-color: var(--lora-accent);
+    color: white;
+    border-color: var(--lora-accent);
+}
+
+/* Switch styles */
+.search-option-switch {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    padding: 4px 0;
+}
+
+.switch {
+    position: relative;
+    display: inline-block;
+    width: 46px;
+    height: 24px;
+}
+
+.switch input {
+    opacity: 0;
+    width: 0;
+    height: 0;
+}
+
+.slider {
+    position: absolute;
+    cursor: pointer;
+    top: 0;
+    left: 0;
+    right: 0;
+    bottom: 0;
+    background-color: #ccc;
+    transition: .4s;
+}
+
+.slider:before {
+    position: absolute;
+    content: "";
+    height: 18px;
+    width: 18px;
+    left: 3px;
+    bottom: 3px;
+    background-color: white;
+    transition: .4s;
+}
+
+input:checked + .slider {
+    background-color: var(--lora-accent);
+}
+
+input:focus + .slider {
+    box-shadow: 0 0 1px var(--lora-accent);
+}
+
+input:checked + .slider:before {
+    transform: translateX(22px);
+}
+
+.slider.round {
+    border-radius: 34px;
+}
+
+.slider.round:before {
+    border-radius: 50%;
+}
+
+/* Mobile adjustments */
+@media (max-width: 768px) {
+    .search-options-panel,
+    .filter-panel {
+        width: calc(100% - 40px);
+        left: 20px;
+        right: 20px;
+        top: 160px; /* Adjusted for mobile layout */
+    }
+}

static/css/components/shared.css

@@ -0,0 +1,111 @@
+/* Local Version Badge */
+.local-badge {
+    display: inline-flex;
+    align-items: center;
+    background: var(--lora-accent);
+    color: var(--lora-text);
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.8em;
+    font-weight: 500;
+    white-space: nowrap;
+    flex-shrink: 0;
+    position: relative;
+    /* Force hardware acceleration to prevent Chrome scroll issues */
+    transform: translateZ(0);
+    will-change: transform;
+}
+
+.local-badge i {
+    margin-right: 4px;
+    font-size: 0.9em;
+}
+
+/* Early Access Badge */
+.early-access-badge {
+    display: inline-flex;
+    align-items: center;
+    background: #00B87A; /* Green for early access */
+    color: white;
+    padding: 4px 8px;
+    border-radius: var(--border-radius-xs);
+    font-size: 0.8em;
+    font-weight: 500;
+    white-space: nowrap;
+    flex-shrink: 0;
+    position: relative;
+    /* Force hardware acceleration to prevent Chrome scroll issues */
+    transform: translateZ(0);
+    will-change: transform;
+}
+
+.early-access-badge i {
+    margin-right: 4px;
+    font-size: 0.9em;
+}
+
+.early-access-info {
+    display: none;
+    position: absolute;
+    top: 100%;
+    right: 0;
+    background: var(--card-bg);
+    border: 1px solid #00B87A;
+    border-radius: var(--border-radius-xs);
+    padding: var(--space-1);
+    margin-top: 4px;
+    font-size: 0.9em;
+    color: var(--text-color);
+    white-space: normal;
+    word-break: break-all;
+    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
+    z-index: 100; /* Higher z-index to ensure it's above other elements */
+    min-width: 300px;
+    max-width: 300px;
+    /* Create a separate layer with hardware acceleration */
+    transform: translateZ(0);
+    /* Use a fixed position to ensure it's in a separate layer from scrollable content */
+    position: fixed;
+    pointer-events: none; /* Don't block mouse events */
+}
+
+.early-access-badge:hover .early-access-info {
+    display: block;
+    pointer-events: auto; /* Allow interaction with the tooltip when visible */
+}
+
+.local-path {
+    display: none;
+    position: absolute;
+    top: 100%;
+    right: 0;
+    background: var(--card-bg);
+    border: 1px solid var(--border-color);
+    border-radius: var(--border-radius-xs);
+    padding: var(--space-1);
+    margin-top: 4px;
+    font-size: 0.9em;
+    color: var(--text-color);
+    white-space: normal;
+    word-break: break-all;
+    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
+    z-index: 100; /* Higher z-index to ensure it's above other elements */
+    min-width: 200px;
+    max-width: 300px;
+    /* Create a separate layer with hardware acceleration */
+    transform: translateZ(0);
+    /* Use a fixed position to ensure it's in a separate layer from scrollable content */
+    position: fixed;
+    pointer-events: none; /* Don't block mouse events */
+}
+
+.local-badge:hover .local-path {
+    display: block;
+    pointer-events: auto; /* Allow interaction with the tooltip when visible */
+}
+
+.error-message {
+    color: var(--lora-error);
+    font-size: 0.9em;
+    margin-top: 4px;
+}

static/css/components/support-modal.css

@@ -1,6 +1,6 @@
 /* Support Modal Styles */
 .support-modal {
-    max-width: 550px;
+    max-width: 570px;
 }
 
 .support-header {
@@ -141,7 +141,7 @@
 
 .support-toggle:hover {
     background: var(--lora-accent);
-    color: white;
+    color: var(--lora-error) !important;
     transform: translateY(-2px);
 }
 

static/css/components/toast.css

@@ -120,4 +120,63 @@
 
 .tooltip:hover::after {
     opacity: 1;
+}
+
+/* Toast Container for stacked notifications */
+.toast-container {
+    position: fixed;
+    top: 0;
+    right: 0;
+    z-index: calc(var(--z-overlay) + 10);
+    display: flex;
+    flex-direction: column;
+    gap: 10px;
+    padding: 20px;
+    pointer-events: none; /* Allow clicking through the container */
+    width: 400px;
+    max-width: 100%;
+}
+
+/* Ensure each toast has pointer events */
+.toast-container .toast {
+    pointer-events: auto;
+    position: relative; /* Override fixed positioning */
+    top: 0 !important; /* Let the container handle positioning */
+    right: 0 !important;
+    margin-bottom: 10px;
+}
+
+/* Add missing warning toast style */
+.toast-warning {
+    border-left: 4px solid var(--lora-warning);
+}
+
+.toast-warning::before {
+    background-image: url("data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 24 24' fill='%23ff9800'%3E%3Cpath d='M1 21h22L12 2 1 21zm12-3h-2v-2h2v2zm0-4h-2v-4h2v4z'/%3E%3C/svg%3E");
+}
+
+/* Improve toast animation */
+.toast {
+    transform: translateX(120%);
+    opacity: 0;
+    transition: transform 0.3s cubic-bezier(0.4, 0, 0.2, 1), 
+                opacity 0.3s cubic-bezier(0.4, 0, 0.2, 1);
+}
+
+.toast.show {
+    transform: translateX(0);
+    opacity: 1;
+}
+
+/* Responsive adjustments */
+@media (max-width: 480px) {
+    .toast-container {
+        width: 100%;
+        padding: 10px;
+    }
+    
+    .toast {
+        width: 100%;
+        max-width: none;
+    }
 }

static/css/components/update-modal.css

@@ -153,56 +153,43 @@
     border-top: 1px solid var(--lora-border);
     margin-top: var(--space-2);
     padding-top: var(--space-2);
-}
-
-/* Toggle switch styles */
-.toggle-switch {
     display: flex;
     align-items: center;
-    gap: 12px;
+    justify-content: flex-start;
+}
+
+/* Override toggle switch styles for update preferences */
+.update-preferences .toggle-switch {
+    position: relative;
+    display: inline-flex;
+    align-items: center;
+    width: auto;
+    height: 24px;
     cursor: pointer;
-    user-select: none;
 }
 
-.toggle-switch input {
-    opacity: 0;
-    width: 0;
-    height: 0;
-    position: absolute;
-}
-
-.toggle-slider {
+.update-preferences .toggle-slider {
     position: relative;
     display: inline-block;
-    width: 40px;
-    height: 20px;
-    background-color: var(--border-color);
-    border-radius: 20px;
-    transition: .4s;
+    width: 50px;
+    height: 24px;
     flex-shrink: 0;
+    margin-right: 10px;
 }
 
-.toggle-slider:before {
-    position: absolute;
-    content: "";
-    height: 16px;
-    width: 16px;
-    left: 2px;
-    bottom: 2px;
-    background-color: white;
-    border-radius: 50%;
-    transition: .4s;
+.update-preferences .toggle-label {
+    margin-left: 0;
+    white-space: nowrap;
+    line-height: 24px;
 }
 
-input:checked + .toggle-slider {
-    background-color: var(--lora-accent);
-}
-
-input:checked + .toggle-slider:before {
-    transform: translateX(20px);
-}
-
-.toggle-label {
-    font-size: 0.9em;
-    color: var(--text-color);
+@media (max-width: 480px) {
+    .update-preferences {
+        flex-direction: row;
+        flex-wrap: wrap;
+    }
+    
+    .update-preferences .toggle-label {
+        margin-top: 5px;
+    }
 }

static/css/layout.css

@@ -1,7 +1,18 @@
+.page-content {
+    height: calc(100vh - 48px); /* Full height minus header */
+    margin-top: 48px; /* Push down below header */
+    overflow-y: auto; /* Enable scrolling here */
+    width: 100%;
+    position: relative;
+    overflow-y: scroll;
+}
+
 .container {
     max-width: 1400px;
     margin: 20px auto;
     padding: 0 15px;
+    position: relative;
+    z-index: var(--z-base);
 }
 
 .controls {
@@ -14,69 +25,101 @@
 .actions {
     display: flex;
     align-items: center;
+    justify-content: space-between;
     gap: var(--space-2);
     flex-wrap: nowrap;
     width: 100%;
 }
 
-/* Search and filter styles moved to components/search-filter.css */
-
-/* Update corner-controls for collapsible behavior */
-.corner-controls {
-    position: fixed;
-    top: 20px;
-    right: 20px;
-    z-index: var(--z-overlay);
+.action-buttons {
     display: flex;
-    flex-direction: column;
     align-items: center;
-    transition: all 0.3s ease;
+    gap: var(--space-2);
+    flex-wrap: nowrap;
 }
 
-.corner-controls-toggle {
-    width: 36px;
-    height: 36px;
-    border-radius: 50%;
-    background: var(--card-bg);
-    border: 1px solid var(--border-color);
-    color: var(--text-color);
+/* Action button styling */
+.control-group {
+    position: relative;
+}
+
+.control-group button {
+    min-width: 100px;
     display: flex;
     align-items: center;
     justify-content: center;
-    cursor: pointer;
+    gap: 4px;
+    border-radius: var(--border-radius-xs);
+    padding: 4px 10px;
+    border: 1px solid var(--border-color);
+    background: var(--card-bg);
+    color: var(--text-color);
+    font-size: 0.85em;
     transition: all 0.2s ease;
-    z-index: 2;
-    margin-bottom: 10px;
+    cursor: pointer;
+    box-shadow: 0 1px 2px rgba(0, 0, 0, 0.05);
 }
 
-.corner-controls-toggle:hover {
+.control-group button:hover {
+    border-color: var(--lora-accent);
+    background: var(--bg-color);
+    transform: translateY(-1px);
+    box-shadow: 0 3px 5px rgba(0, 0, 0, 0.08);
+}
+
+.control-group button:active {
+    transform: translateY(0);
+    box-shadow: 0 1px 2px rgba(0, 0, 0, 0.05);
+}
+
+.control-group button i {
+    opacity: 0.8;
+    transition: opacity 0.2s ease;
+}
+
+.control-group button:hover i {
+    opacity: 1;
+}
+
+/* Active state for buttons that can be toggled */
+.control-group button.active {
     background: var(--lora-accent);
     color: white;
-    transform: translateY(-2px);
+    border-color: var(--lora-accent);
 }
 
-.corner-controls-items {
-    display: flex;
-    flex-direction: column;
-    gap: 10px;
-    opacity: 0;
-    transform: translateY(-10px) scale(0.9);
-    transition: all 0.3s ease;
-    pointer-events: none;
+/* Select dropdown styling */
+.control-group select {
+    min-width: 100px;
+    padding: 4px 26px 4px 10px;
+    border-radius: var(--border-radius-xs);
+    border: 1px solid var(--border-color);
+    background-color: var(--card-bg);
+    color: var(--text-color);
+    font-size: 0.85em;
+    appearance: none;
+    -webkit-appearance: none;
+    -moz-appearance: none;
+    background-image: url("data:image/svg+xml;charset=UTF-8,%3csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 24 24' fill='none' stroke='currentColor' stroke-width='2' stroke-linecap='round' stroke-linejoin='round'%3e%3cpolyline points='6 9 12 15 18 9'%3e%3c/polyline%3e%3c/svg%3e");
+    background-repeat: no-repeat;
+    background-position: right 6px center;
+    background-size: 14px;
+    cursor: pointer;
+    transition: all 0.2s ease;
+    box-shadow: 0 1px 2px rgba(0, 0, 0, 0.05);
 }
 
-/* Expanded state */
-.corner-controls.expanded .corner-controls-items {
-    opacity: 1;
-    transform: translateY(0) scale(1);
-    pointer-events: all;
+.control-group select:hover {
+    border-color: var(--lora-accent);
+    background-color: var(--bg-color);
+    transform: translateY(-1px);
+    box-shadow: 0 3px 5px rgba(0, 0, 0, 0.08);
 }
 
