mulmocast 0.1.2 → 0.1.4

package/lib/types/schema.js

@@ -1,4 +1,5 @@
 import { z } from "zod";
+import { htmlLLMProvider, provider2TTSAgent, provider2ImageAgent, provider2MovieAgent, defaultProviders } from "../utils/provider2agent.js";
 export const langSchema = z.string();
 const URLStringSchema = z.string().url();
 export const localizedTextSchema = z
@@ -20,7 +21,7 @@ export const speechOptionsSchema = z
 })
     .strict();
 const speakerIdSchema = z.string();
-export const text2SpeechProviderSchema = z.enum(["openai", "nijivoice", "google", "elevenlabs"]).default("openai");
+export const text2SpeechProviderSchema = z.enum(Object.keys(provider2TTSAgent)).default(defaultProviders.tts);
 export const speakerDataSchema = z
     .object({
     displayName: z.record(langSchema, z.string()).optional(),
@@ -153,25 +154,31 @@ const mulmoMidiMediaSchema = z
     .strict();
 export const mulmoAudioAssetSchema = z.union([mulmoAudioMediaSchema, mulmoMidiMediaSchema]);
 const imageIdSchema = z.string();
-export const mulmoImageParamsImagesSchema = z.record(imageIdSchema, mulmoImageMediaSchema);
+export const mulmoImagePromptMediaSchema = z
+    .object({
+    type: z.literal("imagePrompt"),
+    prompt: z.string(),
+})
+    .strict();
+export const mulmoImageParamsImagesSchema = z.record(imageIdSchema, z.union([mulmoImageMediaSchema, mulmoImagePromptMediaSchema]));
 export const mulmoFillOptionSchema = z
     .object({
     style: z.enum(["aspectFit", "aspectFill"]).default("aspectFit"),
 })
     .describe("How to handle aspect ratio differences between image and canvas");
-export const text2ImageProviderSchema = z.enum(["openai", "google"]).default("openai");
+export const text2ImageProviderSchema = z.enum(Object.keys(provider2ImageAgent)).default(defaultProviders.text2image);
 // NOTE: This is for UI only. (until we figure out how to use it in mulmoImageParamsSchema)
 export const mulmoOpenAIImageModelSchema = z
     .object({
     provider: z.literal("openai"),
-    model: z.enum(["dall-e-3", "gpt-image-1"]).optional(),
+    model: z.enum(provider2ImageAgent["openai"].models).optional(),
 })
     .strict();
 // NOTE: This is for UI only. (until we figure out how to use it in mulmoImageParamsSchema)
 export const mulmoGoogleImageModelSchema = z
     .object({
     provider: z.literal("google"),
-    model: z.enum(["imagen-3.0-fast-generate-001", "imagen-3.0-generate-002", "imagen-3.0-capability-001"]).optional(),
+    model: z.enum(provider2ImageAgent["google"].models).optional(),
 })
     .strict();
