Add metadata service integration to automatically resolve base model information from checkpoint metadata during recipe import. This replaces the previous approach of relying solely on request parameters and provides more accurate base model information.
- Add _resolve_base_model_from_checkpoint method to fetch base model from metadata provider
- Update recipe import logic to use resolved base model when available
- Add comprehensive tests for base model resolution with fallback behavior
- Remove debug print statement from import parameters
- Import getBaseModelAbbreviation utility function
- Add fallback handling for missing base model values
- Display abbreviated base model names while keeping full name in tooltip
- Maintain "Unknown" label for recipes without base model specification
- Improve user experience by showing cleaner, more readable model identifiers
Add _get_lora_from_version_index method to fetch cached LoRA entries by modelVersionId when hash is unavailable. This improves LoRA enrichment by using version index as fallback when hash is missing, ensuring proper library status, file paths, and preview URLs are set even without hash values.
Update test suite to include version_index in stub cache and add test coverage for version-based lookup functionality.
Add support for importing recipes from remote sources by:
- Adding import_remote_recipe endpoint to RecipeHandlerSet
- Injecting downloader_factory and civitai_client_getter dependencies
- Implementing image download and resource parsing logic
- Supporting Civitai resource payloads with checkpoints and LoRAs
- Adding required imports for regex and temporary file handling
This enables users to import recipes directly from external sources like Civitai without manual file downloads.
- Add auto_organize_exclusions to settings handler proxy keys
- Refactor model file service to handle exclusions relative to model roots
- Improve auto-organize progress reporting for empty operations
- Fix exclusion pattern matching to consider relative paths within model roots
- Ensure proper validation when no model roots are configured
- Add comprehensive cleanup reporting for empty auto-organize operations
- Import rewrite_preview_url utility for optimized image URL handling
- Update thumbnail URL processing for both LoRA and checkpoint entries to use rewritten URLs
- Expand checkpoint metadata with modelId, file size, SHA256 hash, and file name
- Improve error handling and data validation for Civitai API responses
- Maintain backward compatibility with existing data structures
- Extend CSS to style video elements in delete previews
- Add video rendering logic for model version previews
- Use consistent placeholder image for missing previews
- Maintain existing image preview functionality while adding video support
This allows users to see video previews when deleting model versions, improving the user experience for video-based models.
Add shouldShowLicenseFilters method to check if current page is 'recipes' and skip license filter initialization and updates when on recipes page. Also conditionally render license filter section in header template based on current page.
This prevents license filters from appearing on the recipes page where they are not applicable.
- Track when Civitai API returns "Model not found" for default provider
- Use dedicated flag instead of error string comparison for deletion detection
- Ensure archive-sourced models don't get marked as deleted
- Add test coverage for archive source deletion flag behavior
- Fix deletion flag logic to properly handle provider fallback scenarios
- Add optional main_extension parameter to delete_model_artifacts function
- Extract file extension from model filename to handle different file types
- Update model scanner to pass file extension when deleting models
- Add test case for GGUF file deletion to ensure proper cleanup
- Maintain backward compatibility with existing safetensors models
This change allows the model lifecycle service to properly delete GGUF model files along with their associated metadata and preview files, expanding support beyond just safetensors format.
Add ModelUpdateService dependency to ModelLifecycleService to enable synchronization during model deletion. The service is now passed through BaseModelRoutes initialization and used in delete_model to trigger updates when a model is removed. This ensures external systems stay in sync with local model state changes.
Key changes:
- Inject update_service into ModelLifecycleService constructor
- Extract model ID from metadata during deletion
- Call update service sync method after successful deletion
- Add proper type hints and TYPE_CHECKING imports
- Add model_types parameter to ModelListingHandler to support filtering by model type
- Implement get_model_types endpoint in ModelQueryHandler to retrieve available model types
- Register new /api/lm/{prefix}/model-types route for model type queries
- Extend BaseModelService to handle model type filtering in queries
- Support both model_type and civitai_model_type query parameters for backward compatibility
This enables users to filter models by specific types, improving model discovery and organization capabilities.