-/* Expanded state - only expand on hover if not already expanded by click */
-.corner-controls:hover:not(.expanded) .corner-controls-items {
-    opacity: 1;
-    transform: translateY(0) scale(1);
-    pointer-events: all;
+.control-group select:focus {
+    outline: none;
+    border-color: var(--lora-accent);
+    box-shadow: 0 0 0 2px oklch(var(--lora-accent) / 0.15);
 }
 
 /* Ensure hidden class works properly */
@@ -84,46 +127,6 @@
     display: none !important;
 }
 
-/* Update toggle button styles */
-.update-toggle {
-    width: 36px;
-    height: 36px;
-    border-radius: 50%;
-    background: var(--card-bg);
-    border: 1px solid var(--border-color);
-    color: var(--text-color); /* Changed from var(--lora-accent) to match other toggles */
-    display: flex;
-    align-items: center;
-    justify-content: center;
-    cursor: pointer;
-    transition: all 0.2s ease;
-    position: relative;
-}
-
-.update-toggle:hover {
-    background: var(--lora-accent);
-    color: white;
-    transform: translateY(-2px);
-}
-
-/* Update badge styles */
-.update-badge {
-    position: absolute;
-    top: -3px;
-    right: -3px;
-    background-color: var(--lora-error);
-    width: 8px;
-    height: 8px;
-    border-radius: 50%;
-    box-shadow: 0 0 0 2px var(--card-bg);
-}
-
-/* Badge on corner toggle */
-.corner-badge {
-    top: 0;
-    right: 0;
-}
-
 .folder-tags-container {
     position: relative;
     width: 100%;
@@ -131,11 +134,14 @@
 }
 
 .folder-tags {
+    display: flex;
+    gap: 4px;
+    padding: 2px 0;
+    flex-wrap: wrap;
     transition: max-height 0.3s ease, opacity 0.2s ease;
     max-height: 150px; /* Limit height to prevent overflow */
     opacity: 1;
     overflow-y: auto; /* Enable vertical scrolling */
-    padding-right: 40px; /* Make space for the toggle button */
     margin-bottom: 5px; /* Add margin below the tags */
 }
 
@@ -144,13 +150,15 @@
     opacity: 0;
     margin: 0;
     padding-bottom: 0;
+    overflow: hidden;
+}
+
+.toggle-folders-container {
+    margin-left: auto;
 }
 
 /* Toggle Folders Button */
 .toggle-folders-btn {
-    position: absolute;
-    bottom: 0;      /* 固定在容器底部 */
-    right: 0;       /* 固定在容器右侧 */
     width: 36px;
     height: 36px;
     border-radius: 50%;
@@ -162,38 +170,33 @@
     justify-content: center;
     cursor: pointer;
     transition: all 0.3s ease;
-    z-index: 2;
+    box-shadow: 0 1px 2px rgba(0, 0, 0, 0.05);
 }
 
 .toggle-folders-btn:hover {
     background: var(--lora-accent);
     color: white;
     transform: translateY(-2px);
+    box-shadow: 0 3px 6px rgba(0, 0, 0, 0.1);
 }
 
 .toggle-folders-btn i {
     transition: transform 0.3s ease;
 }
 
-/* 折叠状态样式 */
-.folder-tags.collapsed + .toggle-folders-btn {
-    position: static;
-    margin-right: auto; /* 确保按钮在左侧 */
-    transform: translateY(0);
+/* Icon-only button style */
+.icon-only {
+    min-width: unset !important;
+    width: 32px !important;
+    padding: 0 !important;
+    height: 32px !important;
 }
 
-.folder-tags.collapsed + .toggle-folders-btn i {
+/* Rotate icon when folders are collapsed */
+.folder-tags.collapsed ~ .actions .toggle-folders-btn i {
     transform: rotate(180deg);
 }
 
-/* 文件夹标签样式 */
-.folder-tags {
-    display: flex;
-    gap: 4px;
-    padding: 2px 0;
-    flex-wrap: wrap;
-}
-
 /* Add custom scrollbar for better visibility */
 .folder-tags::-webkit-scrollbar {
     width: 6px;
@@ -217,16 +220,25 @@
     cursor: pointer;
     padding: 2px 8px;
     margin: 2px;
-    border: 1px solid #ccc;
+    border: 1px solid var(--border-color);
     border-radius: var(--border-radius-xs);
     display: inline-block;
     line-height: 1.2;
     font-size: 14px;
+    background-color: var(--card-bg);
+    transition: all 0.2s ease;
+}
+
+.tag:hover {
+    border-color: var(--lora-accent);
+    background-color: oklch(var(--lora-accent) / 0.1);
+    transform: translateY(-1px);
 }
 
 .tag.active {
-    background-color: #007bff;
+    background-color: var(--lora-accent);
     color: white;
+    border-color: var(--lora-accent);
 }
 
 /* Back to Top Button */
@@ -239,7 +251,7 @@
     border-radius: 50%;
     background: var(--card-bg);
     border: 1px solid var(--border-color);
-    color: var (--text-color);
+    color: var(--text-color);
     display: flex;
     align-items: center;
     justify-content: center;
@@ -249,6 +261,7 @@
     transform: translateY(10px);
     transition: all 0.3s ease;
     z-index: var(--z-overlay);
+    box-shadow: 0 2px 5px rgba(0, 0, 0, 0.1);
 }
 
 .back-to-top.visible {
@@ -258,84 +271,10 @@
 }
 
 .back-to-top:hover {
-    background: var (--lora-accent);
-    color: white;
-    transform: translateY(-2px);
-}
-
-.theme-toggle {
-    width: 36px;
-    height: 36px;
-    border-radius: 50%;
-    background: var(--card-bg);
-    border: 1px solid var(--border-color);
-    color: var(--text-color);
-    display: flex;
-    align-items: center;
-    justify-content: center;
-    cursor: pointer;
-    transition: all 0.2s ease;
-}
-
-.theme-toggle:hover {
     background: var(--lora-accent);
     color: white;
     transform: translateY(-2px);
-}
-
-.support-toggle {
-    width: 36px;
-    height: 36px;
-    border-radius: 50%;
-    background: var(--card-bg);
-    border: 1px solid var(--border-color);
-    color: var(--lora-error);
-    display: flex;
-    align-items: center;
-    justify-content: center;
-    cursor: pointer;
-    transition: all 0.2s ease;
-}
-
-.support-toggle:hover {
-    background: var(--lora-error);
-    color: white;
-    transform: translateY(-2px);
-}
-
-.support-toggle i {
-    font-size: 1.1em;
-    position: relative;
-    top: 1px;
-    left: -0.5px;
-}
-
-.theme-toggle img {
-    width: 20px;
-    height: 20px;
-}
-
-.theme-toggle .theme-icon {
-    width: 20px;
-    height: 20px;
-    position: absolute;
-    transition: opacity 0.2s ease;
-}
-
-.theme-toggle .light-icon {
-    opacity: 0;
-}
-
-.theme-toggle .dark-icon {
-    opacity: 1;
-}
-
-[data-theme="light"] .theme-toggle .light-icon {
-    opacity: 1;
-}
-
-[data-theme="light"] .theme-toggle .dark-icon {
-    opacity: 0;
+    box-shadow: 0 4px 8px rgba(0, 0, 0, 0.15);
 }
 
 @media (max-width: 768px) {
@@ -343,55 +282,41 @@
         flex-wrap: wrap;
         gap: var(--space-1);
     }
-
-    .controls {
-        flex-direction: column;
-        gap: 15px;
+    
+    .action-buttons {
+        flex-wrap: wrap;
+        gap: var(--space-1);
+        width: 100%;
+    }
+    
+    .toggle-folders-container {
+        margin-left: 0;
+        width: 100%;
+        display: flex;
+        justify-content: flex-end;
     }
     
     .folder-tags-container {
         order: -1;
     }
     
-    .toggle-folders-btn {
-        position: absolute;
-        bottom: 0;
-        right: 0;
-        transform: none; /* 移除transform，防止hover时的位移 */
-    }
-    
     .toggle-folders-btn:hover {
-        transform: none; /* 移动端下禁用hover效果 */
-    }
-    
-    .folder-tags.collapsed + .toggle-folders-btn {
-        position: relative;
-        transform: none;
+        transform: none; /* Disable hover effects on mobile */
     }
 
-    .corner-controls {
-        top: 10px;
-        right: 10px;
+    .control-group button:hover {
+        transform: none; /* Disable hover effects on mobile */
     }
-    
-    .corner-controls-items {
-        display: none;
+
+    .control-group select:hover {
+        transform: none; /* Disable hover effects on mobile */
     }
-    
-    .corner-controls.expanded .corner-controls-items {
-        display: flex;
+
+    .tag:hover {
+        transform: none; /* Disable hover effects on mobile */
     }
 
     .back-to-top {
         bottom: 60px; /* Give some extra space from bottom on mobile */
     }
 }
-
-/* Standardize button widths in controls */
-.control-group button {
-    min-width: 100px;
-    display: flex;
-    align-items: center;
-    justify-content: center;
-    gap: 6px;
-}

static/css/style.css

@@ -5,6 +5,7 @@
 @import 'layout.css';
 
 /* Import Components */
+@import 'components/header.css';
 @import 'components/card.css';
 @import 'components/modal.css';
 @import 'components/download-modal.css';
@@ -16,6 +17,9 @@
 @import 'components/support-modal.css';
 @import 'components/search-filter.css';
 @import 'components/bulk.css';
+@import 'components/shared.css';
+@import 'components/filter-indicator.css';
+@import 'components/initialization.css';
 
 .initialization-notice {
     display: flex;

static/images/site.webmanifest

@@ -1 +1 @@
-{"name":"","short_name":"","icons":[{"src":"/android-chrome-192x192.png","sizes":"192x192","type":"image/png"},{"src":"/android-chrome-512x512.png","sizes":"512x512","type":"image/png"}],"theme_color":"#ffffff","background_color":"#ffffff","display":"standalone"}
+{"name":"","short_name":"","icons":[{"src":"/loras_static/images/android-chrome-192x192.png","sizes":"192x192","type":"image/png"},{"src":"/loras_static/images/android-chrome-512x512.png","sizes":"512x512","type":"image/png"}],"theme_color":"#ffffff","background_color":"#ffffff","display":"standalone"}