 export const mulmoImageParamsSchema = z
@@ -232,6 +239,7 @@ export const mulmoBeatSchema = z
     audioParams: beatAudioParamsSchema.optional(), // beat specific parameters
     movieParams: z
         .object({
+        model: z.string().optional(),
         fillOption: mulmoFillOptionSchema.optional(),
         speed: z.number().optional().describe("Speed of the video. 1.0 is normal speed. 0.5 is half speed. 2.0 is double speed."),
     })
@@ -265,20 +273,20 @@ export const mulmoSpeechParamsSchema = z
     speakers: speakerDictionarySchema,
 })
     .strict();
-export const text2HtmlImageProviderSchema = z.enum(["openai", "anthropic"]).default("openai");
-export const text2MovieProviderSchema = z.enum(["google", "replicate"]).default("google");
+export const text2HtmlImageProviderSchema = z.enum(htmlLLMProvider).default(defaultProviders.text2Html);
+export const text2MovieProviderSchema = z.enum(Object.keys(provider2MovieAgent)).default(defaultProviders.text2movie);
 // NOTE: This is UI only. (until we figure out how to use it in mulmoMovieParamsSchema)
 export const mulmoGoogleMovieModelSchema = z
     .object({
     provider: z.literal("google"),
-    model: z.enum(["veo-2.0-generate-001"]).optional(),
+    model: z.enum(provider2MovieAgent.google.models).optional(),
 })
     .strict();
 // NOTE: This is UI only. (until we figure out how to use it in mulmoMovieParamsSchema)
 export const mulmoReplicateMovieModelSchema = z
     .object({
     provider: z.literal("replicate"),
-    model: z.enum(["bytedance/seedance-1-lite", "kwaivgi/kling-v2.1", "google/veo-3"]).optional(),
+    model: z.enum(provider2MovieAgent.replicate.models).optional(),
 })
     .strict();
 export const mulmoTransitionSchema = z.object({
@@ -329,7 +337,7 @@ export const mulmoReferenceSchema = z.object({
     url: URLStringSchema,
     title: z.string().optional(),
     description: z.string().optional(),
-    type: z.enum(["article", "paper", "image", "video", "audio"]).default("article"),
+    type: z.union([z.enum(["article", "paper", "image", "video", "audio"]), z.string()]).default("article"),
 });
 export const mulmoScriptSchema = mulmoPresentationStyleSchema
     .extend({
@@ -378,6 +386,7 @@ export const mulmoSessionStateSchema = z.object({
         multiLingual: z.record(z.number().int(), z.boolean()),
         caption: z.record(z.number().int(), z.boolean()),
         html: z.record(z.number().int(), z.boolean()),
+        imageReference: z.record(z.number().int(), z.boolean()),
     }),
 });
 export const mulmoStudioSchema = z

