Fixes a critical bug in FTS query building where multi-word searches
with field restrictions incorrectly used OR between all word+field
combinations instead of requiring ALL words to match within at least
one field.
Example: searching "cute cat" in {title, tags} previously produced:
title:cute* OR title:cat* OR tags:cute* OR tags:cat*
Which matched recipes with ANY word in ANY field.
Now produces:
(title:cute* title:cat*) OR (tags:cute* tags:cat*)
Which requires ALL words to match within at least one field.
Also adds fallback to fuzzy search when FTS returns empty results,
improving search reliability.
Co-Authored-By: Claude <noreply@anthropic.com>
Temporary workaround for issues #772 and #774 where valid previews
are rejected. Path validation is disabled until proper fix for
preview root path handling is implemented.
- Log preview root rebuilding with counts of different root types
- Add detailed debug output when preview paths are rejected
- Improve visibility into path mapping and validation processes
Previously, `map_path_to_link` and `map_link_to_path` returned the original input path when no mapping was found, instead of the normalized version. This could cause inconsistencies when paths with different representations (e.g., trailing slashes) were used. Now both methods consistently return the normalized path, ensuring uniform path handling throughout the application.
Add two new test cases to verify preview path validation behavior on Windows:
1. `test_is_preview_path_allowed_case_insensitive_on_windows`: Ensures path validation is case-insensitive on Windows, addressing issues where drive letters and paths with different cases should match. This resolves GitHub issues #772 and #774.
2. `test_is_preview_path_allowed_rejects_prefix_without_separator`: Prevents false positives by ensuring paths are only allowed when they match the root path exactly followed by a separator, not just sharing a common prefix.
Use os.path.normcase to ensure case-insensitive path matching on Windows, addressing issues where drive letter case mismatches (e.g., 'a:/folder' vs 'A:/folder') prevented correct detection of paths under preview roots. Replace Path.relative_to() with string-based comparison for consistent behavior across platforms.
- Move slider handle value labels 6px upward in both DualRangeSlider and SingleSlider components
- Add consistent line-height of 14px to ensure proper text alignment
- Improves visual spacing and readability of value labels during slider interaction
- Add execution_seed and next_seed parameters to support deterministic randomization across batch executions
- Separate UI display generation from execution stack generation to maintain consistency in batch queues
- Update LoraService to accept optional seed parameter for reproducible randomization
- Ensure each execution with a different seed produces unique results without affecting global random state
Align visual design of Lora Randomizer widget with Loras widget for
consistent UI/UX across the node interface.
Changes:
- Unified border-radius system (4px→6px for containers, 6px for inputs)
- Standardized padding (12px→6px for widget container)
- Reduced slider height (32px→24px) following desktop tool best practices
- Aligned font sizes (12px→13px for labels, 11px→12px for buttons)
- Unified spacing system (16px→6px for sections, 8px→6px for gaps)
- Adjusted widget minimum height (510px→448px) to reflect layout changes
- Introduce LoRA Randomizer system with LoRA Pool and Randomizer nodes
- Add recipe folders, bulk operations, search, sorting, and favorites
- Enable video recipe support and ComfyUI Nodes 2.0 compatibility
- Include performance improvements for faster startup and loading
- Update example workflow for LoRA Randomizer template reference
Add LoraRandomizer extension that monitors node mode changes and triggers
updates to connected downstream trigger word toggle nodes, matching the
behavior implemented for Lora Stacker nodes.
Add `.lm-lora-lock-button` to the list of elements that should not trigger drag initialization in the LoRA widget event handler. This prevents unintended drag actions when interacting with the lock button, improving user experience and interaction clarity.
Move cleanupNavigationShortcuts() call before setting navigationModelType
to ensure the correct model type is preserved when using left/right arrow
keys to navigate between models. Previously, the cleanup would immediately
nullify navigationModelType, causing type-specific modal sections (like
trigger words for embeddings and usage tips for loras) to disappear.
- Add fetch polyfill to test setup for jsdom environment
- Update context menu test to match new implementation that uses fetch API
- Remove deprecated handleDownloadButton expectation
- Fix mock indices for multiple fetch calls
Resolves test failures from commit b0f0158 which refactored GlobalContextMenu
to use fetch API directly instead of calling exampleImagesManager.