static/js/api/baseModelApi.js

@@ -0,0 +1,499 @@
+// filepath: d:\Workspace\ComfyUI\custom_nodes\ComfyUI-Lora-Manager\static\js\api\baseModelApi.js
+import { state, getCurrentPageState } from '../state/index.js';
+import { showToast } from '../utils/uiHelpers.js';
+import { showDeleteModal, confirmDelete } from '../utils/modalUtils.js';
+import { getSessionItem } from '../utils/storageHelpers.js';
+
+/**
+ * Shared functionality for handling models (loras and checkpoints)
+ */
+
+// Generic function to load more models with pagination
+export async function loadMoreModels(options = {}) {
+    const {
+        resetPage = false,
+        updateFolders = false,
+        modelType = 'lora', // 'lora' or 'checkpoint'
+        createCardFunction,
+        endpoint = '/api/loras'
+    } = options;
+
+    const pageState = getCurrentPageState();
+    
+    if (pageState.isLoading || (!pageState.hasMore && !resetPage)) return;
+    
+    pageState.isLoading = true;
+    document.body.classList.add('loading');
+    
+    try {
+        // Reset to first page if requested
+        if (resetPage) {
+            pageState.currentPage = 1;
+            // Clear grid if resetting
+            const gridId = modelType === 'checkpoint' ? 'checkpointGrid' : 'loraGrid';
+            const grid = document.getElementById(gridId);
+            if (grid) grid.innerHTML = '';
+        }
+        
+        const params = new URLSearchParams({
+            page: pageState.currentPage,
+            page_size: pageState.pageSize || 20,
+            sort_by: pageState.sortBy
+        });
+        
+        if (pageState.activeFolder !== null) {
+            params.append('folder', pageState.activeFolder);
+        }
+
+        // Add search parameters if there's a search term
+        if (pageState.filters?.search) {
+            params.append('search', pageState.filters.search);
+            params.append('fuzzy', 'true');
+            
+            // Add search option parameters if available
+            if (pageState.searchOptions) {
+                params.append('search_filename', pageState.searchOptions.filename.toString());
+                params.append('search_modelname', pageState.searchOptions.modelname.toString());
+                if (pageState.searchOptions.tags !== undefined) {
+                    params.append('search_tags', pageState.searchOptions.tags.toString());
+                }
+                params.append('recursive', (pageState.searchOptions?.recursive ?? false).toString());
+            }
+        }
+        
+        // Add filter parameters if active
+        if (pageState.filters) {
+            // Handle tags filters
+            if (pageState.filters.tags && pageState.filters.tags.length > 0) {
+                // Checkpoints API expects individual 'tag' parameters, Loras API expects comma-separated 'tags'
+                if (modelType === 'checkpoint') {
+                    pageState.filters.tags.forEach(tag => {
+                        params.append('tag', tag);
+                    });
+                } else {
+                    params.append('tags', pageState.filters.tags.join(','));
+                }
+            }
+            
+            // Handle base model filters
+            if (pageState.filters.baseModel && pageState.filters.baseModel.length > 0) {
+                if (modelType === 'checkpoint') {
+                    pageState.filters.baseModel.forEach(model => {
+                        params.append('base_model', model);
+                    });
+                } else {
+                    params.append('base_models', pageState.filters.baseModel.join(','));
+                }
+            }
+        }
+
+        // Add model-specific parameters
+        if (modelType === 'lora') {
+            // Check for recipe-based filtering parameters from session storage
+            const filterLoraHash = getSessionItem('recipe_to_lora_filterLoraHash');
+            const filterLoraHashes = getSessionItem('recipe_to_lora_filterLoraHashes');
+
+            // Add hash filter parameter if present
+            if (filterLoraHash) {
+                params.append('lora_hash', filterLoraHash);
+            } 
+            // Add multiple hashes filter if present
+            else if (filterLoraHashes) {
+                try {
+                    if (Array.isArray(filterLoraHashes) && filterLoraHashes.length > 0) {
+                        params.append('lora_hashes', filterLoraHashes.join(','));
+                    }
+                } catch (error) {
+                    console.error('Error parsing lora hashes from session storage:', error);
+                }
+            }
+        }
+
+        const response = await fetch(`${endpoint}?${params}`);
+        if (!response.ok) {
+            throw new Error(`Failed to fetch models: ${response.statusText}`);
+        }
+        
+        const data = await response.json();
+        
+        const gridId = modelType === 'checkpoint' ? 'checkpointGrid' : 'loraGrid';
+        const grid = document.getElementById(gridId);
+        
+        if (data.items.length === 0 && pageState.currentPage === 1) {
+            grid.innerHTML = `<div class="no-results">No ${modelType}s found in this folder</div>`;
+            pageState.hasMore = false;
+        } else if (data.items.length > 0) {
+            pageState.hasMore = pageState.currentPage < data.total_pages;
+            
+            // Append model cards using the provided card creation function
+            data.items.forEach(model => {
+                const card = createCardFunction(model);
+                grid.appendChild(card);
+            });
+            
+            // Increment the page number AFTER successful loading
+            pageState.currentPage++;
+        } else {
+            pageState.hasMore = false;
+        }
+
+        if (updateFolders && data.folders) {
+            updateFolderTags(data.folders);
+        }
+        
+    } catch (error) {
+        console.error(`Error loading ${modelType}s:`, error);
+        showToast(`Failed to load ${modelType}s: ${error.message}`, 'error');
+    } finally {
+        pageState.isLoading = false;
+        document.body.classList.remove('loading');
+    }
+}
+
+// Update folder tags in the UI
+export function updateFolderTags(folders) {
+    const folderTagsContainer = document.querySelector('.folder-tags');
+    if (!folderTagsContainer) return;
+
+    // Keep track of currently selected folder
+    const pageState = getCurrentPageState();
+    const currentFolder = pageState.activeFolder;
+
+    // Create HTML for folder tags
+    const tagsHTML = folders.map(folder => {
+        const isActive = folder === currentFolder;
+        return `<div class="tag ${isActive ? 'active' : ''}" data-folder="${folder}">${folder}</div>`;
+    }).join('');
+
+    // Update the container
+    folderTagsContainer.innerHTML = tagsHTML;
+
+    // Reattach click handlers and ensure the active tag is visible
+    const tags = folderTagsContainer.querySelectorAll('.tag');
+    tags.forEach(tag => {
+        if (typeof toggleFolder === 'function') {
+            tag.addEventListener('click', toggleFolder);
+        }
+        if (tag.dataset.folder === currentFolder) {
+            tag.scrollIntoView({ behavior: 'smooth', block: 'nearest' });
+        }
+    });
+}
+
+// Generic function to replace a model preview
+export function replaceModelPreview(filePath, modelType = 'lora') {
+    // Open file picker
+    const input = document.createElement('input');
+    input.type = 'file';
+    input.accept ='image/*,video/mp4'; 
+    
+    input.onchange = async function() {
+        if (!input.files || !input.files[0]) return;
+        
+        const file = input.files[0];
+        await uploadPreview(filePath, file, modelType);
+    };
+    
+    input.click();
+}
+
+// Delete a model (generic)
+export function deleteModel(filePath, modelType = 'lora') {
+    if (modelType === 'checkpoint') {
+        confirmDelete('Are you sure you want to delete this checkpoint?', () => {
+            performDelete(filePath, modelType);
+        });
+    } else {
+        showDeleteModal(filePath);
+    }
+}
+
+// Reset and reload models
+export async function resetAndReload(options = {}) {
+    const {
+        updateFolders = false,
+        modelType = 'lora', 
+        loadMoreFunction
+    } = options;
+    
+    const pageState = getCurrentPageState();
+    
+    // Reset pagination and load more models
+    if (typeof loadMoreFunction === 'function') {
+        await loadMoreFunction(true, updateFolders);
+    }
+}
+
+// Generic function to refresh models
+export async function refreshModels(options = {}) {
+    const { 
+        modelType = 'lora',
+        scanEndpoint = '/api/loras/scan',
+        resetAndReloadFunction
+    } = options;
+    
+    try {
+        state.loadingManager.showSimpleLoading(`Refreshing ${modelType}s...`);
+        
+        const response = await fetch(scanEndpoint);
+        
+        if (!response.ok) {
+            throw new Error(`Failed to refresh ${modelType}s: ${response.status} ${response.statusText}`);
+        }
+        
+        if (typeof resetAndReloadFunction === 'function') {
+            await resetAndReloadFunction();
+        }
+        
+        showToast(`Refresh complete`, 'success');
+    } catch (error) {
+        console.error(`Refresh failed:`, error);
+        showToast(`Failed to refresh ${modelType}s`, 'error');
+    } finally {
+        state.loadingManager.hide();
+        state.loadingManager.restoreProgressBar();
+    }
+}
+
+// Generic fetch from Civitai
+export async function fetchCivitaiMetadata(options = {}) {
+    const {
+        modelType = 'lora',
+        fetchEndpoint = '/api/fetch-all-civitai',
+        resetAndReloadFunction
+    } = options;
+    
+    let ws = null;
+    
+    await state.loadingManager.showWithProgress(async (loading) => {
+        try {
+            const wsProtocol = window.location.protocol === 'https:' ? 'wss://' : 'ws://';
+            ws = new WebSocket(`${wsProtocol}${window.location.host}/ws/fetch-progress`);
+            
+            const operationComplete = new Promise((resolve, reject) => {
+                ws.onmessage = (event) => {
+                    const data = JSON.parse(event.data);
+                    
+                    switch(data.status) {
+                        case 'started':
+                            loading.setStatus('Starting metadata fetch...');
+                            break;
+                            
+                        case 'processing':
+                            const percent = ((data.processed / data.total) * 100).toFixed(1);
+                            loading.setProgress(percent);
+                            loading.setStatus(
+                                `Processing (${data.processed}/${data.total}) ${data.current_name}`
+                            );
+                            break;
+                            
+                        case 'completed':
+                            loading.setProgress(100);
+                            loading.setStatus(
+                                `Completed: Updated ${data.success} of ${data.processed} ${modelType}s`
+                            );
+                            resolve();
+                            break;
+                            
+                        case 'error':
+                            reject(new Error(data.error));
+                            break;
+                    }
+                };
+                
+                ws.onerror = (error) => {
+                    reject(new Error('WebSocket error: ' + error.message));
+                };
+            });
+            
+            await new Promise((resolve, reject) => {
+                ws.onopen = resolve;
+                ws.onerror = reject;
+            });
+            
+            const requestBody = modelType === 'checkpoint' 
+                ? JSON.stringify({ model_type: 'checkpoint' }) 
+                : JSON.stringify({});
+                
+            const response = await fetch(fetchEndpoint, {
+                method: 'POST',
+                headers: { 'Content-Type': 'application/json' },
+                body: requestBody
+            });
+            
+            if (!response.ok) {
+                throw new Error('Failed to fetch metadata');
+            }
+            
+            await operationComplete;
+            
+            if (typeof resetAndReloadFunction === 'function') {
+                await resetAndReloadFunction();
+            }
+            
+        } catch (error) {
+            console.error('Error fetching metadata:', error);
+            showToast('Failed to fetch metadata: ' + error.message, 'error');
+        } finally {
+            if (ws) {
+                ws.close();
+            }
+        }
+    }, {
+        initialMessage: 'Connecting...',
+        completionMessage: 'Metadata update complete'
+    });
+}
+
+// Generic function to refresh single model metadata
+export async function refreshSingleModelMetadata(filePath, modelType = 'lora') {
+    try {
+        state.loadingManager.showSimpleLoading('Refreshing metadata...');
+        
+        const endpoint = modelType === 'checkpoint' 
+            ? '/api/checkpoints/fetch-civitai'
+            : '/api/fetch-civitai';
+            
+        const response = await fetch(endpoint, {
+            method: 'POST',
+            headers: {
+                'Content-Type': 'application/json'
+            },
+            body: JSON.stringify({ file_path: filePath })
+        });
+
+        if (!response.ok) {
+            throw new Error('Failed to refresh metadata');
+        }
+
+        const data = await response.json();
+        
+        if (data.success) {
+            showToast('Metadata refreshed successfully', 'success');
+            return true;
+        } else {
+            throw new Error(data.error || 'Failed to refresh metadata');
+        }
+    } catch (error) {
+        console.error('Error refreshing metadata:', error);
+        showToast(error.message, 'error');
+        return false;
+    } finally {
+        state.loadingManager.hide();
+        state.loadingManager.restoreProgressBar();
+    }
+}
+
+// Private methods
+
+// Upload a preview image
+async function uploadPreview(filePath, file, modelType = 'lora') {
+    const loadingOverlay = document.getElementById('loading-overlay');
+    const loadingStatus = document.querySelector('.loading-status');
+    
+    try {
+        if (loadingOverlay) loadingOverlay.style.display = 'flex';
+        if (loadingStatus) loadingStatus.textContent = 'Uploading preview...';
+        
+        const formData = new FormData();
+        
+        // Use appropriate parameter names and endpoint based on model type
+        // Prepare common form data
+        formData.append('preview_file', file);
+        formData.append('model_path', filePath);
+
+        // Set endpoint based on model type
+        const endpoint = modelType === 'checkpoint' 
+            ? '/api/checkpoints/replace-preview'
+            : '/api/replace_preview';
+        
+        const response = await fetch(endpoint, {
+            method: 'POST',
+            body: formData
+        });
+        
+        if (!response.ok) {
+            throw new Error('Upload failed');
+        }
+
+        const data = await response.json();
+        
+        // Update the card preview in UI
+        const card = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
+        if (card) {
+            const previewContainer = card.querySelector('.card-preview');
+            const oldPreview = previewContainer.querySelector('img, video');
+            
+            // For LoRA models, use timestamp to prevent caching
+            if (modelType === 'lora') {
+                state.previewVersions?.set(filePath, Date.now());
+            }
+            
+            const timestamp = Date.now();
+            const previewUrl = data.preview_url ? 
+                `${data.preview_url}?t=${timestamp}` : 
+                `/api/model/preview_image?path=${encodeURIComponent(filePath)}&t=${timestamp}`;
+            
+            // Create appropriate element based on file type
+            if (file.type.startsWith('video/')) {
+                const video = document.createElement('video');
+                video.controls = true;
+                video.autoplay = true;
+                video.muted = true;
+                video.loop = true;
+                video.src = previewUrl;
+                oldPreview.replaceWith(video);
+            } else {
+                const img = document.createElement('img');
+                img.src = previewUrl;
+                oldPreview.replaceWith(img);
+            }
+            
+            showToast('Preview updated successfully', 'success');
+        }
+    } catch (error) {
+        console.error('Error uploading preview:', error);
+        showToast('Failed to upload preview image', 'error');
+    } finally {
+        if (loadingOverlay) loadingOverlay.style.display = 'none';
+    }
+}
+
+// Private function to perform the delete operation
+async function performDelete(filePath, modelType = 'lora') {
+    try {
+        showToast(`Deleting ${modelType}...`, 'info');
+        
+        const response = await fetch('/api/model/delete', {
+            method: 'POST',
+            headers: {
+                'Content-Type': 'application/json'
+            },
+            body: JSON.stringify({
+                file_path: filePath,
+                model_type: modelType
+            })
+        });
+        
+        if (!response.ok) {
+            throw new Error(`Failed to delete ${modelType}: ${response.status} ${response.statusText}`);
+        }
+        
+        const data = await response.json();
+        
+        if (data.success) {
+            // Remove the card from UI
+            const card = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
+            if (card) {
+                card.remove();
+            }
+            
+            showToast(`${modelType} deleted successfully`, 'success');
+        } else {
+            throw new Error(data.error || `Failed to delete ${modelType}`);
+        }
+    } catch (error) {
+        console.error(`Error deleting ${modelType}:`, error);
+        showToast(`Failed to delete ${modelType}: ${error.message}`, 'error');
+    }
+}

static/js/api/checkpointApi.js

@@ -0,0 +1,57 @@
+import { createCheckpointCard } from '../components/CheckpointCard.js';
+import {
+    loadMoreModels,
+    resetAndReload as baseResetAndReload,
+    refreshModels as baseRefreshModels,
+    deleteModel as baseDeleteModel,
+    replaceModelPreview,
+    fetchCivitaiMetadata
+} from './baseModelApi.js';
+
+// Load more checkpoints with pagination
+export async function loadMoreCheckpoints(resetPagination = true) {
+    return loadMoreModels({
+        resetPage: resetPagination,
+        updateFolders: true,
+        modelType: 'checkpoint',
+        createCardFunction: createCheckpointCard,
+        endpoint: '/api/checkpoints'
+    });
+}
+
+// Reset and reload checkpoints
+export async function resetAndReload() {
+    return baseResetAndReload({
+        updateFolders: true,
+        modelType: 'checkpoint',
+        loadMoreFunction: loadMoreCheckpoints
+    });
+}
+
+// Refresh checkpoints
+export async function refreshCheckpoints() {
+    return baseRefreshModels({
+        modelType: 'checkpoint',
+        scanEndpoint: '/api/checkpoints/scan',
+        resetAndReloadFunction: resetAndReload
+    });
+}
+
+// Delete a checkpoint
+export function deleteCheckpoint(filePath) {
+    return baseDeleteModel(filePath, 'checkpoint');
+}
+
+// Replace checkpoint preview
+export function replaceCheckpointPreview(filePath) {
+    return replaceModelPreview(filePath, 'checkpoint');
+}
+
+// Fetch metadata from Civitai for checkpoints
+export async function fetchCivitai() {
+    return fetchCivitaiMetadata({
+        modelType: 'checkpoint',
+        fetchEndpoint: '/api/checkpoints/fetch-all-civitai',
+        resetAndReloadFunction: resetAndReload
+    });
+}

static/js/api/loraApi.js

@@ -1,250 +1,38 @@
-import { state } from '../state/index.js';
-import { showToast } from '../utils/uiHelpers.js';
 import { createLoraCard } from '../components/LoraCard.js';
-import { initializeInfiniteScroll } from '../utils/infiniteScroll.js';
-import { showDeleteModal } from '../utils/modalUtils.js';
-import { toggleFolder } from '../utils/uiHelpers.js';
+import {
+    loadMoreModels,
+    resetAndReload as baseResetAndReload,
+    refreshModels as baseRefreshModels,
+    deleteModel as baseDeleteModel,
+    replaceModelPreview,
+    fetchCivitaiMetadata,
+    refreshSingleModelMetadata
+} from './baseModelApi.js';
 