package/lib/types/type.d.ts

@@ -1,4 +1,4 @@
-import { langSchema, localizedTextSchema, mulmoBeatSchema, mulmoScriptSchema, mulmoStudioSchema, mulmoStudioBeatSchema, mulmoStoryboardSchema, mulmoStoryboardSceneSchema, mulmoStudioMultiLingualSchema, mulmoStudioMultiLingualDataSchema, speakerDictionarySchema, mulmoImageParamsSchema, mulmoImageParamsImagesSchema, mulmoFillOptionSchema, mulmoMovieParamsSchema, mulmoSpeechParamsSchema, textSlideParamsSchema, speechOptionsSchema, speakerDataSchema, mulmoCanvasDimensionSchema, mulmoScriptTemplateSchema, mulmoScriptTemplateFileSchema, text2ImageProviderSchema, text2HtmlImageProviderSchema, text2MovieProviderSchema, text2SpeechProviderSchema, mulmoPresentationStyleSchema, multiLingualTextsSchema, mulmoMermaidMediaSchema, mulmoTextSlideMediaSchema, mulmoMarkdownMediaSchema, mulmoImageMediaSchema, mulmoChartMediaSchema, mediaSourceSchema, mulmoSessionStateSchema, mulmoOpenAIImageModelSchema, mulmoGoogleImageModelSchema, mulmoGoogleMovieModelSchema, mulmoReplicateMovieModelSchema } from "./schema.js";
+import { langSchema, localizedTextSchema, mulmoBeatSchema, mulmoScriptSchema, mulmoStudioSchema, mulmoStudioBeatSchema, mulmoStoryboardSchema, mulmoStoryboardSceneSchema, mulmoStudioMultiLingualSchema, mulmoStudioMultiLingualDataSchema, speakerDictionarySchema, mulmoImageParamsSchema, mulmoImageParamsImagesSchema, mulmoFillOptionSchema, mulmoMovieParamsSchema, mulmoSpeechParamsSchema, textSlideParamsSchema, speechOptionsSchema, speakerDataSchema, mulmoCanvasDimensionSchema, mulmoScriptTemplateSchema, mulmoScriptTemplateFileSchema, text2ImageProviderSchema, text2HtmlImageProviderSchema, text2MovieProviderSchema, text2SpeechProviderSchema, mulmoPresentationStyleSchema, multiLingualTextsSchema, mulmoMermaidMediaSchema, mulmoTextSlideMediaSchema, mulmoMarkdownMediaSchema, mulmoImageMediaSchema, mulmoChartMediaSchema, mediaSourceSchema, mulmoSessionStateSchema, mulmoOpenAIImageModelSchema, mulmoGoogleImageModelSchema, mulmoGoogleMovieModelSchema, mulmoReplicateMovieModelSchema, mulmoImagePromptMediaSchema } from "./schema.js";
 import { pdf_modes, pdf_sizes, storyToScriptGenerateMode } from "../utils/const.js";
 import { LLM } from "../utils/utils.js";
 import { z } from "zod";
@@ -35,6 +35,7 @@ export type MulmoOpenAIImageModel = z.infer<typeof mulmoOpenAIImageModelSchema>;
 export type MulmoGoogleImageModel = z.infer<typeof mulmoGoogleImageModelSchema>;
 export type MulmoGoogleMovieModel = z.infer<typeof mulmoGoogleMovieModelSchema>;
 export type MulmoReplicateMovieModel = z.infer<typeof mulmoReplicateMovieModelSchema>;
+export type MulmoImagePromptMedia = z.infer<typeof mulmoImagePromptMediaSchema>;
 export type MulmoTextSlideMedia = z.infer<typeof mulmoTextSlideMediaSchema>;
 export type MulmoMarkdownMedia = z.infer<typeof mulmoMarkdownMediaSchema>;
 export type MulmoImageMedia = z.infer<typeof mulmoImageMediaSchema>;
@@ -90,7 +91,7 @@ export type Text2HtmlAgentInfo = {
 export type BeatMediaType = "movie" | "image";
 export type StoryToScriptGenerateMode = (typeof storyToScriptGenerateMode)[keyof typeof storyToScriptGenerateMode];
 export type SessionType = "audio" | "image" | "video" | "multiLingual" | "caption" | "pdf";
-export type BeatSessionType = "audio" | "image" | "multiLingual" | "caption" | "movie" | "html";
+export type BeatSessionType = "audio" | "image" | "multiLingual" | "caption" | "movie" | "html" | "imageReference";
 export type SessionProgressEvent = {
     kind: "session";
     sessionType: SessionType;

package/lib/utils/const.d.ts

@@ -9,4 +9,3 @@ export declare const storyToScriptGenerateMode: {
     stepWise: string;
     oneStep: string;
 };
-export declare const defaultOpenAIImageModel = "dall-e-3";

package/lib/utils/const.js

@@ -9,4 +9,3 @@ export const storyToScriptGenerateMode = {
     stepWise: "step_wise",
     oneStep: "one_step",
 };
-export const defaultOpenAIImageModel = "dall-e-3";

