use a listener to recalculate trigger phrases when model or lora list changes

invokeai/frontend/web/src/app/store/middleware/listenerMiddleware/index.ts

@@ -55,6 +55,7 @@ import { addUpscaleRequestedListener } from 'app/store/middleware/listenerMiddle
 import { addWorkflowLoadRequestedListener } from 'app/store/middleware/listenerMiddleware/listeners/workflowLoadRequested';
 import type { AppDispatch, RootState } from 'app/store/store';
 
+import { addPromptTriggerListChanged } from './listeners/promptTriggerListChanged';
 import { addSetDefaultSettingsListener } from './listeners/setDefaultSettings';
 
 export const listenerMiddleware = createListenerMiddleware();
@@ -153,7 +154,8 @@ addFirstListImagesListener(startAppListening);
 // Ad-hoc upscale workflwo
 addUpscaleRequestedListener(startAppListening);
 
-// Dynamic prompts
+// Prompts
 addDynamicPromptsListener(startAppListening);
+addPromptTriggerListChanged(startAppListening);
 
 addSetDefaultSettingsListener(startAppListening);

invokeai/frontend/web/src/app/store/middleware/listenerMiddleware/listeners/promptTriggerListChanged.ts

@@ -0,0 +1,39 @@
+import { isAnyOf } from '@reduxjs/toolkit';
+import type { AppStartListening } from 'app/store/middleware/listenerMiddleware';
+import { loraAdded, loraRemoved } from 'features/lora/store/loraSlice';
+import { modelChanged, triggerPhrasesChanged } from 'features/parameters/store/generationSlice';
+import { modelsApi } from 'services/api/endpoints/models';
+
+const matcher = isAnyOf(loraAdded, loraRemoved, modelChanged);
+
+export const addPromptTriggerListChanged = (startAppListening: AppStartListening) => {
+  startAppListening({
+    matcher,
+    effect: async (action, { dispatch, getState, cancelActiveListeners }) => {
+      cancelActiveListeners();
+      const state = getState();
+      const { model: mainModel } = state.generation;
+      const { loras } = state.lora;
+
+      let triggerPhrases: string[] = [];
+
+      if (!mainModel) {
+        dispatch(triggerPhrasesChanged([]));
+        return;
+      }
+
+      const { data: mainModelData } = await dispatch(modelsApi.endpoints.getModelMetadata.initiate(mainModel.key));
+      triggerPhrases = mainModelData?.trigger_phrases || [];
+
+      for (let index = 0; index < Object.values(loras).length; index++) {
+        const lora = Object.values(loras)[index];
+        if (lora) {
+          const { data: loraData } = await dispatch(modelsApi.endpoints.getModelMetadata.initiate(lora.model.key));
+          triggerPhrases = [...triggerPhrases, ...(loraData?.trigger_phrases || [])];
+        }
+      }
+
+      dispatch(triggerPhrasesChanged(triggerPhrases));
+    },
+  });
+};

invokeai/frontend/web/src/features/parameters/store/generationSlice.ts

@@ -56,6 +56,7 @@ const initialGenerationState: GenerationState = {
   shouldUseCpuNoise: true,
   shouldShowAdvancedOptions: false,
   aspectRatio: { ...initialAspectRatioState },
+  triggerPhrases: [],
 };
 
 export const generationSlice = createSlice({
@@ -207,6 +208,9 @@ export const generationSlice = createSlice({
     aspectRatioChanged: (state, action: PayloadAction<AspectRatioState>) => {
       state.aspectRatio = action.payload;
     },
+    triggerPhrasesChanged: (state, action: PayloadAction<string[]>) => {
+      state.triggerPhrases = action.payload;
+    },
   },
   extraReducers: (builder) => {
     builder.addCase(configChanged, (state, action) => {
@@ -285,6 +289,7 @@ export const {
   heightChanged,
   widthRecalled,
   heightRecalled,
+  triggerPhrasesChanged,
 } = generationSlice.actions;
 