-export async function loadMoreLoras(boolUpdateFolders = false) {
-    if (state.isLoading || !state.hasMore) return;
-    
-    state.isLoading = true;
-    try {
-        const params = new URLSearchParams({
-            page: state.currentPage,
-            page_size: 20,
-            sort_by: state.sortBy
-        });
-        
-        // 使用 state 中的 searchManager 获取递归搜索状态
-        const isRecursiveSearch = state.searchManager?.isRecursiveSearch ?? false;
-        
-        if (state.activeFolder !== null) {
-            params.append('folder', state.activeFolder);
-            params.append('recursive', isRecursiveSearch.toString());
-        }
-
-        // Add search parameters if there's a search term
-        const searchInput = document.getElementById('searchInput');
-        if (searchInput && searchInput.value.trim()) {
-            params.append('search', searchInput.value.trim());
-            params.append('fuzzy', 'true');
-        }
-        
-        // Add filter parameters if active
-        if (state.filters && state.filters.baseModel && state.filters.baseModel.length > 0) {
-            // Convert the array of base models to a comma-separated string
-            params.append('base_models', state.filters.baseModel.join(','));
-        }
-
-        console.log('Loading loras with params:', params.toString());
-
-        const response = await fetch(`/api/loras?${params}`);
-        if (!response.ok) {
-            throw new Error(`Failed to fetch loras: ${response.statusText}`);
-        }
-        
-        const data = await response.json();
-        console.log('Received data:', data);
-        
-        if (data.items.length === 0 && state.currentPage === 1) {
-            const grid = document.getElementById('loraGrid');
-            grid.innerHTML = '<div class="no-results">No loras found in this folder</div>';
-            state.hasMore = false;
-        } else if (data.items.length > 0) {
-            state.hasMore = state.currentPage < data.total_pages;
-            state.currentPage++;
-            appendLoraCards(data.items);
-            
-            const sentinel = document.getElementById('scroll-sentinel');
-            if (sentinel && state.observer) {
-                state.observer.observe(sentinel);
-            }
-        } else {
-            state.hasMore = false;
-        }
-
-        if (boolUpdateFolders && data.folders) {
-            updateFolderTags(data.folders);
-        }
-        
-    } catch (error) {
-        console.error('Error loading loras:', error);
-        showToast('Failed to load loras: ' + error.message, 'error');
-    } finally {
-        state.isLoading = false;
-    }
-}
-
-function updateFolderTags(folders) {
-    const folderTagsContainer = document.querySelector('.folder-tags');
-    if (!folderTagsContainer) return;
-
-    // Keep track of currently selected folder
-    const currentFolder = state.activeFolder;
-
-    // Create HTML for folder tags
-    const tagsHTML = folders.map(folder => {
-        const isActive = folder === currentFolder;
-        return `<div class="tag ${isActive ? 'active' : ''}" data-folder="${folder}">${folder}</div>`;
-    }).join('');
-
-    // Update the container
-    folderTagsContainer.innerHTML = tagsHTML;
-
-    // Reattach click handlers and ensure the active tag is visible
-    const tags = folderTagsContainer.querySelectorAll('.tag');
-    tags.forEach(tag => {
-        tag.addEventListener('click', toggleFolder);
-        if (tag.dataset.folder === currentFolder) {
-            tag.scrollIntoView({ behavior: 'smooth', block: 'nearest' });
-        }
+export async function loadMoreLoras(resetPage = false, updateFolders = false) {
+    return loadMoreModels({
+        resetPage,
+        updateFolders,
+        modelType: 'lora',
+        createCardFunction: createLoraCard,
+        endpoint: '/api/loras'
     });
 }
 
 export async function fetchCivitai() {
-    let ws = null;
-    
-    await state.loadingManager.showWithProgress(async (loading) => {
-        try {
-            ws = new WebSocket(`ws://${window.location.host}/ws/fetch-progress`);
-            
-            const operationComplete = new Promise((resolve, reject) => {
-                ws.onmessage = (event) => {
-                    const data = JSON.parse(event.data);
-                    
-                    switch(data.status) {
-                        case 'started':
-                            loading.setStatus('Starting metadata fetch...');
-                            break;
-                            
-                        case 'processing':
-                            const percent = ((data.processed / data.total) * 100).toFixed(1);
-                            loading.setProgress(percent);
-                            loading.setStatus(
-                                `Processing (${data.processed}/${data.total}) ${data.current_name}`
-                            );
-                            break;
-                            
-                        case 'completed':
-                            loading.setProgress(100);
-                            loading.setStatus(
-                                `Completed: Updated ${data.success} of ${data.processed} loras`
-                            );
-                            resolve();
-                            break;
-                            
-                        case 'error':
-                            reject(new Error(data.error));
-                            break;
-                    }
-                };
-                
-                ws.onerror = (error) => {
-                    reject(new Error('WebSocket error: ' + error.message));
-                };
-            });
-            
-            await new Promise((resolve, reject) => {
-                ws.onopen = resolve;
-                ws.onerror = reject;
-            });
-            
-            const response = await fetch('/api/fetch-all-civitai', {
-                method: 'POST',
-                headers: { 'Content-Type': 'application/json' }
-            });
-            
-            if (!response.ok) {
-                throw new Error('Failed to fetch metadata');
-            }
-            
-            await operationComplete;
-            
-            await resetAndReload();
-            
-        } catch (error) {
-            console.error('Error fetching metadata:', error);
-            showToast('Failed to fetch metadata: ' + error.message, 'error');
-        } finally {
-            if (ws) {
-                ws.close();
-            }
-        }
-    }, {
-        initialMessage: 'Connecting...',
-        completionMessage: 'Metadata update complete'
+    return fetchCivitaiMetadata({
+        modelType: 'lora',
+        fetchEndpoint: '/api/fetch-all-civitai',
+        resetAndReloadFunction: resetAndReload
     });
 }
 
 export async function deleteModel(filePath) {
-    showDeleteModal(filePath);
+    return baseDeleteModel(filePath, 'lora');
 }
 
 export async function replacePreview(filePath) {
-    const loadingOverlay = document.getElementById('loading-overlay');
-    const loadingStatus = document.querySelector('.loading-status');
-    
-    const input = document.createElement('input');
-    input.type = 'file';
-    input.accept = 'image/*,video/mp4';
-    
-    input.onchange = async function() {
-        if (!input.files || !input.files[0]) return;
-        
-        const file = input.files[0];
-        const formData = new FormData();
-        formData.append('preview_file', file);
-        formData.append('model_path', filePath);
-        
-        try {
-            loadingOverlay.style.display = 'flex';
-            loadingStatus.textContent = 'Uploading preview...';
-            
-            const response = await fetch('/api/replace_preview', {
-                method: 'POST',
-                body: formData
-            });
-            
-            if (!response.ok) {
-                throw new Error('Upload failed');
-            }
-
-            const data = await response.json();
-            
-            // 更新预览版本
-            state.previewVersions.set(filePath, Date.now());
-            
-            // 更新卡片显示
-            const card = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
-            const previewContainer = card.querySelector('.card-preview');
-            const oldPreview = previewContainer.querySelector('img, video');
-            
-            const previewUrl = `${data.preview_url}?t=${state.previewVersions.get(filePath)}`;
-            
-            if (file.type.startsWith('video/')) {
-                const video = document.createElement('video');
-                video.controls = true;
-                video.autoplay = true;
-                video.muted = true;
-                video.loop = true;
-                video.src = previewUrl;
-                oldPreview.replaceWith(video);
-            } else {
-                const img = document.createElement('img');
-                img.src = previewUrl;
-                oldPreview.replaceWith(img);
-            }
-            
-        } catch (error) {
-            console.error('Error uploading preview:', error);
-            alert('Failed to upload preview image');
-        } finally {
-            loadingOverlay.style.display = 'none';
-        }
-    };
-    
-    input.click();
+    return replaceModelPreview(filePath, 'lora');
 }
 
 export function appendLoraCards(loras) {
@@ -253,76 +41,45 @@ export function appendLoraCards(loras) {
     
     loras.forEach(lora => {
         const card = createLoraCard(lora);
-        if (sentinel) {
-            grid.insertBefore(card, sentinel);
-        } else {
-            grid.appendChild(card);
-        }
+        grid.appendChild(card);
     });
 }
 
-export async function resetAndReload(boolUpdateFolders = false) {
-    console.log('Resetting with state:', { ...state });
-    
-    state.currentPage = 1;
-    state.hasMore = true;
-    state.isLoading = false;
-    
-    const grid = document.getElementById('loraGrid');
-    grid.innerHTML = '';
-    
-    const sentinel = document.createElement('div');
-    sentinel.id = 'scroll-sentinel';
-    grid.appendChild(sentinel);
-    
-    initializeInfiniteScroll();
-    
-    await loadMoreLoras(boolUpdateFolders);
+export async function resetAndReload(updateFolders = false) {
+    return baseResetAndReload({
+        updateFolders,
+        modelType: 'lora',
+        loadMoreFunction: loadMoreLoras
+    });
 }
 
 export async function refreshLoras() {
-    try {
-        state.loadingManager.showSimpleLoading('Refreshing loras...');
-        await resetAndReload();
-        showToast('Refresh complete', 'success');
-    } catch (error) {
-        console.error('Refresh failed:', error);
-        showToast('Failed to refresh loras', 'error');
-    } finally {
-        state.loadingManager.hide();
-        state.loadingManager.restoreProgressBar();
-    }
+    return baseRefreshModels({
+        modelType: 'lora',
+        scanEndpoint: '/api/loras/scan',
+        resetAndReloadFunction: resetAndReload
+    });
 }
 
 export async function refreshSingleLoraMetadata(filePath) {
+    const success = await refreshSingleModelMetadata(filePath, 'lora');
+    if (success) {
+        // Reload the current view to show updated data
+        await resetAndReload();
+    }
+}
+
+export async function fetchModelDescription(modelId, filePath) {
     try {
-        state.loadingManager.showSimpleLoading('Refreshing metadata...');
-        const response = await fetch('/api/fetch-civitai', {
-            method: 'POST',
-            headers: {
-                'Content-Type': 'application/json'
-            },
-            body: JSON.stringify({ file_path: filePath })
-        });
-
-        if (!response.ok) {
-            throw new Error('Failed to refresh metadata');
-        }
-
-        const data = await response.json();
+        const response = await fetch(`/api/lora-model-description?model_id=${modelId}&file_path=${encodeURIComponent(filePath)}`);
         