package/lib/utils/context.d.ts

@@ -49,7 +49,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                 height: number;
             };
             speechParams: {
-                provider: "openai" | "nijivoice" | "google" | "elevenlabs";
+                provider: string;
                 speakers: Record<string, {
                     voiceId: string;
                     displayName?: Record<string, string> | undefined;
@@ -57,7 +57,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                         speed?: number | undefined;
                         instruction?: string | undefined;
                     } | undefined;
-                    provider?: "openai" | "nijivoice" | "google" | "elevenlabs" | undefined;
+                    provider?: string | undefined;
                 }>;
             };
             beats: {
@@ -193,9 +193,10 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                     type: "midi";
                     source: string;
                 } | undefined;
+                imagePrompt?: string | undefined;
                 description?: string | undefined;
                 imageParams?: {
-                    provider: "openai" | "google";
+                    provider: string;
                     style?: string | undefined;
                     model?: string | undefined;
                     moderation?: string | undefined;
@@ -214,6 +215,9 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                             path: string;
                             kind: "path";
                         };
+                    } | {
+                        type: "imagePrompt";
+                        prompt: string;
                     }> | undefined;
                 } | undefined;
                 audioParams?: {
@@ -221,6 +225,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                 } | undefined;
                 movieParams?: {
                     speed?: number | undefined;
+                    model?: string | undefined;
                     fillOption?: {
                         style: "aspectFit" | "aspectFill";
                     } | undefined;
@@ -236,7 +241,6 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                     lang?: string | undefined;
                 } | undefined;
                 imageNames?: string[] | undefined;
-                imagePrompt?: string | undefined;
                 moviePrompt?: string | undefined;
                 htmlPrompt?: {
                     prompt: string;
@@ -249,7 +253,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
             title?: string | undefined;
             description?: string | undefined;
             imageParams?: {
-                provider: "openai" | "google";
+                provider: string;
                 style?: string | undefined;
                 model?: string | undefined;
                 moderation?: string | undefined;
@@ -268,10 +272,13 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                         path: string;
                         kind: "path";
                     };
+                } | {
+                    type: "imagePrompt";
+                    prompt: string;
                 }> | undefined;
             } | undefined;
             movieParams?: {
-                provider?: "google" | "replicate" | undefined;
+                provider?: string | undefined;
                 model?: string | undefined;
                 fillOption?: {
                     style: "aspectFit" | "aspectFill";
@@ -282,7 +289,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                 } | undefined;
             } | undefined;
             htmlImageParams?: {
-                provider: "openai" | "anthropic";
+                provider: string;
                 model?: string | undefined;
             } | undefined;
             textSlideParams?: {
@@ -293,7 +300,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                 lang?: string | undefined;
             } | undefined;
             references?: {
-                type: "image" | "audio" | "article" | "paper" | "video";
+                type: string;
                 url: string;
                 title?: string | undefined;
                 description?: string | undefined;
@@ -322,6 +329,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
             multiLingual: {};
             caption: {};
             html: {};
+            imageReference: {};
         };
     };
     presentationStyle: {
@@ -356,7 +364,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
             height: number;
         };
         speechParams: {
-            provider: "openai" | "nijivoice" | "google" | "elevenlabs";
+            provider: string;
             speakers: Record<string, {
                 voiceId: string;
                 displayName?: Record<string, string> | undefined;
@@ -364,11 +372,11 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                     speed?: number | undefined;
                     instruction?: string | undefined;
                 } | undefined;
-                provider?: "openai" | "nijivoice" | "google" | "elevenlabs" | undefined;
+                provider?: string | undefined;
             }>;
         };
         imageParams?: {
-            provider: "openai" | "google";
+            provider: string;
             style?: string | undefined;
             model?: string | undefined;
             moderation?: string | undefined;
@@ -387,10 +395,13 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
                     path: string;
                     kind: "path";
                 };
+            } | {
+                type: "imagePrompt";
+                prompt: string;
             }> | undefined;
         } | undefined;
         movieParams?: {
-            provider?: "google" | "replicate" | undefined;
+            provider?: string | undefined;
             model?: string | undefined;
             fillOption?: {
                 style: "aspectFit" | "aspectFill";
@@ -401,7 +412,7 @@ export declare const initializeContextFromFiles: (files: FileObject, raiseError:
             } | undefined;
         } | undefined;
         htmlImageParams?: {
-            provider: "openai" | "anthropic";
+            provider: string;
             model?: string | undefined;
         } | undefined;
         textSlideParams?: {

package/lib/utils/context.js

@@ -57,6 +57,7 @@ const initSessionState = () => {
             multiLingual: {},
             caption: {},
             html: {},
+            imageReference: {},
         },
     };
 };

