admin/infinite-canvas
1
0
Fork 0
mirror of https://github.com/basketikun/infinite-canvas.git synced 2026-08-05 17:04:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat(config): enhance model configuration with support for audio models and improve UI for user preferences

Browse Source
This commit is contained in:
HouYunFei
2026-06-02 17:24:27 +08:00
parent 39793e890d
commit 38ec654748
4 changed files with 151 additions and 34 deletions
Download Patch File Download Diff File Expand all files Collapse all files
web/src/stores/use-config-store.ts
+61 -5
View File
@@ -15,12 +15,17 @@ export type AiConfig = {
imageModel: string;
videoModel: string;
textModel: string;
audioModel: string;
videoSeconds: string;
vquality: string;
videoGenerateAudio: string;
videoWatermark: string;
systemPrompt: string;
models: string[];
imageModels: string[];
videoModels: string[];
textModels: string[];
audioModels: string[];
quality: string;
size: string;
count: string;
@@ -28,7 +33,7 @@ export type AiConfig = {
};
export const CONFIG_STORE_KEY = "infinite-canvas:ai_config_store";
export type ModelCapability = "image" | "video" | "text";
export type ModelCapability = "image" | "video" | "text" | "audio";
export const defaultConfig: AiConfig = {
channelMode: "local",
@@ -38,12 +43,17 @@ export const defaultConfig: AiConfig = {
imageModel: "gpt-image-2",
videoModel: "grok-imagine-video",
textModel: "gpt-5.5",
audioModel: "",
videoSeconds: "6",
vquality: "720",
videoGenerateAudio: "true",
videoWatermark: "false",
systemPrompt: "",
models: [],
imageModels: [],
videoModels: [],
textModels: [],
audioModels: [],
quality: "auto",
size: "1:1",
count: "1",
@@ -71,18 +81,25 @@ function resolveEffectiveConfig(config: AiConfig, modelChannel: AdminPublicSetti
const textModels = filterModelsByCapability(models, "text");
const imageModels = filterModelsByCapability(models, "image");
const videoModels = filterModelsByCapability(models, "video");
const audioModels = filterModelsByCapability(models, "audio");
const fallbackTextModel = validDefault(modelChannel.defaultTextModel, textModels) || preferredModel(textModels, isTextModelName);
const fallbackModel = validDefault(modelChannel.defaultModel, textModels) || fallbackTextModel;
const fallbackImageModel = validDefault(modelChannel.defaultImageModel, imageModels) || preferredModel(imageModels, isImageModelName);
const fallbackVideoModel = validDefault(modelChannel.defaultVideoModel, videoModels) || preferredModel(videoModels, isVideoModelName);
const fallbackAudioModel = preferredModel(audioModels, isAudioModelName);
return {
...config,
channelMode,
models,
imageModels,
videoModels,
textModels,
audioModels,
model: textModels.includes(config.model) ? config.model : fallbackModel,
imageModel: imageModels.includes(config.imageModel) ? config.imageModel : fallbackImageModel,
videoModel: videoModels.includes(config.videoModel) ? config.videoModel : fallbackVideoModel,
textModel: textModels.includes(config.textModel) ? config.textModel : fallbackTextModel || fallbackModel,
audioModel: audioModels.includes(config.audioModel) ? config.audioModel : fallbackAudioModel,
systemPrompt: modelChannel.systemPrompt,
};
}
@@ -102,17 +119,23 @@ function isVideoModelName(model: string) {
function isImageModelName(model: string) {
const value = model.toLowerCase();
return !isVideoModelName(model) && (value.includes("seedream") || value.includes("gpt-image") || value.includes("image") || value.includes("dall-e") || value.includes("dalle") || value.includes("imagen") || value.includes("flux") || value.includes("sdxl") || value.includes("stable-diffusion") || value.includes("midjourney"));
return !isVideoModelName(model) && !isAudioModelName(model) && (value.includes("seedream") || value.includes("gpt-image") || value.includes("image") || value.includes("dall-e") || value.includes("dalle") || value.includes("imagen") || value.includes("flux") || value.includes("sdxl") || value.includes("stable-diffusion") || value.includes("midjourney"));