-        if (data.success) {
-            showToast('Metadata refreshed successfully', 'success');
-            // Reload the current view to show updated data
-            await resetAndReload();
-        } else {
-            throw new Error(data.error || 'Failed to refresh metadata');
+        if (!response.ok) {
+            throw new Error(`Failed to fetch model description: ${response.statusText}`);
         }
+        
+        return await response.json();
     } catch (error) {
-        console.error('Error refreshing metadata:', error);
-        showToast(error.message, 'error');
-    } finally {
-        state.loadingManager.hide();
-        state.loadingManager.restoreProgressBar();
+        console.error('Error fetching model description:', error);
+        throw error;
     }
 }

static/js/checkpoints.js

@@ -0,0 +1,55 @@
+import { appCore } from './core.js';
+import { initializeInfiniteScroll } from './utils/infiniteScroll.js';
+import { confirmDelete, closeDeleteModal } from './utils/modalUtils.js';
+import { createPageControls } from './components/controls/index.js';
+import { loadMoreCheckpoints } from './api/checkpointApi.js';
+import { CheckpointDownloadManager } from './managers/CheckpointDownloadManager.js';
+
+// Initialize the Checkpoints page
+class CheckpointsPageManager {
+    constructor() {
+        // Initialize page controls
+        this.pageControls = createPageControls('checkpoints');
+        
+        // Initialize checkpoint download manager
+        window.checkpointDownloadManager = new CheckpointDownloadManager();
+        
+        // Expose only necessary functions to global scope
+        this._exposeRequiredGlobalFunctions();
+    }
+    
+    _exposeRequiredGlobalFunctions() {
+        // Minimal set of functions that need to remain global
+        window.confirmDelete = confirmDelete;
+        window.closeDeleteModal = closeDeleteModal;
+        
+        // Add loadCheckpoints function to window for FilterManager compatibility
+        window.checkpointManager = {
+            loadCheckpoints: (reset) => loadMoreCheckpoints(reset)
+        };
+    }
+    
+    async initialize() {
+        // Initialize page-specific components
+        this.pageControls.restoreFolderFilter();
+        this.pageControls.initFolderTagsVisibility();
+        
+        // Initialize infinite scroll
+        initializeInfiniteScroll('checkpoints');
+        
+        // Initialize common page features
+        appCore.initializePageFeatures();
+        
+        console.log('Checkpoints Manager initialized');
+    }
+}
+
+// Initialize everything when DOM is ready
+document.addEventListener('DOMContentLoaded', async () => {
+    // Initialize core application
+    await appCore.initialize();
+    
+    // Initialize checkpoints page
+    const checkpointsPage = new CheckpointsPageManager();
+    await checkpointsPage.initialize();
+});

static/js/components/CheckpointCard.js

@@ -0,0 +1,313 @@
+import { showToast } from '../utils/uiHelpers.js';
+import { state } from '../state/index.js';
+import { showCheckpointModal } from './checkpointModal/index.js';
+import { NSFW_LEVELS } from '../utils/constants.js';
+import { replaceCheckpointPreview as apiReplaceCheckpointPreview } from '../api/checkpointApi.js';
+
+export function createCheckpointCard(checkpoint) {
+    const card = document.createElement('div');
+    card.className = 'lora-card';  // Reuse the same class for styling
+    card.dataset.sha256 = checkpoint.sha256;
+    card.dataset.filepath = checkpoint.file_path;
+    card.dataset.name = checkpoint.model_name;
+    card.dataset.file_name = checkpoint.file_name;
+    card.dataset.folder = checkpoint.folder;
+    card.dataset.modified = checkpoint.modified;
+    card.dataset.file_size = checkpoint.file_size;
+    card.dataset.from_civitai = checkpoint.from_civitai;
+    card.dataset.notes = checkpoint.notes || '';
+    card.dataset.base_model = checkpoint.base_model || 'Unknown';
+
+    // Store metadata if available
+    if (checkpoint.civitai) {
+        card.dataset.meta = JSON.stringify(checkpoint.civitai || {});
+    }
+    
+    // Store tags if available
+    if (checkpoint.tags && Array.isArray(checkpoint.tags)) {
+        card.dataset.tags = JSON.stringify(checkpoint.tags);
+    }
+
+    if (checkpoint.modelDescription) {
+        card.dataset.modelDescription = checkpoint.modelDescription;
+    }
+
+    // Store NSFW level if available
+    const nsfwLevel = checkpoint.preview_nsfw_level !== undefined ? checkpoint.preview_nsfw_level : 0;
+    card.dataset.nsfwLevel = nsfwLevel;
+    
+    // Determine if the preview should be blurred based on NSFW level and user settings
+    const shouldBlur = state.settings.blurMatureContent && nsfwLevel > NSFW_LEVELS.PG13;
+    if (shouldBlur) {
+        card.classList.add('nsfw-content');
+    }
+
+    // Determine preview URL
+    const previewUrl = checkpoint.preview_url || '/loras_static/images/no-preview.png';
+    const version = state.previewVersions ? state.previewVersions.get(checkpoint.file_path) : null;
+    const versionedPreviewUrl = version ? `${previewUrl}?t=${version}` : previewUrl;
+
+    // Determine NSFW warning text based on level
+    let nsfwText = "Mature Content";
+    if (nsfwLevel >= NSFW_LEVELS.XXX) {
+        nsfwText = "XXX-rated Content";
+    } else if (nsfwLevel >= NSFW_LEVELS.X) {
+        nsfwText = "X-rated Content";
+    } else if (nsfwLevel >= NSFW_LEVELS.R) {
+        nsfwText = "R-rated Content";
+    }
+
+    // Check if autoplayOnHover is enabled for video previews
+    const autoplayOnHover = state.global?.settings?.autoplayOnHover || false;
+    const isVideo = previewUrl.endsWith('.mp4');
+    const videoAttrs = autoplayOnHover ? 'controls muted loop' : 'controls autoplay muted loop';
+
+    card.innerHTML = `
+        <div class="card-preview ${shouldBlur ? 'blurred' : ''}">
+            ${isVideo ? 
+                `<video ${videoAttrs}>
+                    <source src="${versionedPreviewUrl}" type="video/mp4">
+                </video>` :
+                `<img src="${versionedPreviewUrl}" alt="${checkpoint.model_name}">`
+            }
+            <div class="card-header">
+                ${shouldBlur ? 
+                  `<button class="toggle-blur-btn" title="Toggle blur">
+                      <i class="fas fa-eye"></i>
+                  </button>` : ''}
+                <span class="base-model-label ${shouldBlur ? 'with-toggle' : ''}" title="${checkpoint.base_model}">
+                    ${checkpoint.base_model}
+                </span>
+                <div class="card-actions">
+                    <i class="fas fa-globe" 
+                       title="${checkpoint.from_civitai ? 'View on Civitai' : 'Not available from Civitai'}"
+                       ${!checkpoint.from_civitai ? 'style="opacity: 0.5; cursor: not-allowed"' : ''}>
+                    </i>
+                    <i class="fas fa-copy" 
+                       title="Copy Checkpoint Name">
+                    </i>
+                    <i class="fas fa-trash" 
+                       title="Delete Model">
+                    </i>
+                </div>
+            </div>
+            ${shouldBlur ? `
+                <div class="nsfw-overlay">
+                    <div class="nsfw-warning">
+                        <p>${nsfwText}</p>
+                        <button class="show-content-btn">Show</button>
+                    </div>
+                </div>
+            ` : ''}
+            <div class="card-footer">
+                <div class="model-info">
+                    <span class="model-name">${checkpoint.model_name}</span>
+                </div>
+                <div class="card-actions">
+                    <i class="fas fa-image" 
+                       title="Replace Preview Image">
+                    </i>
+                </div>
+            </div>
+        </div>
+    `;
+
+    // Main card click event
+    card.addEventListener('click', () => {
+        // Show checkpoint details modal
+        const checkpointMeta = {
+            sha256: card.dataset.sha256,
+            file_path: card.dataset.filepath,
+            model_name: card.dataset.name,
+            file_name: card.dataset.file_name,
+            folder: card.dataset.folder,
+            modified: card.dataset.modified,
+            file_size: parseInt(card.dataset.file_size || '0'),
+            from_civitai: card.dataset.from_civitai === 'true',
+            base_model: card.dataset.base_model,
+            notes: card.dataset.notes || '',
+            preview_url: versionedPreviewUrl,
+            // Parse civitai metadata from the card's dataset
+            civitai: (() => {
+                try {
+                    return JSON.parse(card.dataset.meta || '{}');
+                } catch (e) {
+                    console.error('Failed to parse civitai metadata:', e);
+                    return {}; // Return empty object on error
+                }
+            })(),
+            tags: (() => {
+                try {
+                    return JSON.parse(card.dataset.tags || '[]');
+                } catch (e) {
+                    console.error('Failed to parse tags:', e);
+                    return []; // Return empty array on error
+                }
+            })(),
+            modelDescription: card.dataset.modelDescription || ''
+        };
+        showCheckpointModal(checkpointMeta);
+    });
+
+    // Toggle blur button functionality
+    const toggleBlurBtn = card.querySelector('.toggle-blur-btn');
+    if (toggleBlurBtn) {
+        toggleBlurBtn.addEventListener('click', (e) => {
+            e.stopPropagation();
+            const preview = card.querySelector('.card-preview');
+            const isBlurred = preview.classList.toggle('blurred');
+            const icon = toggleBlurBtn.querySelector('i');
+            
+            // Update the icon based on blur state
+            if (isBlurred) {
+                icon.className = 'fas fa-eye';
+            } else {
+                icon.className = 'fas fa-eye-slash';
+            }
+            
+            // Toggle the overlay visibility
+            const overlay = card.querySelector('.nsfw-overlay');
+            if (overlay) {
+                overlay.style.display = isBlurred ? 'flex' : 'none';
+            }
+        });
+    }
+
+    // Show content button functionality
+    const showContentBtn = card.querySelector('.show-content-btn');
+    if (showContentBtn) {
+        showContentBtn.addEventListener('click', (e) => {
+            e.stopPropagation();
+            const preview = card.querySelector('.card-preview');
+            preview.classList.remove('blurred');
+            
+            // Update the toggle button icon
+            const toggleBtn = card.querySelector('.toggle-blur-btn');
+            if (toggleBtn) {
+                toggleBtn.querySelector('i').className = 'fas fa-eye-slash';
+            }
+            
+            // Hide the overlay
+            const overlay = card.querySelector('.nsfw-overlay');
+            if (overlay) {
+                overlay.style.display = 'none';
+            }
+        });
+    }
+
+    // Copy button click event
+    card.querySelector('.fa-copy')?.addEventListener('click', async e => {
+        e.stopPropagation();
+        const checkpointName = card.dataset.file_name;
+        
+        try {
+            // Modern clipboard API
+            if (navigator.clipboard && window.isSecureContext) {
+                await navigator.clipboard.writeText(checkpointName);
+            } else {
+                // Fallback for older browsers
+                const textarea = document.createElement('textarea');
+                textarea.value = checkpointName;
+                textarea.style.position = 'absolute';
+                textarea.style.left = '-99999px';
+                document.body.appendChild(textarea);
+                textarea.select();
+                document.execCommand('copy');
+                document.body.removeChild(textarea);
+            }
+            showToast('Checkpoint name copied', 'success');
+        } catch (err) {
+            console.error('Copy failed:', err);
+            showToast('Copy failed', 'error');
+        }
+    });
+
+    // Civitai button click event
+    if (checkpoint.from_civitai) {
+        card.querySelector('.fa-globe')?.addEventListener('click', e => {
+            e.stopPropagation();
+            openCivitai(checkpoint.model_name);
+        });
+    }
+
+    // Delete button click event
+    card.querySelector('.fa-trash')?.addEventListener('click', e => {
+        e.stopPropagation();
+        deleteCheckpoint(checkpoint.file_path);
+    });
+
+    // Replace preview button click event
+    card.querySelector('.fa-image')?.addEventListener('click', e => {
+        e.stopPropagation();
+        replaceCheckpointPreview(checkpoint.file_path);
+    });
+
+    // Add autoplayOnHover handlers for video elements if needed
+    const videoElement = card.querySelector('video');
+    if (videoElement && autoplayOnHover) {
+        const cardPreview = card.querySelector('.card-preview');
+        
+        // Remove autoplay attribute and pause initially
+        videoElement.removeAttribute('autoplay');
+        videoElement.pause();
+        
+        // Add mouse events to trigger play/pause
+        cardPreview.addEventListener('mouseenter', () => {
+            videoElement.play();
+        });
+        
+        cardPreview.addEventListener('mouseleave', () => {
+            videoElement.pause();
+            videoElement.currentTime = 0;
+        });
+    }
+
+    return card;
+}
+
+// These functions will be implemented in checkpointApi.js
+function openCivitai(modelName) {
+    // Check if the global function exists (registered by PageControls)
+    if (window.openCivitai) {
+        window.openCivitai(modelName);
+    } else {
+        // Fallback implementation
+        const card = document.querySelector(`.lora-card[data-name="${modelName}"]`);
+        if (!card) return;
+        
+        const metaData = JSON.parse(card.dataset.meta || '{}');
+        const civitaiId = metaData.modelId;
+        const versionId = metaData.id;
+        
+        // Build URL
+        if (civitaiId) {
+            let url = `https://civitai.com/models/${civitaiId}`;
+            if (versionId) {
+                url += `?modelVersionId=${versionId}`;
+            }
+            window.open(url, '_blank');
+        } else {
+            // If no ID, try searching by name
+            window.open(`https://civitai.com/models?query=${encodeURIComponent(modelName)}`, '_blank');
+        }
+    }
+}
+
+function deleteCheckpoint(filePath) {
+    if (window.deleteCheckpoint) {
+        window.deleteCheckpoint(filePath);
+    } else {
+        // Use the modal delete functionality
+        import('../utils/modalUtils.js').then(({ showDeleteModal }) => {
+            showDeleteModal(filePath, 'checkpoint');
+        });
+    }
+}
+
+function replaceCheckpointPreview(filePath) {
+    if (window.replaceCheckpointPreview) {
+        window.replaceCheckpointPreview(filePath);
+    } else {
+        apiReplaceCheckpointPreview(filePath);
+    }
+}

static/js/components/ContextMenu.js

@@ -1,9 +1,13 @@
 import { refreshSingleLoraMetadata } from '../api/loraApi.js';
+import { showToast, getNSFWLevelName } from '../utils/uiHelpers.js';
+import { NSFW_LEVELS } from '../utils/constants.js';
+import { getStorageItem } from '../utils/storageHelpers.js';
 
 export class LoraContextMenu {
     constructor() {
         this.menu = document.getElementById('loraContextMenu');
         this.currentCard = null;
+        this.nsfwSelector = document.getElementById('nsfwLevelSelector');
         this.init();
     }
 