package/lib/utils/ffmpeg_utils.d.ts

@@ -12,4 +12,4 @@ export declare const FfmpegContextPushFormattedAudio: (context: FfmpegContext, s
 export declare const FfmpegContextInputFormattedAudio: (context: FfmpegContext, input: string, duration?: number | undefined, inputOptions?: string[]) => string;
 export declare const FfmpegContextGenerateOutput: (context: FfmpegContext, output: string, options?: string[]) => Promise<number>;
 export declare const ffmpegGetMediaDuration: (filePath: string) => Promise<number>;
-export declare const extractImageFromMovie: (movieFile: string, imagePath: string) => Promise<void>;
+export declare const extractImageFromMovie: (movieFile: string, imagePath: string) => Promise<object>;

package/lib/utils/ffmpeg_utils.js

@@ -77,7 +77,7 @@ export const extractImageFromMovie = (movieFile, imagePath) => {
         ffmpeg(movieFile)
             .outputOptions(["-frames:v 1"])
             .output(imagePath)
-            .on("end", () => resolve())
+            .on("end", () => resolve({}))
             .on("error", (err) => reject(err))
             .run();
     });

package/lib/utils/file.js

@@ -29,9 +29,9 @@ export function readMulmoScriptFile(arg2, errorMessage) {
             fileName: parsedPath.name,
         };
     }
-    catch (__error) {
+    catch (error) {
         if (errorMessage) {
-            GraphAILogger.info("read file format is broken.");
+            GraphAILogger.info("read file format is broken.", error);
         }
         return null;
     }
@@ -159,8 +159,8 @@ export const readTemplatePrompt = (templateName) => {
     const template = JSON.parse(templateData);
     const script = (() => {
         if (template.scriptName) {
-            const script = readScriptTemplateFile(template.scriptName);
-            return { ...script, ...(template.presentationStyle ?? {}) };
+            const scriptData = readScriptTemplateFile(template.scriptName);
+            return { ...scriptData, ...(template.presentationStyle ?? {}) };
         }
         return undefined;
     })();

package/lib/utils/filters.js

@@ -7,17 +7,16 @@ import { writingMessage } from "./file.js";
 import { text2hash } from "./utils.js";
 import { MulmoStudioContextMethods } from "../methods/mulmo_studio_context.js";
 export const fileCacheAgentFilter = async (context, next) => {
-    const { namedInputs } = context;
-    const { file, force, mulmoContext, index, sessionType } = namedInputs;
+    const { force, file, index, mulmoContext, sessionType } = context.namedInputs.cache;
     const shouldUseCache = async () => {
-        if (force) {
+        if (force && force.some((element) => element)) {
             return false;
         }
         try {
             await fsPromise.access(file);
             return true;
         }
-        catch (__e) {
+        catch {
             return false;
         }
     };

package/lib/utils/markdown.js

@@ -18,7 +18,7 @@ export const renderHTMLToImage = async (html, outputPath, width, height, isMerma
         }, { timeout: 20000 });
     }
     // Step 3: Capture screenshot of the page (which contains the Markdown-rendered HTML)
-    await page.screenshot({ path: outputPath, omitBackground: omitBackground });
+    await page.screenshot({ path: outputPath, omitBackground });
     await browser.close();
 };
 export const renderMarkdownToImage = async (markdown, style, outputPath, width, height) => {