- Normalize string quotes to double quotes across all constants for consistency
- Add trailing commas in dictionaries and lists to improve diff readability
- Expand DIFFUSION_MODEL_BASE_MODELS with additional Wan Video and Qwen models
- Fix comment spacing in NSFW_LEVELS dictionary
- Maintain all existing functionality while improving code style
When force=true is passed via API, models in failed_models set are
re-downloaded instead of being skipped. On successful download, model is
removed from failed_models set.
This provides a manual batch repair mechanism for users when CivitAI
media server is temporarily down and causes empty folders.
Changes:
- Backend: Add force parameter to start_download(), _download_all_example_images(), _process_model()
- Backend: Skip failed_models check when force=true
- Backend: Remove model from failed_models on successful force retry
- Frontend: GlobalContextMenu now calls API with force=true directly
- Tests: Update mock to accept force parameter
Introduce a new RecipeFTSIndex class that provides fast prefix-based search across recipe fields (title, tags, LoRA names/models, prompts) using SQLite's FTS5 extension. The implementation supports sub-100ms search times for large datasets (20k+ recipes) and includes asynchronous indexing, incremental updates, and comprehensive unit tests.
The default_unet_root setting was not being synced from backend to frontend
because it was missing from the _SYNC_KEYS tuple in misc_handlers.py. This
caused the "Default Diffusion Model Root" setting to always display "No Default"
even when a valid path was configured in settings.json.
CivitAI does not distinguish between checkpoint and diffusion model types -
both are labeled as "checkpoint". For certain base model types like
"ZImageTurbo", all models are actually diffusion models and should be
saved to the unet/diffusion model folder instead of the checkpoint folder.
- Add DIFFUSION_MODEL_BASE_MODELS constant for known diffusion model types
- Add default_unet_root setting with auto-set logic
- Route downloads to unet folder when baseModel matches known diffusion types
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add failed_model_timestamps to track when models fail
- Retry failed models after 24-hour cooldown period
- Skip retry if example folder already has files
- Skip retry if failure was less than 24 hours ago
- Log count of failed models with retry message
- Fix unbound snapshot variable in exception path
- Remove duplicate/unreachable directory check code
- Update string quotes to double quotes (PEP 8)
This fixes the issue where failed models were permanently skipped in
auto-download mode, even when their example folders were empty.
- Fix infinite reinitialization loop by only validating stale widget.inputEl when it's actually in DOM
- Improve findWidgetInputElement to specifically search for textarea for text widgets, avoiding mismatches with checkbox inputs on nodes like WanVideo Lora Select that have toggle switches
- Add data-node-id based element search as primary strategy for better reliability across rendering modes
- Fix autocomplete initialization to properly handle element DOM state transitions
Fixes autocomplete failing after Canvas ↔ Vue DOM mode switches and WanVideo node always failing to trigger autocomplete.
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Update DebugMetadata node to return metadata via ComfyUI's UI system instead of websocket
- Add new JsonDisplayWidget Vue component for displaying metadata in the UI
- Remove dependency on PromptServer and websocket communication
- Improve error handling with proper UI feedback
- Maintain backward compatibility with existing metadata collection system
- Replace custom button creation and attachment logic with built-in actionBarButtons API
- Remove manual DOM manipulation for button positioning and retry logic
- Add custom styling and icon replacement for better visual integration
- Maintain existing functionality for opening LoRA Manager in same/new window
- Simplify extension setup by leveraging ComfyUI's extension system
When users paste CivArchive URLs, the system now fetches metadata from
CivArchive API first instead of Civitai. This prevents download failures
when a model has been deleted from Civitai but remains available on
CivArchive with alternative mirrors.
Changes:
- Source-aware metadata fetching: Uses CivArchive API when source='civarchive'
- URL prioritization: Prefers non-Civitai mirrors for CivArchive downloads
- Fallback mechanism: Falls back to default provider if CivArchive fails
Fixes#769
- Fix lora entry click-to-select broken after pointer events refactoring
- Only stopPropagation() after pointer moves beyond 3 pixel threshold
- This allows click events to fire on lora entries for selection
- Applied to all drag handlers: initDrag, initHeaderDrag, initReorderDrag
- Fix strength value display to always show 2 decimal places
- Use toFixed(2) when updating strength input during drag
- Ensures consistent display (e.g., "1.00" instead of "1", "1.40" instead of "1.4")
- Use pointer events (pointerdown/pointermove/pointerup/pointercancel) with proper capture
- Fix drag not updating strength values by avoiding re-renders during drag
- Fix cursor stuck in resize state by ensuring proper cleanup
- Fix cursor showing wrong icon on hover (should be pointer)
- Ensure strength values display fixed width with 2 decimal places
- Remove unnecessary data-capture-wheel attribute (no wheel adjustment in loras widget)
- Add font-variant-numeric: tabular-nums for consistent number display
This ensures loras widget works consistently in both Canvas and Vue DOM render modes.