@@ -58,10 +62,274 @@ export class LoraContextMenu {
                 case 'refresh-metadata':
                     refreshSingleLoraMetadata(this.currentCard.dataset.filepath);
                     break;
+                case 'set-nsfw':
+                    this.showNSFWLevelSelector(null, null, this.currentCard);
+                    break;
             }
             
             this.hideMenu();
         });
+
+        // Initialize NSFW Level Selector events
+        this.initNSFWSelector();
+    }
+
+    initNSFWSelector() {
+        // Close button
+        const closeBtn = this.nsfwSelector.querySelector('.close-nsfw-selector');
+        closeBtn.addEventListener('click', () => {
+            this.nsfwSelector.style.display = 'none';
+        });
+
+        // Level buttons
+        const levelButtons = this.nsfwSelector.querySelectorAll('.nsfw-level-btn');
+        levelButtons.forEach(btn => {
+            btn.addEventListener('click', async () => {
+                const level = parseInt(btn.dataset.level);
+                const filePath = this.nsfwSelector.dataset.cardPath;
+                
+                if (!filePath) return;
+                
+                try {
+                    await this.saveModelMetadata(filePath, { preview_nsfw_level: level });
+                    
+                    // Update card data
+                    const card = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
+                    if (card) {
+                        let metaData = {};
+                        try {
+                            metaData = JSON.parse(card.dataset.meta || '{}');
+                        } catch (err) {
+                            console.error('Error parsing metadata:', err);
+                        }
+                        
+                        metaData.preview_nsfw_level = level;
+                        card.dataset.meta = JSON.stringify(metaData);
+                        card.dataset.nsfwLevel = level.toString();
+                        
+                        // Apply blur effect immediately
+                        this.updateCardBlurEffect(card, level);
+                    }
+                    
+                    showToast(`Content rating set to ${getNSFWLevelName(level)}`, 'success');
+                    this.nsfwSelector.style.display = 'none';
+                } catch (error) {
+                    showToast(`Failed to set content rating: ${error.message}`, 'error');
+                }
+            });
+        });
+        
+        // Close when clicking outside
+        document.addEventListener('click', (e) => {
+            if (this.nsfwSelector.style.display === 'block' && 
+                !this.nsfwSelector.contains(e.target) && 
+                !e.target.closest('.context-menu-item[data-action="set-nsfw"]')) {
+                this.nsfwSelector.style.display = 'none';
+            }
+        });
+    }
+
+    async saveModelMetadata(filePath, data) {
+        const response = await fetch('/api/loras/save-metadata', {
+            method: 'POST',
+            headers: {
+                'Content-Type': 'application/json',
+            },
+            body: JSON.stringify({
+                file_path: filePath,
+                ...data
+            })
+        });
+    
+        if (!response.ok) {
+            throw new Error('Failed to save metadata');
+        }
+        
+        return await response.json();
+    }
+
+    updateCardBlurEffect(card, level) {
+        // Get user settings for blur threshold
+        const blurThreshold = parseInt(getStorageItem('nsfwBlurLevel') || '4');
+        
+        // Get card preview container
+        const previewContainer = card.querySelector('.card-preview');
+        if (!previewContainer) return;
+        
+        // Get preview media element
+        const previewMedia = previewContainer.querySelector('img') || previewContainer.querySelector('video');
+        if (!previewMedia) return;
+        
+        // Check if blur should be applied
+        if (level >= blurThreshold) {
+            // Add blur class to the preview container
+            previewContainer.classList.add('blurred');
+            
+            // Get or create the NSFW overlay
+            let nsfwOverlay = previewContainer.querySelector('.nsfw-overlay');
+            if (!nsfwOverlay) {
+                // Create new overlay
+                nsfwOverlay = document.createElement('div');
+                nsfwOverlay.className = 'nsfw-overlay';
+                
+                // Create and configure the warning content
+                const warningContent = document.createElement('div');
+                warningContent.className = 'nsfw-warning';
+                
+                // Determine NSFW warning text based on level
+                let nsfwText = "Mature Content";
+                if (level >= NSFW_LEVELS.XXX) {
+                    nsfwText = "XXX-rated Content";
+                } else if (level >= NSFW_LEVELS.X) {
+                    nsfwText = "X-rated Content";
+                } else if (level >= NSFW_LEVELS.R) {
+                    nsfwText = "R-rated Content";
+                }
+                
+                // Add warning text and show button
+                warningContent.innerHTML = `
+                    <p>${nsfwText}</p>
+                    <button class="show-content-btn">Show</button>
+                `;
+                
+                // Add click event to the show button
+                const showBtn = warningContent.querySelector('.show-content-btn');
+                showBtn.addEventListener('click', (e) => {
+                    e.stopPropagation();
+                    previewContainer.classList.remove('blurred');
+                    nsfwOverlay.style.display = 'none';
+                    
+                    // Update toggle button icon if it exists
+                    const toggleBtn = card.querySelector('.toggle-blur-btn');
+                    if (toggleBtn) {
+                        toggleBtn.querySelector('i').className = 'fas fa-eye-slash';
+                    }
+                });
+                
+                nsfwOverlay.appendChild(warningContent);
+                previewContainer.appendChild(nsfwOverlay);
+            } else {
+                // Update existing overlay
+                const warningText = nsfwOverlay.querySelector('p');
+                if (warningText) {
+                    let nsfwText = "Mature Content";
+                    if (level >= NSFW_LEVELS.XXX) {
+                        nsfwText = "XXX-rated Content";
+                    } else if (level >= NSFW_LEVELS.X) {
+                        nsfwText = "X-rated Content";
+                    } else if (level >= NSFW_LEVELS.R) {
+                        nsfwText = "R-rated Content";
+                    }
+                    warningText.textContent = nsfwText;
+                }
+                nsfwOverlay.style.display = 'flex';
+            }
+            
+            // Get or create the toggle button in the header
+            const cardHeader = previewContainer.querySelector('.card-header');
+            if (cardHeader) {
+                let toggleBtn = cardHeader.querySelector('.toggle-blur-btn');
+                
+                if (!toggleBtn) {
+                    toggleBtn = document.createElement('button');
+                    toggleBtn.className = 'toggle-blur-btn';
+                    toggleBtn.title = 'Toggle blur';
+                    toggleBtn.innerHTML = '<i class="fas fa-eye"></i>';
+                    
+                    // Add click event to toggle button
+                    toggleBtn.addEventListener('click', (e) => {
+                        e.stopPropagation();
+                        const isBlurred = previewContainer.classList.toggle('blurred');
+                        const icon = toggleBtn.querySelector('i');
+                        
+                        // Update icon and overlay visibility
+                        if (isBlurred) {
+                            icon.className = 'fas fa-eye';
+                            nsfwOverlay.style.display = 'flex';
+                        } else {
+                            icon.className = 'fas fa-eye-slash';
+                            nsfwOverlay.style.display = 'none';
+                        }
+                    });
+                    
+                    // Add to the beginning of header
+                    cardHeader.insertBefore(toggleBtn, cardHeader.firstChild);
+                    
+                    // Update base model label class
+                    const baseModelLabel = cardHeader.querySelector('.base-model-label');
+                    if (baseModelLabel && !baseModelLabel.classList.contains('with-toggle')) {
+                        baseModelLabel.classList.add('with-toggle');
+                    }
+                } else {
+                    // Update existing toggle button
+                    toggleBtn.querySelector('i').className = 'fas fa-eye';
+                }
+            }
+        } else {
+            // Remove blur
+            previewContainer.classList.remove('blurred');
+            
+            // Hide overlay if it exists
+            const overlay = previewContainer.querySelector('.nsfw-overlay');
+            if (overlay) overlay.style.display = 'none';
+            
+            // Update or remove toggle button
+            const toggleBtn = card.querySelector('.toggle-blur-btn');
+            if (toggleBtn) {
+                // We'll leave the button but update the icon
+                toggleBtn.querySelector('i').className = 'fas fa-eye-slash';
+            }
+        }
+    }
+
+    showNSFWLevelSelector(x, y, card) {
+        const selector = document.getElementById('nsfwLevelSelector');
+        const currentLevelEl = document.getElementById('currentNSFWLevel');
+        
+        // Get current NSFW level
+        let currentLevel = 0;
+        try {
+            const metaData = JSON.parse(card.dataset.meta || '{}');
+            currentLevel = metaData.preview_nsfw_level || 0;
+            
+            // Update if we have no recorded level but have a dataset attribute
+            if (!currentLevel && card.dataset.nsfwLevel) {
+                currentLevel = parseInt(card.dataset.nsfwLevel) || 0;
+            }
+        } catch (err) {
+            console.error('Error parsing metadata:', err);
+        }
+        
+        currentLevelEl.textContent = getNSFWLevelName(currentLevel);
+        
+        // Position the selector
+        if (x && y) {
+            const viewportWidth = document.documentElement.clientWidth;
+            const viewportHeight = document.documentElement.clientHeight;
+            const selectorRect = selector.getBoundingClientRect();
+            
+            // Center the selector if no coordinates provided
+            let finalX = (viewportWidth - selectorRect.width) / 2;
+            let finalY = (viewportHeight - selectorRect.height) / 2;
+            
+            selector.style.left = `${finalX}px`;
+            selector.style.top = `${finalY}px`;
+        }
+        
+        // Highlight current level button
+        document.querySelectorAll('.nsfw-level-btn').forEach(btn => {
+            if (parseInt(btn.dataset.level) === currentLevel) {
+                btn.classList.add('active');
+            } else {
+                btn.classList.remove('active');
+            }
+        });
+        
+        // Store reference to current card
+        selector.dataset.cardPath = card.dataset.filepath;
+        
+        // Show selector
+        selector.style.display = 'block';
     }
 
     showMenu(x, y, card) {

static/js/components/Header.js

@@ -0,0 +1,82 @@
+import { updateService } from '../managers/UpdateService.js';
+import { toggleTheme } from '../utils/uiHelpers.js';
+import { SearchManager } from '../managers/SearchManager.js';
+import { FilterManager } from '../managers/FilterManager.js';
+import { initPageState } from '../state/index.js';
+
+/**
+ * Header.js - Manages the application header behavior across different pages
+ * Handles initialization of appropriate search and filter managers based on current page
+ */
+export class HeaderManager {
+    constructor() {
+      this.currentPage = this.detectCurrentPage();
+      initPageState(this.currentPage);
+      this.searchManager = null;
+      this.filterManager = null;
+      
+      // Initialize appropriate managers based on current page
+      this.initializeManagers();
+      
+      // Set up common header functionality
+      this.initializeCommonElements();
+    }
+    
+    detectCurrentPage() {
+      const path = window.location.pathname;
+      if (path.includes('/loras/recipes')) return 'recipes';
+      if (path.includes('/checkpoints')) return 'checkpoints';
+      if (path.includes('/loras')) return 'loras';
+      return 'unknown';
+    }
+    
+    initializeManagers() {
+      // Initialize SearchManager for all page types
+      this.searchManager = new SearchManager({ page: this.currentPage });
+      window.searchManager = this.searchManager;
+      
+      // Initialize FilterManager for all page types that have filters
+      if (document.getElementById('filterButton')) {
+        this.filterManager = new FilterManager({ page: this.currentPage });
+        window.filterManager = this.filterManager;
+      }
+    }
+    
+    initializeCommonElements() {
+      // Handle theme toggle
+      const themeToggle = document.querySelector('.theme-toggle');
+      if (themeToggle) {
+        themeToggle.addEventListener('click', () => {
+          if (typeof toggleTheme === 'function') {
+            toggleTheme();
+          }
+        });
+      }
+      
+      // Handle settings toggle
+      const settingsToggle = document.querySelector('.settings-toggle');
+      if (settingsToggle) {
+        settingsToggle.addEventListener('click', () => {
+          if (window.settingsManager) {
+            window.settingsManager.toggleSettings();
+          }
+        });
+      }
+      
+      // Handle update toggle
+      const updateToggle = document.getElementById('updateToggleBtn');
+      if (updateToggle) {
+        updateToggle.addEventListener('click', () => {
+          updateService.toggleUpdateModal();
+        });
+      }
+      
+      // Handle support toggle
+      const supportToggle = document.getElementById('supportToggleBtn');
+      if (supportToggle) {
+        supportToggle.addEventListener('click', () => {
+          // Handle support panel logic
+        });
+      }
+    }
+}

static/js/components/LoraCard.js

@@ -1,7 +1,9 @@
-import { showToast } from '../utils/uiHelpers.js';
+import { showToast, openCivitai } from '../utils/uiHelpers.js';
 import { state } from '../state/index.js';
-import { showLoraModal } from './LoraModal.js';
+import { showLoraModal } from './loraModal/index.js';
 import { bulkManager } from '../managers/BulkManager.js';
+import { NSFW_LEVELS } from '../utils/constants.js';
+import { replacePreview, deleteModel } from '../api/loraApi.js'
 
 export function createLoraCard(lora) {
     const card = document.createElement('div');
@@ -18,6 +20,24 @@ export function createLoraCard(lora) {
     card.dataset.usage_tips = lora.usage_tips;
     card.dataset.notes = lora.notes;
     card.dataset.meta = JSON.stringify(lora.civitai || {});
+    
+    // Store tags and model description
+    if (lora.tags && Array.isArray(lora.tags)) {
+        card.dataset.tags = JSON.stringify(lora.tags);
+    }
+    if (lora.modelDescription) {
+        card.dataset.modelDescription = lora.modelDescription;
+    }
+
+    // Store NSFW level if available
+    const nsfwLevel = lora.preview_nsfw_level !== undefined ? lora.preview_nsfw_level : 0;
+    card.dataset.nsfwLevel = nsfwLevel;
+    
+    // Determine if the preview should be blurred based on NSFW level and user settings
+    const shouldBlur = state.settings.blurMatureContent && nsfwLevel > NSFW_LEVELS.PG13;
+    if (shouldBlur) {
+        card.classList.add('nsfw-content');
+    }
 