package/lib/utils/preprocess.d.ts

@@ -44,7 +44,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
             height: number;
         };
         speechParams: {
-            provider: "openai" | "nijivoice" | "google" | "elevenlabs";
+            provider: string;
             speakers: Record<string, {
                 voiceId: string;
                 displayName?: Record<string, string> | undefined;
@@ -52,7 +52,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
                     speed?: number | undefined;
                     instruction?: string | undefined;
                 } | undefined;
-                provider?: "openai" | "nijivoice" | "google" | "elevenlabs" | undefined;
+                provider?: string | undefined;
             }>;
         };
         beats: {
@@ -188,9 +188,10 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
                 type: "midi";
                 source: string;
             } | undefined;
+            imagePrompt?: string | undefined;
             description?: string | undefined;
             imageParams?: {
-                provider: "openai" | "google";
+                provider: string;
                 style?: string | undefined;
                 model?: string | undefined;
                 moderation?: string | undefined;
@@ -209,6 +210,9 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
                         path: string;
                         kind: "path";
                     };
+                } | {
+                    type: "imagePrompt";
+                    prompt: string;
                 }> | undefined;
             } | undefined;
             audioParams?: {
@@ -216,6 +220,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
             } | undefined;
             movieParams?: {
                 speed?: number | undefined;
+                model?: string | undefined;
                 fillOption?: {
                     style: "aspectFit" | "aspectFill";
                 } | undefined;
@@ -231,7 +236,6 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
                 lang?: string | undefined;
             } | undefined;
             imageNames?: string[] | undefined;
-            imagePrompt?: string | undefined;
             moviePrompt?: string | undefined;
             htmlPrompt?: {
                 prompt: string;
@@ -244,7 +248,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
         title?: string | undefined;
         description?: string | undefined;
         imageParams?: {
-            provider: "openai" | "google";
+            provider: string;
             style?: string | undefined;
             model?: string | undefined;
             moderation?: string | undefined;
@@ -263,10 +267,13 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
                     path: string;
                     kind: "path";
                 };
+            } | {
+                type: "imagePrompt";
+                prompt: string;
             }> | undefined;
         } | undefined;
         movieParams?: {
-            provider?: "google" | "replicate" | undefined;
+            provider?: string | undefined;
             model?: string | undefined;
             fillOption?: {
                 style: "aspectFit" | "aspectFill";
@@ -277,7 +284,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
             } | undefined;
         } | undefined;
         htmlImageParams?: {
-            provider: "openai" | "anthropic";
+            provider: string;
             model?: string | undefined;
         } | undefined;
         textSlideParams?: {
@@ -288,7 +295,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
             lang?: string | undefined;
         } | undefined;
         references?: {
-            type: "image" | "audio" | "article" | "paper" | "video";
+            type: string;
             url: string;
             title?: string | undefined;
             description?: string | undefined;

package/lib/utils/provider2agent.d.ts

@@ -0,0 +1,72 @@
+export declare const defaultProviders: {
+    tts: string;
+    text2image: string;
+    text2movie: string;
+    text2Html: string;
+    llm: string;
+};
+export declare const provider2TTSAgent: {
+    nijivoice: {
+        agentName: string;
+        hasLimitedConcurrency: boolean;
+    };
+    openai: {
+        agentName: string;
+        hasLimitedConcurrency: boolean;
+    };
+    google: {
+        agentName: string;
+        hasLimitedConcurrency: boolean;
+    };
+    elevenlabs: {
+        agentName: string;
+        hasLimitedConcurrency: boolean;
+    };
+};
+export declare const provider2ImageAgent: {
+    openai: {
+        agentName: string;
+        defaultModel: string;
+        models: string[];
+    };
+    google: {
+        agentName: string;
+        defaultModel: string;
+        models: string[];
+    };
+};
+export declare const provider2MovieAgent: {
+    replicate: {
+        agentName: string;
+        models: string[];
+    };
+    google: {
+        agentName: string;
+        models: string[];
+    };
+};
+export declare const provider2LLMAgent: {
+    readonly openai: {
+        readonly agentName: "openAIAgent";
+        readonly defaultModel: "gpt-4o";
+        readonly max_tokens: 8192;
+    };
+    readonly anthropic: {
+        readonly agentName: "anthropicAgent";
+        readonly defaultModel: "claude-3-7-sonnet-20250219";
+        readonly max_tokens: 8192;
+    };
+    readonly gemini: {
+        readonly agentName: "geminiAgent";
+        readonly defaultModel: "gemini-1.5-flash";
+        readonly max_tokens: 8192;
+    };
+    readonly groq: {
+        readonly agentName: "groqAgent";
+        readonly defaultModel: "llama3-8b-8192";
+        readonly max_tokens: 4096;
+    };
+};
+export declare const llm: (keyof typeof provider2LLMAgent)[];
+export type LLM = keyof typeof provider2LLMAgent;
+export declare const htmlLLMProvider: string[];