 export const { selectOptimalDimension } = generationSlice.selectors;

invokeai/frontend/web/src/features/parameters/store/types.ts

@@ -51,6 +51,7 @@ export interface GenerationState {
   shouldUseCpuNoise: boolean;
   shouldShowAdvancedOptions: boolean;
   aspectRatio: AspectRatioState;
+  triggerPhrases: string[];
 }
 
 export type PayloadActionWithOptimalDimension<T = void> = PayloadAction<T, string, { optimalDimension: number }>;

invokeai/frontend/web/src/features/prompt/PromptTriggerSelect.tsx

@@ -1,13 +1,13 @@
-import type { ChakraProps, ComboboxOnChange } from '@invoke-ai/ui-library';
+import type { ChakraProps, ComboboxOnChange, ComboboxOption } from '@invoke-ai/ui-library';
 import { Combobox, FormControl } from '@invoke-ai/ui-library';
-import { skipToken } from '@reduxjs/toolkit/query';
 import { useAppSelector } from 'app/store/storeHooks';
+import type { GroupBase, OptionsOrGroups } from 'chakra-react-select';
 import type { PromptTriggerSelectProps } from 'features/prompt/types';
 import { t } from 'i18next';
 import { map } from 'lodash-es';
 import { memo, useCallback, useMemo } from 'react';
 import { useTranslation } from 'react-i18next';
-import { useGetModelMetadataQuery, useGetTextualInversionModelsQuery } from 'services/api/endpoints/models';
+import { useGetTextualInversionModelsQuery } from 'services/api/endpoints/models';
 
 const noOptionsMessage = () => t('prompt.noMatchingTriggers');
 
@@ -15,10 +15,9 @@ export const PromptTriggerSelect = memo(({ onSelect, onClose }: PromptTriggerSel
   const { t } = useTranslation();
 
   const currentBaseModel = useAppSelector((s) => s.generation.model?.base);
-  const currentModelKey = useAppSelector((s) => s.generation.model?.key);
+  const triggerPhrases = useAppSelector((s) => s.generation.triggerPhrases);
 
   const { data, isLoading } = useGetTextualInversionModelsQuery();
-  const { data: metadata } = useGetModelMetadataQuery(currentModelKey ?? skipToken);
 
   const _onChange = useCallback<ComboboxOnChange>(
     (v) => {
@@ -32,34 +31,28 @@ export const PromptTriggerSelect = memo(({ onSelect, onClose }: PromptTriggerSel
     [onSelect]
   );
 
-  const embeddingOptions = useMemo(() => {
+  const options = useMemo(() => {
-    if (!data) {
+    let embeddingOptions: OptionsOrGroups<ComboboxOption, GroupBase<ComboboxOption>> = [];
-      return [];
-    }
 
+    if (data) {
       const compatibleEmbeddingsArray = map(data.entities).filter((model) => model.base === currentBaseModel);
 
-    return [
+      embeddingOptions = [
         {
           label: t('prompt.compatibleEmbeddings'),
           options: compatibleEmbeddingsArray.map((model) => ({ label: model.name, value: `<${model.name}>` })),
         },
       ];
-  }, [data, currentBaseModel, t]);
-
-  const options = useMemo(() => {
-    if (!metadata || !metadata.trigger_phrases) {
-      return [...embeddingOptions];
     }
 
     const metadataOptions = [
       {
         label: t('modelManager.triggerPhrases'),
-        options: metadata.trigger_phrases.map((phrase) => ({ label: phrase, value: phrase })),
+        options: triggerPhrases.map((phrase) => ({ label: phrase, value: phrase })),
       },
     ];
     return [...metadataOptions, ...embeddingOptions];
-  }, [embeddingOptions, metadata, t]);
+  }, [data, currentBaseModel, triggerPhrases, t]);
 
   return (
     <FormControl>