     // Apply selection state if in bulk mode and this card is in the selected set
     if (state.bulkMode && state.selectedLoras.has(lora.file_path)) {
@@ -28,16 +48,35 @@ export function createLoraCard(lora) {
     const previewUrl = lora.preview_url || '/loras_static/images/no-preview.png';
     const versionedPreviewUrl = version ? `${previewUrl}?t=${version}` : previewUrl;
 
+    // Determine NSFW warning text based on level
+    let nsfwText = "Mature Content";
+    if (nsfwLevel >= NSFW_LEVELS.XXX) {
+        nsfwText = "XXX-rated Content";
+    } else if (nsfwLevel >= NSFW_LEVELS.X) {
+        nsfwText = "X-rated Content";
+    } else if (nsfwLevel >= NSFW_LEVELS.R) {
+        nsfwText = "R-rated Content";
+    }
+
+    // Check if autoplayOnHover is enabled for video previews
+    const autoplayOnHover = state.global.settings.autoplayOnHover || false;
+    const isVideo = previewUrl.endsWith('.mp4');
+    const videoAttrs = autoplayOnHover ? 'controls muted loop' : 'controls autoplay muted loop';
+
     card.innerHTML = `
-        <div class="card-preview">
-            ${previewUrl.endsWith('.mp4') ? 
-                `<video controls autoplay muted loop>
+        <div class="card-preview ${shouldBlur ? 'blurred' : ''}">
+            ${isVideo ? 
+                `<video ${videoAttrs}>
                     <source src="${versionedPreviewUrl}" type="video/mp4">
                 </video>` :
                 `<img src="${versionedPreviewUrl}" alt="${lora.model_name}">`
             }
             <div class="card-header">
-                <span class="base-model-label" title="${lora.base_model}">
+                ${shouldBlur ? 
+                  `<button class="toggle-blur-btn" title="Toggle blur">
+                      <i class="fas fa-eye"></i>
+                  </button>` : ''}
+                <span class="base-model-label ${shouldBlur ? 'with-toggle' : ''}" title="${lora.base_model}">
                     ${lora.base_model}
                 </span>
                 <div class="card-actions">
@@ -53,6 +92,14 @@ export function createLoraCard(lora) {
                     </i>
                 </div>
             </div>
+            ${shouldBlur ? `
+                <div class="nsfw-overlay">
+                    <div class="nsfw-warning">
+                        <p>${nsfwText}</p>
+                        <button class="show-content-btn">Show</button>
+                    </div>
+                </div>
+            ` : ''}
             <div class="card-footer">
                 <div class="model-info">
                     <span class="model-name">${lora.model_name}</span>
@@ -86,12 +133,69 @@ export function createLoraCard(lora) {
                 base_model: card.dataset.base_model,
                 usage_tips: card.dataset.usage_tips,
                 notes: card.dataset.notes,
-                civitai: JSON.parse(card.dataset.meta || '{}')
+                // Parse civitai metadata from the card's dataset
+                civitai: (() => {
+                    try {
+                        // Attempt to parse the JSON string
+                        return JSON.parse(card.dataset.meta || '{}');
+                    } catch (e) {
+                        console.error('Failed to parse civitai metadata:', e);
+                        return {}; // Return empty object on error
+                    }
+                })(),
+                tags: JSON.parse(card.dataset.tags || '[]'),
+                modelDescription: card.dataset.modelDescription || ''
             };
             showLoraModal(loraMeta);
         }
     });
 
+    // Toggle blur button functionality
+    const toggleBlurBtn = card.querySelector('.toggle-blur-btn');
+    if (toggleBlurBtn) {
+        toggleBlurBtn.addEventListener('click', (e) => {
+            e.stopPropagation();
+            const preview = card.querySelector('.card-preview');
+            const isBlurred = preview.classList.toggle('blurred');
+            const icon = toggleBlurBtn.querySelector('i');
+            
+            // Update the icon based on blur state
+            if (isBlurred) {
+                icon.className = 'fas fa-eye';
+            } else {
+                icon.className = 'fas fa-eye-slash';
+            }
+            
+            // Toggle the overlay visibility
+            const overlay = card.querySelector('.nsfw-overlay');
+            if (overlay) {
+                overlay.style.display = isBlurred ? 'flex' : 'none';
+            }
+        });
+    }
+
+    // Show content button functionality
+    const showContentBtn = card.querySelector('.show-content-btn');
+    if (showContentBtn) {
+        showContentBtn.addEventListener('click', (e) => {
+            e.stopPropagation();
+            const preview = card.querySelector('.card-preview');
+            preview.classList.remove('blurred');
+            
+            // Update the toggle button icon
+            const toggleBtn = card.querySelector('.toggle-blur-btn');
+            if (toggleBtn) {
+                toggleBtn.querySelector('i').className = 'fas fa-eye-slash';
+            }
+            
+            // Hide the overlay
+            const overlay = card.querySelector('.nsfw-overlay');
+            if (overlay) {
+                overlay.style.display = 'none';
+            }
+        });
+    }
+
     // Copy button click event
     card.querySelector('.fa-copy')?.addEventListener('click', async e => {
         e.stopPropagation();
@@ -148,6 +252,26 @@ export function createLoraCard(lora) {
             actionGroup.style.display = 'none';
         });
     }
+    
+    // Add autoplayOnHover handlers for video elements if needed
+    const videoElement = card.querySelector('video');
+    if (videoElement && autoplayOnHover) {
+        const cardPreview = card.querySelector('.card-preview');
+        
+        // Remove autoplay attribute and pause initially
+        videoElement.removeAttribute('autoplay');
+        videoElement.pause();
+        
+        // Add mouse events to trigger play/pause
+        cardPreview.addEventListener('mouseenter', () => {
+            videoElement.play();
+        });
+        
+        cardPreview.addEventListener('mouseleave', () => {
+            videoElement.pause();
+            videoElement.currentTime = 0;
+        });
+    }
 
     return card;
 }
@@ -159,11 +283,26 @@ export function updateCardsForBulkMode(isBulkMode) {
     
     document.body.classList.toggle('bulk-mode', isBulkMode);
     
-    document.querySelectorAll('.lora-card').forEach(card => {
+    // Get all lora cards
+    const loraCards = document.querySelectorAll('.lora-card');
+    
+    loraCards.forEach(card => {
+        // Get all action containers for this card
         const actions = card.querySelectorAll('.card-actions');
-        actions.forEach(actionGroup => {
-            actionGroup.style.display = isBulkMode ? 'none' : 'flex';
-        });
+        
+        // Handle display property based on mode
+        if (isBulkMode) {
+            // Hide actions when entering bulk mode
+            actions.forEach(actionGroup => {
+                actionGroup.style.display = 'none';
+            });
+        } else {
+            // Ensure actions are visible when exiting bulk mode
+            actions.forEach(actionGroup => {
+                // We need to reset to default display style which is flex
+                actionGroup.style.display = 'flex';
+            });
+        }
     });
     