package/lib/utils/provider2agent.js

@@ -0,0 +1,81 @@
+export const defaultProviders = {
+    tts: "openai",
+    text2image: "openai",
+    text2movie: "google",
+    text2Html: "openai",
+    llm: "openai",
+};
+export const provider2TTSAgent = {
+    nijivoice: {
+        agentName: "ttsNijivoiceAgent",
+        hasLimitedConcurrency: true,
+    },
+    openai: {
+        agentName: "ttsOpenaiAgent",
+        hasLimitedConcurrency: false,
+    },
+    google: {
+        agentName: "ttsGoogleAgent",
+        hasLimitedConcurrency: false,
+    },
+    elevenlabs: {
+        agentName: "ttsElevenlabsAgent",
+        hasLimitedConcurrency: true,
+    },
+};
+export const provider2ImageAgent = {
+    openai: {
+        agentName: "imageOpenaiAgent",
+        defaultModel: "gpt-image-1",
+        models: ["dall-e-3", "gpt-image-1"],
+    },
+    google: {
+        agentName: "imageGoogleAgent",
+        defaultModel: "imagen-3.0-fast-generate-001",
+        models: ["imagen-3.0-fast-generate-001", "imagen-3.0-generate-002", "imagen-3.0-capability-001"],
+    },
+};
+export const provider2MovieAgent = {
+    replicate: {
+        agentName: "movieReplicateAgent",
+        models: [
+            "bytedance/seedance-1-lite",
+            "bytedance/seedance-1-pro",
+            "kwaivgi/kling-v1.6-pro",
+            "kwaivgi/kling-v2.1",
+            "google/veo-2",
+            "google/veo-3",
+            "google/veo-3-fast",
+            "minimax/video-01",
+        ],
+    },
+    google: {
+        agentName: "movieGoogleAgent",
+        models: ["veo-2.0-generate-001"],
+    },
+};
+// : Record<LLM, { agent: string; defaultModel: string; max_tokens: number }>
+export const provider2LLMAgent = {
+    openai: {
+        agentName: "openAIAgent",
+        defaultModel: "gpt-4o",
+        max_tokens: 8192,
+    },
+    anthropic: {
+        agentName: "anthropicAgent",
+        defaultModel: "claude-3-7-sonnet-20250219",
+        max_tokens: 8192,
+    },
+    gemini: {
+        agentName: "geminiAgent",
+        defaultModel: "gemini-1.5-flash",
+        max_tokens: 8192,
+    },
+    groq: {
+        agentName: "groqAgent",
+        defaultModel: "llama3-8b-8192",
+        max_tokens: 4096,
+    },
+};
+export const llm = Object.keys(provider2LLMAgent);
+export const htmlLLMProvider = ["openai", "anthropic"];

package/lib/utils/string.js

@@ -20,18 +20,18 @@ export const recursiveSplitJa = (text) => {
     const delimiters = ["。", "？", "！", "、"];
     return delimiters
         .reduce((textData, delimiter) => {
-        return textData.map((text) => splitIntoSentencesJa(text, delimiter, 7)).flat(1);
+        return textData.map((textInner) => splitIntoSentencesJa(textInner, delimiter, 7)).flat(1);
     }, [text])
         .flat(1);
 };
 export function replacePairsJa(str, replacements) {
-    replacements.forEach(({ from, to }) => {
+    return replacements.reduce((tmp, current) => {
+        const { from, to } = current;
         // Escape any special regex characters in the 'from' string.
         const escapedFrom = from.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
         const regex = new RegExp(escapedFrom, "g");
-        str = str.replace(regex, to);
-    });
-    return str;
+        return tmp.replace(regex, to);
+    }, str);
 }
 export const replacementsJa = [
     { from: "Anthropic", to: "アンスロピック" },