- Remove unused isNewer variable calculation
- Use dividerThresholdVersionId instead of latestLibraryVersionId for badge logic
- Add test case to verify newer version badge appears with same-base strategy
- Ensures correct badge display when filtering by same base model versions
Add four new license restriction keys to all locale files:
- noImageSell: "No selling generated content"
- noRentCivit: "No Civitai generation"
- noRent: "No generation services"
- noSell: "No selling models"
These additions provide comprehensive coverage for various commercial and generation restrictions in the licensing system, ensuring proper localization across all supported languages.
Add CSS classes and JavaScript logic for the base filter toggle button in the versions toolbar. The filter allows users to switch between showing all versions or only versions matching the current base model. Includes styling for different states (active, hover, disabled) and accessibility features like screen reader support.
Add new "updateFlags" section to settings navigation and implement update flag strategy configuration. The strategy allows users to choose when update badges appear:
- Match updates by base model (only show when new release shares same base model)
- Flag any available update (show whenever newer version exists)
Includes translations for English, German, Spanish, and French locales.
Add support for configurable update flag strategy with new "same_base" mode that considers base model versions when determining update availability. The strategy is controlled by the "update_flag_strategy" setting.
When strategy is set to "same_base":
- Uses get_records_bulk instead of has_updates_bulk
- Compares model versions against highest local versions per base model
- Provides more granular update detection based on base model relationships
Fallback to existing bulk or individual update checks when:
- Strategy is not "same_base"
- Bulk operations fail
- Records are unavailable
This enables more precise update flagging for models sharing common bases.
- Add CheckpointLoaderKJ to NODE_EXTRACTORS mapping for KJNodes support
- Enhance model filename generation in SaveImage to handle different data types
- Add proper type checking and fallback for model metadata values
- Improve robustness when processing checkpoint paths for filename generation
Update trigger word validation message across all language files to reflect increased character limit. The change allows users to create longer trigger words, providing more flexibility in trigger word creation while maintaining the existing maximum count of 30 trigger words.
Change `load_metadata` method to return a tuple containing both the metadata object and a boolean success flag instead of just the metadata object. This provides clearer error handling and allows callers to distinguish between successful loads with null metadata versus actual load failures.
Add `allow_strength_adjustment` parameter to enable mouse wheel adjustment of trigger word strengths. When enabled, strength values are preserved and can be modified interactively. Also improves trigger word parsing by handling whitespace more consistently and adding debug logging for trigger data inspection.
Remove visual styling for tags with modified strength values. The gold border and gradient background were previously applied to tags with strength values other than 1.0, but this visual distinction is no longer needed. This simplifies the tag styling logic and maintains consistent appearance across all tags regardless of their strength values.
Change from loadedGraphNode to nodeCreated lifecycle method to ensure proper widget initialization timing. Wrap widget creation and highlight logic in requestAnimationFrame to prevent race conditions with node setup. This ensures the trigger word toggle widget functions correctly when nodes are created.
- Add optional license_flags parameter to model preview API endpoint
- Include license flags in response when requested via query parameter
- Add CSS styles for license overlay and icons in tooltip
- Implement license flag parsing and icon mapping logic
- Display license restrictions as icons in preview tooltip overlay
This allows users to see model license restrictions directly in the preview tooltip without needing to navigate to detailed model information pages.
- Change lifecycle hook from nodeCreated to loadedGraphNode in Lora Loader, Lora Stacker, and TriggerWord Toggle nodes
- Remove requestAnimationFrame wrappers as loadedGraphNode ensures proper initialization timing
- Maintain same functionality for restoring saved values and widget initialization
- Improves reliability by using the appropriate node lifecycle event
- Update frontend tag filter to cycle through include/exclude/clear states
- Add backend support for tag_include and tag_exclude query parameters
- Maintain backward compatibility with legacy tag parameter
- Store tag states as dictionary with 'include'/'exclude' values
- Update test matrix documentation to reflect new tag behavior
The changes enable more granular tag filtering where users can now explicitly include or exclude specific tags, rather than just adding tags to a simple inclusion list. This provides better control over search results and improves the filtering user experience.