     // Apply selection state to cards if entering bulk mode

static/js/components/LoraModal.js

@@ -1,561 +0,0 @@
-import { showToast } from '../utils/uiHelpers.js';
-import { state } from '../state/index.js';
-
-export function showLoraModal(lora) {
-    const escapedWords = lora.civitai?.trainedWords?.length ? 
-        lora.civitai.trainedWords.map(word => word.replace(/'/g, '\\\'')) : [];
-
-    const content = `
-        <div class="modal-content">
-            <button class="close" onclick="modalManager.closeModal('loraModal')">&times;</button>
-            <header class="modal-header">
-                <div class="editable-field model-name-field">
-                    <h2 class="model-name-content" contenteditable="true" spellcheck="false">${lora.model_name}</h2>
-                    <button class="save-btn" onclick="saveModelName('${lora.file_path}')">
-                        <i class="fas fa-save"></i>
-                    </button>
-                </div>
-            </header>
-
-            <div class="modal-body">
-                <div class="info-section">
-                    <div class="info-grid">
-                        <div class="info-item">
-                            <label>Version</label>
-                            <span>${lora.civitai.name || 'N/A'}</span>
-                        </div>
-                        <div class="info-item">
-                            <label>File Name</label>
-                            <div class="file-name-wrapper" onclick="copyFileName('${lora.file_name}')">
-                                <span id="file-name">${lora.file_name || 'N/A'}</span>
-                                <i class="fas fa-copy" title="Copy file name"></i>
-                            </div>
-                        </div>
-                        <div class="info-item location-size">
-                            <div class="location-wrapper">
-                                <label>Location</label>
-                                <span class="file-path">${lora.file_path.replace(/[^/]+$/, '') || 'N/A'}</span>
-                            </div>
-                        </div>
-                        <div class="info-item base-size">
-                            <div class="base-wrapper">
-                                <label>Base Model</label>
-                                <span>${lora.base_model || 'N/A'}</span>
-                            </div>
-                            <div class="size-wrapper">
-                                <label>Size</label>
-                                <span>${formatFileSize(lora.file_size)}</span>
-                            </div>
-                        </div>
-                        <div class="info-item usage-tips">
-                            <label>Usage Tips</label>
-                            <div class="editable-field">
-                                <div class="preset-controls">
-                                    <select id="preset-selector">
-                                        <option value="">Add preset parameter...</option>
-                                        <option value="strength_min">Strength Min</option>
-                                        <option value="strength_max">Strength Max</option>
-                                        <option value="strength">Strength</option>
-                                        <option value="clip_skip">Clip Skip</option>
-                                    </select>
-                                    <input type="number" id="preset-value" step="0.01" placeholder="Value" style="display:none;">
-                                    <button class="add-preset-btn">Add</button>
-                                </div>
-                                <div class="preset-tags">
-                                    ${renderPresetTags(parsePresets(lora.usage_tips))}
-                                </div>
-                            </div>
-                        </div>
-                        ${renderTriggerWords(escapedWords)}
-                        <div class="info-item notes">
-                            <label>Additional Notes</label>
-                            <div class="editable-field">
-                                <div class="notes-content" contenteditable="true" spellcheck="false">${lora.notes || 'Add your notes here...'}</div>
-                                <button class="save-btn" onclick="saveNotes('${lora.file_path}')">
-                                    <i class="fas fa-save"></i>
-                                </button>
-                            </div>
-                        </div>
-                        <div class="info-item full-width">
-                            <label>About this version</label>
-                            <div class="description-text">${lora.description || 'N/A'}</div>
-                        </div>
-                    </div>
-
-                </div>
-
-                ${renderShowcaseImages(lora.civitai.images)}
-            </div>
-        </div>
-    `;
-    
-    modalManager.showModal('loraModal', content);
-    setupEditableFields();
-    setupShowcaseScroll();
-}
-
-// 添加复制文件名的函数
-window.copyFileName = async function(fileName) {
-    try {
-        await navigator.clipboard.writeText(fileName);
-        showToast('File name copied', 'success');
-    } catch (err) {
-        console.error('Copy failed:', err);
-        showToast('Copy failed', 'error');
-    }
-};
-
-// Add function to save model name
-window.saveModelName = async function(filePath) {
-    const modelNameElement = document.querySelector('.model-name-content');
-    const newModelName = modelNameElement.textContent.trim();
-    
-    // Validate model name
-    if (!newModelName) {
-        showToast('Model name cannot be empty', 'error');
-        return;
-    }
-    
-    // Check if model name is too long (limit to 100 characters)
-    if (newModelName.length > 100) {
-        showToast('Model name is too long (maximum 100 characters)', 'error');
-        // Truncate the displayed text
-        modelNameElement.textContent = newModelName.substring(0, 100);
-        return;
-    }
-    
-    try {
-        await saveModelMetadata(filePath, { model_name: newModelName });
-        
-        // Update the corresponding lora card's dataset and display
-        const loraCard = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
-        if (loraCard) {
-            loraCard.dataset.model_name = newModelName;
-            const titleElement = loraCard.querySelector('.card-title');
-            if (titleElement) {
-                titleElement.textContent = newModelName;
-            }
-        }
-        
-        showToast('Model name updated successfully', 'success');
-        
-        // Reload the page to reflect the sorted order
-        setTimeout(() => {
-            window.location.reload();
-        }, 1500);
-    } catch (error) {
-        showToast('Failed to update model name', 'error');
-    }
-};
-
-function setupEditableFields() {
-    const editableFields = document.querySelectorAll('.editable-field [contenteditable]');
-    
-    editableFields.forEach(field => {
-        field.addEventListener('focus', function() {
-            if (this.textContent === 'Add your notes here...' || 
-                this.textContent === 'Save usage tips here..') {
-                this.textContent = '';
-            }
-        });
-
-        field.addEventListener('blur', function() {
-            if (this.textContent.trim() === '') {
-                if (this.classList.contains('model-name-content')) {
-                    // Restore original model name if empty
-                    const filePath = document.querySelector('.modal-content')
-                        .querySelector('.file-path').textContent + 
-                        document.querySelector('.modal-content')
-                        .querySelector('#file-name').textContent + '.safetensors';
-                    const loraCard = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
-                    if (loraCard) {
-                        this.textContent = loraCard.dataset.model_name;
-                    }
-                } else if (this.classList.contains('usage-tips-content')) {
-                    this.textContent = 'Save usage tips here..';
-                } else {
-                    this.textContent = 'Add your notes here...';
-                }
-            }
-        });
-        
-        // Add input validation for model name
-        if (field.classList.contains('model-name-content')) {
-            field.addEventListener('input', function() {
-                // Limit model name length
-                if (this.textContent.length > 100) {
-                    this.textContent = this.textContent.substring(0, 100);
-                    // Place cursor at the end
-                    const range = document.createRange();
-                    const sel = window.getSelection();
-                    range.setStart(this.childNodes[0], 100);
-                    range.collapse(true);
-                    sel.removeAllRanges();
-                    sel.addRange(range);
-                    
-                    showToast('Model name is limited to 100 characters', 'warning');
-                }
-            });
-            
-            field.addEventListener('keydown', function(e) {
-                if (e.key === 'Enter') {
-                    e.preventDefault();
-                    const filePath = document.querySelector('.modal-content')
-                        .querySelector('.file-path').textContent + 
-                        document.querySelector('.modal-content')
-                        .querySelector('#file-name').textContent + '.safetensors';
-                    saveModelName(filePath);
-                }
-            });
-        }
-    });
-
-    const presetSelector = document.getElementById('preset-selector');
-    const presetValue = document.getElementById('preset-value');
-    const addPresetBtn = document.querySelector('.add-preset-btn');
-    const presetTags = document.querySelector('.preset-tags');
-
-    presetSelector.addEventListener('change', function() {
-        const selected = this.value;
-        if (selected) {
-            presetValue.style.display = 'inline-block';
-            presetValue.min = selected.includes('strength') ? 0 : 1;
-            presetValue.max = selected.includes('strength') ? 1 : 12;
-            presetValue.step = selected.includes('strength') ? 0.01 : 1;
-            if (selected === 'clip_skip') {
-                presetValue.type = 'number';
-                presetValue.step = 1;
-            }
-            // Add auto-focus
-            setTimeout(() => presetValue.focus(), 0);
-        } else {
-            presetValue.style.display = 'none';
-        }
-    });
-
-    addPresetBtn.addEventListener('click', async function() {
-        const key = presetSelector.value;
-        const value = presetValue.value;
-        
-        if (!key || !value) return;
-
-        const filePath = document.querySelector('.modal-content')
-            .querySelector('.file-path').textContent + 
-            document.querySelector('.modal-content')
-            .querySelector('#file-name').textContent + '.safetensors';
-
-        const loraCard = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
-        const currentPresets = parsePresets(loraCard.dataset.usage_tips);
-        
-        currentPresets[key] = parseFloat(value);
-        const newPresetsJson = JSON.stringify(currentPresets);
-
-        await saveModelMetadata(filePath, { 
-            usage_tips: newPresetsJson
-        });
-
-        loraCard.dataset.usage_tips = newPresetsJson;
-        presetTags.innerHTML = renderPresetTags(currentPresets);
-        
-        presetSelector.value = '';
-        presetValue.value = '';
-        presetValue.style.display = 'none';
-    });
-
-    // Add keydown event listeners for notes
-    const notesContent = document.querySelector('.notes-content');
-    if (notesContent) {
-        notesContent.addEventListener('keydown', async function(e) {
-            if (e.key === 'Enter') {
-                if (e.shiftKey) {
-                    // Allow shift+enter for new line
-                    return;
-                }
-                e.preventDefault();
-                const filePath = document.querySelector('.modal-content')
-                    .querySelector('.file-path').textContent + 
-                    document.querySelector('.modal-content')
-                    .querySelector('#file-name').textContent + '.safetensors';
-                await saveNotes(filePath);
-            }
-        });
-    }
-
-    // Add keydown event for preset value
-    presetValue.addEventListener('keydown', function(e) {
-        if (e.key === 'Enter') {
-            e.preventDefault();
-            addPresetBtn.click();
-        }
-    });
-}
-
-window.saveNotes = async function(filePath) {
-    const content = document.querySelector('.notes-content').textContent;
-    try {
-        await saveModelMetadata(filePath, { notes: content });
-
-        // Update the corresponding lora card's dataset
-        const loraCard = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
-        if (loraCard) {
-            loraCard.dataset.notes = content;
-        }
-
-        showToast('Notes saved successfully', 'success');
-    } catch (error) {
-        showToast('Failed to save notes', 'error');
-    }
-};
-
-async function saveModelMetadata(filePath, data) {
-    const response = await fetch('/loras/api/save-metadata', {
-        method: 'POST',
-        headers: {
-            'Content-Type': 'application/json',
-        },
-        body: JSON.stringify({
-            file_path: filePath,
-            ...data
-        })
-    });
-
-    if (!response.ok) {
-        throw new Error('Failed to save metadata');
-    }
-}
-
-function renderTriggerWords(words) {
-    if (!words.length) return `
-        <div class="info-item full-width trigger-words">
-            <label>Trigger Words</label>
-            <span>No trigger word needed</span>
-        </div>
-    `;
-    
-    return `
-        <div class="info-item full-width trigger-words">
-            <label>Trigger Words</label>
-            <div class="trigger-words-tags">
-                ${words.map(word => `
-                    <div class="trigger-word-tag" onclick="copyTriggerWord('${word}')">
-                        <span class="trigger-word-content">${word}</span>
-                        <span class="trigger-word-copy">
-                            <i class="fas fa-copy"></i>
-                        </span>
-                    </div>
-                `).join('')}
-            </div>
-        </div>
-    `;
-}
-
-function renderShowcaseImages(images) {
-    if (!images?.length) return '';
-    
-    return `
-        <div class="showcase-section">
-            <div class="scroll-indicator" onclick="toggleShowcase(this)">
-                <i class="fas fa-chevron-down"></i>
-                <span>Scroll or click to show ${images.length} examples</span>
-            </div>
-            <div class="carousel collapsed">
-                <div class="carousel-container">
-                    ${images.map(img => {
-                        // 计算适当的展示高度：
-                        // 1. 保持原始宽高比
-                        // 2. 限制最大高度为视窗高度的60%
-                        // 3. 确保最小高度为容器宽度的40%
-                        const aspectRatio = (img.height / img.width) * 100;
-                        const containerWidth = 800; // modal content的最大宽度
-                        const minHeightPercent = 40; // 最小高度为容器宽度的40%
-                        const maxHeightPercent = (window.innerHeight * 0.6 / containerWidth) * 100;
-                        const heightPercent = Math.max(
-                            minHeightPercent,
-                            Math.min(maxHeightPercent, aspectRatio)
-                        );
-                        
-                        if (img.type === 'video') {
-                            return `
-                                <div class="media-wrapper" style="padding-bottom: ${heightPercent}%">
-                                    <video controls autoplay muted loop crossorigin="anonymous" 
-                                           referrerpolicy="no-referrer" data-src="${img.url}"
-                                           class="lazy">
-                                        <source data-src="${img.url}" type="video/mp4">
-                                        Your browser does not support video playback
-                                    </video>
-                                </div>
-                            `;
-                        }
-                        return `
-                            <div class="media-wrapper" style="padding-bottom: ${heightPercent}%">
-                                <img data-src="${img.url}" 
-                                     alt="Preview" 
-                                     crossorigin="anonymous" 
-                                     referrerpolicy="no-referrer"
-                                     width="${img.width}"
-                                     height="${img.height}"
-                                     class="lazy"> 
-                            </div>
-                        `;
-                    }).join('')}
-                </div>
-            </div>
-            <button class="back-to-top" onclick="scrollToTop(this)">
-                <i class="fas fa-arrow-up"></i>
-            </button>
-        </div>
-    `;
-}
-
-export function toggleShowcase(element) {
-    const carousel = element.nextElementSibling;
-    const isCollapsed = carousel.classList.contains('collapsed');
-    const indicator = element.querySelector('span');
-    const icon = element.querySelector('i');
-    
-    carousel.classList.toggle('collapsed');
-    
-    if (isCollapsed) {
-        const count = carousel.querySelectorAll('.media-wrapper').length;
-        indicator.textContent = `Scroll or click to hide examples`;
-        icon.classList.replace('fa-chevron-down', 'fa-chevron-up');
-        initLazyLoading(carousel);
-    } else {
-        const count = carousel.querySelectorAll('.media-wrapper').length;
-        indicator.textContent = `Scroll or click to show ${count} examples`;
-        icon.classList.replace('fa-chevron-up', 'fa-chevron-down');
-    }
-}
-
-// Add lazy loading initialization
-function initLazyLoading(container) {
-    const lazyElements = container.querySelectorAll('.lazy');
-    
-    const lazyLoad = (element) => {
-        if (element.tagName.toLowerCase() === 'video') {
-            element.src = element.dataset.src;
-            element.querySelector('source').src = element.dataset.src;
-            element.load();
-        } else {
-            element.src = element.dataset.src;
-        }
-        element.classList.remove('lazy');
-    };
-
-    const observer = new IntersectionObserver((entries) => {
-        entries.forEach(entry => {
-            if (entry.isIntersecting) {
-                lazyLoad(entry.target);
-                observer.unobserve(entry.target);
-            }
-        });
-    });
-
-    lazyElements.forEach(element => observer.observe(element));
-}
-
-export function setupShowcaseScroll() {
-    // Change from modal-content to window/document level
-    document.addEventListener('wheel', (event) => {
-        const modalContent = document.querySelector('.modal-content');
-        if (!modalContent) return;
-
-        const showcase = modalContent.querySelector('.showcase-section');
-        if (!showcase) return;
-        
-        const carousel = showcase.querySelector('.carousel');
-        const scrollIndicator = showcase.querySelector('.scroll-indicator');
-        
-        if (carousel?.classList.contains('collapsed') && event.deltaY > 0) {
-            const isNearBottom = modalContent.scrollHeight - modalContent.scrollTop - modalContent.clientHeight < 100;
-            
-            if (isNearBottom) {
-                toggleShowcase(scrollIndicator);
-                event.preventDefault();
-            }
-        }
-    }, { passive: false }); // Add passive: false option here
-
-    // Keep the existing scroll tracking code
-    const modalContent = document.querySelector('.modal-content');
-    if (modalContent) {
-        modalContent.addEventListener('scroll', () => {
-            const backToTopBtn = modalContent.querySelector('.back-to-top');
-            if (backToTopBtn) {
-                if (modalContent.scrollTop > 300) {
-                    backToTopBtn.classList.add('visible');
-                } else {
-                    backToTopBtn.classList.remove('visible');
-                }
-            }
-        });
-    }
-}
-
-export function scrollToTop(button) {
-    const modalContent = button.closest('.modal-content');
-    if (modalContent) {
-        modalContent.scrollTo({
-            top: 0,
-            behavior: 'smooth'
-        });
-    }
-}
-
-function parsePresets(usageTips) {
-    if (!usageTips || usageTips === 'Save usage tips here..') return {};
-    try {
-        return JSON.parse(usageTips);
-    } catch {
-        return {};
-    }
-}
-
-function renderPresetTags(presets) {
-    return Object.entries(presets).map(([key, value]) => `
-        <div class="preset-tag" data-key="${key}">
-            <span>${formatPresetKey(key)}: ${value}</span>
-            <i class="fas fa-times" onclick="removePreset('${key}')"></i>
-        </div>
-    `).join('');
-}
-
-function formatPresetKey(key) {
-    return key.split('_').map(word => 
-        word.charAt(0).toUpperCase() + word.slice(1)
-    ).join(' ');
-}
-
-window.removePreset = async function(key) {
-    const filePath = document.querySelector('.modal-content')
-            .querySelector('.file-path').textContent + 
-            document.querySelector('.modal-content')
-            .querySelector('#file-name').textContent + '.safetensors';
-    const loraCard = document.querySelector(`.lora-card[data-filepath="${filePath}"]`);
-    const currentPresets = parsePresets(loraCard.dataset.usage_tips);
-    
-    delete currentPresets[key];
-    const newPresetsJson = JSON.stringify(currentPresets);
-
-    await saveModelMetadata(filePath, { 
-        usage_tips: newPresetsJson 
-    });
-    
-    loraCard.dataset.usage_tips = newPresetsJson;
-    document.querySelector('.preset-tags').innerHTML = renderPresetTags(currentPresets);
-};
-
-// 添加文件大小格式化函数
-function formatFileSize(bytes) {
-    if (!bytes) return 'N/A';
-    const units = ['B', 'KB', 'MB', 'GB'];
-    let size = bytes;
-    let unitIndex = 0;
-    
-    while (size >= 1024 && unitIndex < units.length - 1) {
-        size /= 1024;
-        unitIndex++;
-    }
-    
-    return `${size.toFixed(1)} ${units[unitIndex]}`;
-}