Replace graph reconfigure with targeted widget updates so applying resolved models does not re-trigger node context analysis or duplicate dialog refreshes. Add workflow refresh suppression state, defer ComfyUI model catalog refresh until after widget linking, and track missing-workflow signatures separately from general workflow changes.
- Skip fuzzy matching for already-resolved model references in the backend
- Keep existing nodes in their previous result positions during incremental workflow analysis
- Simplify frontend workflow sync and add a content-preserving refresh path
- Preserve scroll position and suppress progress overlay during background Missing/Loaded Models refresh
- Add backend and frontend tests for the new incremental refresh behavior
- Preserve existing image metadata when re-normalizing CivitAI model images
- Add extractComfyWorkflow helper to recover pasteable editor workflows from image metadata, including nested/double-encoded JSON and non-finite number tolerance
- Surface recovered workflows as a copyable button in the model info image preview under Other metadata
- Add focused tests for workflow extraction, metadata preservation, and edge cases
Replace the Windows-only Explorer behavior with a host-file-manager
implementation that supports Windows, macOS, and Linux. The endpoint
now validates input, handles frontend backslash paths on POSIX, returns
structured errors, and reports the launcher/platform. Update README
wording and add tests for the new module.
Queue workflow model reference selection requests in the dialog, wait for
workflow analysis to complete before attempting selection, and handle
superseded or timed-out requests gracefully.
- Track pending selection requests on the dialog with status fields
- Apply queued selections from lifecycle graph methods after analysis
- Await loadWorkflowData/loadLoadedModels so the caller can proceed
after tab content is actually ready
- Add timeout handling with user-visible warnings when selection cannot
complete
- Cover queueing, superseding, waiting, and completion in JS tests
Treat control_after_generate widgets that are implicitly inserted after seed
widgets as non-model widgets when analyzing dynamic node layouts. This
prevents the resolver from attempting to resolve control values such as
'fixed', 'randomize', 'increment', and 'decrement' as model paths.
Treat widget values under the configs and custom_nodes dynamic
categories as non-model references so the workflow analyzer does
not falsely mark them as missing models. Added a regression test
to verify a checkpoint widget is still detected when a config
widget is present.
Backend:
- Keep CivArchive source page URL distinct from HuggingFace download mirrors in download metadata
- Handle CivitAI hash lookup HTTP errors and unexpected JSON shapes without caching failures
- Allow network JSON requests to return arrays as well as objects
Frontend:
- Escape each item individually when interpolating arrays into HTML templates
- Prefer the selected provider's page URL over cross-provider mirrors in context menus and download targets
- Reuse source link resolution for model info display
Tests:
- Add coverage for CivArchive metadata URL isolation, HTML array escaping, provider array responses, and CivitAI hash lookup caching behavior
Generic widget names such as model_name are no longer treated as local
model references by default. Only folder-backed schemas, known widget
indexes, or a model-typed combo/output pairing now admit an extensionless
value.
This prevents metadata/search nodes (for example Civitai Hash Fetcher)
with a model_name STRING input from being incorrectly reported as missing
models.
Added test_model_name_string_used_as_metadata_query_is_ignored to cover
the regression.
- Accept explicit model URNs without requiring model widget metadata
- Restrict model scanning to widgets with category/model hints or model-like filenames
- Remove over-broad single-widget-value detection path
- Add tests for cached model text, custom selectors, and constant/model slot edge cases
Add openResolverForDetectedMissingModels() in model_resolver.js to
keep the current resolver display state when missing models are
detected: if the resolver dialog is already visible, schedule an
active workflow refresh instead of re-activating the button; if it
is hidden, activate the resolver button as before.
Update the auto-open-on-missing path to use the new method, and
adjust/add tests to cover the preserved display mode behavior.
Dariusz L