Add new translation keys for model filter interface:
- license
- noCreditRequired
- allowSellingGeneratedContent
These labels support new filtering options for model licensing and content usage permissions, enabling users to filter models based on their license requirements and commercial usage rights.
Add support for filtering models by license requirements:
- credit_required: filter models that require credits or allow free use
- allow_selling_generated_content: filter models based on commercial usage rights
These filters use license_flags bitmask to determine model permissions and enable users to find models that match their specific usage requirements and budget constraints.
- Extract and preserve strength values from trigger words in format "(word:strength)"
- Maintain strength formatting when filtering active trigger words in both group and individual modes
- Update active state tracking to handle strength-modified words correctly
- Ensure backward compatibility with existing trigger word formats
- Import applySelectionHighlight in lora_loader and lora_stacker
- Pass onSelectionChange callback to loras_widget to handle selection changes
- Implement selection tracking and payload building in loras_widget
- Emit selection changes when LoRA selection is modified
- Update tags_widget to support highlighted tag styling
This provides visual feedback when LoRAs are selected by highlighting associated trigger words in the interface.
Add property descriptor to listen for mode changes in Lora Loader and Lora Stacker nodes. When node mode changes, automatically update connected trigger word toggle nodes and downstream loader nodes to maintain synchronization between node modes and trigger word states.
- Lora Loader: Updates connected trigger words when mode changes
- Lora Stacker: Updates connected trigger words and downstream loaders when mode changes
- Both nodes log mode changes for debugging purposes
Add node mode checks to ensure trigger words are only updated when the stacker node is active (mode 0 for Always or mode 3 for On Trigger). This prevents unnecessary updates when the node is inactive (mode 2 for Never or mode 4 for Bypass), improving performance and ensuring trigger words reflect the actual active state of the node.
The changes include:
- Adding mode checks before updating active LoRA names in the stacker callback
- Modifying collectActiveLorasFromChain to only include active nodes
- Adding comments to clarify node mode behavior
Add RateLimitRetryingProvider and _RateLimitRetryHelper classes to handle rate limiting with exponential backoff retries. Update get_metadata_provider function to automatically wrap providers with rate limit handling. This improves reliability when external APIs return rate limit errors by implementing automatic retries with configurable delays and jitter.
Add hard reset and clean operations to ensure a clean working directory
before switching branches or checking out release tags. This prevents
local changes from interfering with the update process and ensures
consistent behavior across both nightly and release update paths.
Add license-related translation keys including credit requirements, derivative restrictions, and license sharing permissions. This supports displaying proper license information and restrictions in the UI across all supported languages (DE, EN, ES, FR, HE, JA, KO, RU).
- Simplify error handling logic by checking for "not found" message directly
- Extract model data fetching into separate _fetch_model_data method
- Extract version enrichment into separate _enrich_version_with_model_data method
- Improve logging consistency using %s formatting
- Rename variables for better clarity (result -> version, e -> exc)
- Update DEFAULT_LICENSE_FLAGS from 57 to 127 to enable all commercial modes by default
- Replace CommercialUseLevel enum with bitwise commercial permission handling
- Simplify commercial value normalization and validation using allowed values set
- Adjust bit shifting in license flag construction to accommodate new commercial bits structure
- Remove CommercialUseLevel from exports and update tests accordingly
- Improve handling of empty commercial use values with proper type checking
The changes streamline commercial permission processing and align with CivitAI's default license configuration while maintaining backward compatibility.