}
function isAudioModelName(model: string) {
const value = model.toLowerCase();
return value.includes("audio") || value.includes("tts") || value.includes("speech") || value.includes("whisper") || value.includes("transcribe") || value.includes("voice") || value.includes("music") || value.includes("sound");
}
function isTextModelName(model: string) {
return !isImageModelName(model) && !isVideoModelName(model);
return !isImageModelName(model) && !isVideoModelName(model) && !isAudioModelName(model);
}
export function modelMatchesCapability(model: string, capability?: ModelCapability) {
if (!capability) return true;
if (capability === "image") return isImageModelName(model);
if (capability === "video") return isVideoModelName(model);
if (capability === "audio") return isAudioModelName(model);
return isTextModelName(model);
}
@@ -120,6 +143,15 @@ export function filterModelsByCapability(models: string[], capability?: ModelCap
return capability ? models.filter((model) => modelMatchesCapability(model, capability)) : models;
}
export function selectableModelsByCapability(config: AiConfig, capability?: ModelCapability) {
if (!capability) return config.models;
return config[modelListKey(capability)];
}
function modelListKey(capability: ModelCapability) {
return `${capability}Models` as "imageModels" | "videoModels" | "textModels" | "audioModels";
}
function isAiConfigReady(config: AiConfig, model: string) {
return Boolean(model.trim()) && (config.channelMode === "remote" || Boolean(config.baseUrl.trim() && config.apiKey.trim()));
}
@@ -157,13 +189,37 @@ export const useConfigStore = create<ConfigStore>()(
name: CONFIG_STORE_KEY,
partialize: (state) => ({ config: state.config }),
merge: (persisted, current) => {
const config = { ...defaultConfig, ...((persisted as Partial<ConfigStore>).config || {}) };
return { ...current, config: { ...config, channelMode: config.channelMode || "remote", imageModel: config.imageModel || config.model, videoModel: config.videoModel || "grok-imagine-video", textModel: config.textModel || config.model, videoSeconds: config.videoSeconds || "6", vquality: config.vquality || "720", videoGenerateAudio: config.videoGenerateAudio || "true", videoWatermark: config.videoWatermark || "false", canvasImageCount: config.canvasImageCount || "3" } };
const persistedConfig = ((persisted as Partial<ConfigStore>).config || {}) as Partial<AiConfig>;
const config = { ...defaultConfig, ...persistedConfig };
return {
...current,
config: {
...config,
channelMode: config.channelMode || "remote",
imageModel: config.imageModel || config.model,
videoModel: config.videoModel || "grok-imagine-video",
textModel: config.textModel || config.model,
audioModel: config.audioModel || "",
videoSeconds: config.videoSeconds || "6",
vquality: config.vquality || "720",
videoGenerateAudio: config.videoGenerateAudio || "true",
videoWatermark: config.videoWatermark || "false",
canvasImageCount: config.canvasImageCount || "3",
imageModels: Array.isArray(persistedConfig.imageModels) ? normalizeModelList(config.imageModels) : filterModelsByCapability(config.models, "image"),
videoModels: Array.isArray(persistedConfig.videoModels) ? normalizeModelList(config.videoModels) : filterModelsByCapability(config.models, "video"),
textModels: Array.isArray(persistedConfig.textModels) ? normalizeModelList(config.textModels) : filterModelsByCapability(config.models, "text"),
audioModels: Array.isArray(persistedConfig.audioModels) ? normalizeModelList(config.audioModels) : filterModelsByCapability(config.models, "audio"),
},
};
},
},
),
);
function normalizeModelList(models: string[]) {
return Array.from(new Set((models || []).map((model) => model.trim()).filter(Boolean)));
}
export function useEffectiveConfig() {
const config = useConfigStore((state) => state.config);
const modelChannel = useConfigStore((state) => state.publicSettings?.modelChannel || null);