- Add folder_include parameter support in backend API handlers
- Add folder_include to FilterCriteria and implement multi-folder filtering logic
- Update frontend to send all include folders instead of only the first
- Add tests for single/multiple include folders, include with exclude, and non-recursive filtering
In Vue DOM render mode, widget.inputEl is not in the DOM, causing autocomplete to fail. This commit:
- Adds findWidgetInputElement() helper to search DOM for actual input elements
- Checks if widget.inputEl is in document before using it
- Falls back to DOM search for Vue-rendered widgets using .lg-node-widget containers
- Implements async initialization with retry logic (20 attempts, 50ms interval)
- Adds debug logging for troubleshooting
- Prevents duplicate initialization with isInitializing flag
Fixes autocomplete functionality for Lora Loader nodes when ComfyUI uses Vue DOM rendering instead of canvas rendering.
Add data-capture-wheel attribute to SingleSlider and DualRangeSlider
components to prevent wheel events from being intercepted by the canvas
in ComfyUI's new Vue DOM render mode. This allows mouse wheel to work
for adjusting slider values while still enabling workflow zoom on
non-interactive widget areas.
Also update event handling to use pointer events with proper stop
propagation and pointer capture for reliable drag operations in both
rendering modes.
Update development guide with Section 8 documenting Vue DOM render mode
event handling patterns and best practices.
Add a clear button (X icon) to the search bars in BaseModelModal and TagsModal. The button appears when there is search text, and clicking it clears the search input and refocuses the search field.
Add support for respecting recommended strength values from LoRA usage_tips
when randomizing LoRA selection.
Features:
- New toggle setting to enable/disable recommended strength respect (default off)
- Scale range slider (0-2, default 0.5-1.0) to adjust recommended values
- Uses recommended strength × random(scale) when feature enabled
- Fallbacks to original Model/Clip Strength range when no recommendation exists
- Clip strength recommendations only apply when using Custom Range mode
Backend changes:
- Parse usage_tips JSON string to extract strength/clipStrength
- Apply scale factor to recommended values during randomization
- Pass new parameters through API route and node
Frontend changes:
- Update RandomizerConfig type with new properties
- Add new UI section with toggle and dual-range slider
- Wire up state management and event handlers
- No layout shift (removed description text)
Tests:
- Add tests for enabled/disabled recommended strength in API routes
- Add test verifying config passed to service
- All existing tests pass
Build: Include compiled Vue widgets
- Add WSL detection and Windows path conversion using wslpath
- Add Docker/Kubernetes detection via /.dockerenv and /proc/1/cgroup
- Implement clipboard fallback for containerized environments
- Update open_file_location handler to detect WSL/Docker before POSIX
- Update open_settings_location handler with same detection logic
- Add clipboard API integration with graceful fallback in frontend
- Add translations for clipboard feature across all 10 languages
- Add unit tests for _is_wsl(), _is_docker(), and _wsl_to_windows_path()
Fixes file manager opening failures in WSL and Docker environments.
Fix issue where mouse cursor flickers between 'grabbing' and 'default'
while dragging slider handles. The cursor now remains 'grabbing'
throughout the entire drag operation regardless of mouse position.
Changes:
- Add dynamic 'is-dragging' class to SingleSlider and DualRangeSlider
- Apply cursor: grabbing to root component when dragging state is active
Replace NODE_CLASS_MAPPINGS.update({...}) with direct assignment
to prevent ComfyUI Manager scanner from detecting test mock nodes
as actual plugin nodes.
The scanner.py pattern '_CLASS_MAPPINGS\.update\s*\(\s*{([^}]*)}\s*\)'
was matching test fixtures that use .update() to register mock nodes,
causing false positive conflict warnings.
- Add blank line after module docstring for better PEP 8 compliance
- Reformat long lines to adhere to 88-character limit using Black-style formatting
- Improve string consistency by using double quotes consistently
- Enhance readability of complex list comprehensions and method calls
- Maintain all existing functionality while improving code structure