Update license restriction labels in ModelModal component to be more descriptive and user-friendly. Changed fallback text and translation keys for various license restrictions including:
- Selling models
- Generation services
- Civitai generation
- Selling generated content
- Creator credit requirements
- Sharing merges
- Permission requirements
The changes make the license restrictions more clear and specific about what actions are prohibited or required.
Update the --lora-border CSS custom property to use a darker, more opaque color. The previous border color was too subtle and lacked sufficient contrast against the background. This change improves visual hierarchy and makes interface elements more distinguishable.
Add visual indicators for commercial license restrictions in the model modal. New CSS classes and JavaScript utilities handle the display of restriction icons for selling, renting, and image usage limitations. The modal header actions container has been restructured to accommodate the new license restriction section.
- Add `.modal-header-actions` and `.license-restrictions` CSS classes
- Implement commercial license icon configuration and rendering logic
- Normalize and sanitize commercial restriction values
- Update header layout to remove bottom margin for better visual alignment
Add license resolution utilities and integrate license information into model metadata processing. The changes include:
- Add `resolve_license_payload` function to extract license data from Civitai model responses
- Integrate license information into model metadata in CivitaiClient and MetadataSyncService
- Add license flags support in model scanning and caching
- Implement CommercialUseLevel enum for standardized license classification
- Update model scanner to handle unknown fields when extracting metadata values
This ensures proper license attribution and compliance when working with Civitai models.
Convert all tags to lowercase in tag processing logic to prevent case sensitivity issues on Windows filesystems. This ensures consistent tag matching and prevents duplicate tags with different cases from being created.
Changes include:
- TagUpdateService now converts tags to lowercase before comparison
- Utils function converts model tags to lowercase before priority resolution
- Test cases updated to reflect lowercase tag expectations
Add duplicate detection to banner recording to prevent multiple entries
for the same banner ID in recent history. This prevents duplicate history
entries when pages refresh or banners are shown multiple times.
- Check if banner ID already exists in recentHistory before adding
- Return early if duplicate found to prevent adding same banner multiple times
- Add comprehensive tests for banner history functionality including:
- Adding new banners to history
- Preventing duplicate entries
- Handling multiple different banners
- Clear history between tests to ensure test isolation
- Remove legacy community support banner tracking variables and logic
- Simplify banner dismissal handling by checking dismissal state before marking
- Replace timer-based community support banner with immediate registration
- Clean up unused constants and legacy storage keys
- Add comprehensive test suite with mocked dependencies
- Improve code maintainability and test coverage
- Update zh-CN locale with Chinese text for community support section
- Add support for Afdian platform for Chinese users alongside existing Ko-fi
- Implement language-based URL routing for support links and tutorials
- Chinese users now see localized content with appropriate payment options (Alipay/WeChat)
- Maintains existing functionality for non-Chinese users
- Add onDragStart and onDragEnd callbacks to initDrag function
- Implement preview suppression during and briefly after strength dragging
- Clear preview timer on drag start/end to prevent tooltip conflicts
- Update tests to verify drag callbacks are properly triggered
This prevents tooltip previews from interfering with drag interactions and provides better control over drag lifecycle events.
Add test coverage for trigger word refresh functionality when LoRA syntax is edited in the input widget. The test verifies that after modifying LoRA syntax in the input field, the connected trigger words are properly updated to reflect the active LoRAs.
Additionally, implement the actual trigger word update logic in lora_loader.js by calling updateConnectedTriggerWords after merging LoRAs, ensuring the UI stays synchronized with the current LoRA state.
Remove redundant details about settings configuration to make installation steps clearer and more concise. The simplified instructions now focus on essential steps without unnecessary explanations about placeholder values and automatic registry generation.
- Add `forceInitialize` option to sidebar initialization to bypass disabled setting
- Refactor sidebar toggle logic to handle initialization promises more reliably
- Improve cleanup behavior when sidebar is disabled
- Ensure proper DOM updates when sidebar state changes
- Maintain container layout consistency during sidebar operations
Will Miao