mulmocast 0.0.8 → 0.0.9

package/lib/types/schema.js

@@ -109,7 +109,9 @@ export const mulmoImageAssetSchema = z.union([
     mulmoPdfMediaSchema,
     mulmoImageMediaSchema,
     mulmoSvgMediaSchema,
-    mulmoMovieMediaSchema,
+    mulmoMovieMediaSchema.extend({
+        mixAudio: z.number().default(1.0),
+    }),
     mulmoTextSlideMediaSchema,
     mulmoChartMediaSchema,
     mulmoMermaidMediaSchema,
@@ -128,12 +130,14 @@ const mulmoMidiMediaSchema = z
 })
     .strict();
 export const mulmoAudioAssetSchema = z.union([mulmoAudioMediaSchema, mulmoMidiMediaSchema]);
+const imageIdSchema = z.string();
 export const mulmoImageParamsSchema = z
     .object({
     model: z.string().optional(), // default: provider specific
     size: z.string().optional(), // default: provider specific
     style: z.string().optional(), // optional image style
     moderation: z.string().optional(), // optional image style
+    images: z.record(imageIdSchema, mulmoImageMediaSchema).optional(),
 })
     .strict();
 export const textSlideParamsSchema = z
@@ -173,6 +177,7 @@ export const mulmoBeatSchema = z
     audioParams: beatAudioParamsSchema.optional(), // beat specific parameters
     speechOptions: speechOptionsSchema.optional(),
     textSlideParams: textSlideParamsSchema.optional(),
+    imageNames: z.array(imageIdSchema).optional(), // list of image names to use for image generation. The default is all images in the imageParams.images.
     imagePrompt: z.string().optional(), // specified or inserted by preprocessor
 })
     .strict();
@@ -231,7 +236,7 @@ export const mulmoReferenceSchema = z.object({
     url: URLStringSchema,
     title: z.string().optional(),
     description: z.string().optional(),
-    type: z.union([z.literal("article"), z.literal("image"), z.literal("video"), z.literal("audio")]).default("article"),
+    type: z.union([z.literal("article"), z.literal("paper"), z.literal("image"), z.literal("video"), z.literal("audio")]).default("article"),
 });
 export const mulmoScriptSchema = mulmoPresentationStyleSchema
     .extend({
@@ -305,6 +310,7 @@ export const mulmoScriptTemplateSchema = z
     description: z.string(),
     systemPrompt: z.string(),
     scriptName: z.string().optional(),
+    presentationStyle: mulmoPresentationStyleSchema.optional(),
 })
     .strict();
 export const mulmoStoryboardSceneSchema = z

package/lib/utils/file.js

@@ -18,13 +18,22 @@ export function readMulmoScriptFile(arg2, errorMessage) {
         return null;
     }
     const scriptData = fs.readFileSync(scriptPath, "utf-8");
-    const script = ([".yaml", ".yml"].includes(path.extname(scriptPath).toLowerCase()) ? yamlParse(scriptData) : JSON.parse(scriptData));
-    const parsedPath = path.parse(scriptPath);
-    return {
-        mulmoData: script,
-        mulmoDataPath: scriptPath,
-        fileName: parsedPath.name,
-    };
+    try {
+        const script = ([".yaml", ".yml"].includes(path.extname(scriptPath).toLowerCase()) ? yamlParse(scriptData) : JSON.parse(scriptData));
+        const parsedPath = path.parse(scriptPath);
+        return {
+            mulmoData: script,
+            mulmoDataPath: scriptPath,
+            fileName: parsedPath.name,
+        };
+    }
+    catch (__error) {
+        if (errorMessage) {
+            GraphAILogger.info("read file format is broken.");
+            process.exit(1);
+        }
+        return null;
+    }
 }
 export const fetchMulmoScriptFile = async (url) => {
     try {
@@ -120,10 +129,12 @@ export const readScriptTemplateFile = (scriptName) => {
 export const readTemplatePrompt = (templateName) => {
     const templatePath = getTemplateFilePath(templateName);
     const templateData = fs.readFileSync(templatePath, "utf-8");
-    const template = mulmoScriptTemplateSchema.parse(JSON.parse(templateData));
+    // NOTE: We don't want to schema parse the template here to eliminate default values.
+    const template = JSON.parse(templateData);
     const script = (() => {
         if (template.scriptName) {
-            return readScriptTemplateFile(template.scriptName);
+            const script = readScriptTemplateFile(template.scriptName);
+            return { ...script, ...(template.presentationStyle ?? {}) };
         }
         return undefined;
     })();

package/lib/utils/pdf.d.ts

@@ -2,6 +2,7 @@ import { PDFFont } from "pdf-lib";
 export declare const fontSize = 12;
 export declare const textMargin = 6;
 export declare const drawSize: (fitWidth: boolean, expectWidth: number, expectHeight: number, origWidth: number, origHeight: number) => {
+    containerWidth: number;
     drawWidth: number;
     drawHeight: number;
 };

package/lib/utils/pdf.js

@@ -6,6 +6,7 @@ export const drawSize = (fitWidth, expectWidth, expectHeight, origWidth, origHei
         const scale = drawWidth / origWidth;
         const drawHeight = origHeight * scale;
         return {
+            containerWidth: expectWidth,
             drawWidth,
             drawHeight,
         };
@@ -14,6 +15,7 @@ export const drawSize = (fitWidth, expectWidth, expectHeight, origWidth, origHei
     const scale = drawHeight / origHeight;
     const drawWidth = origWidth * scale;
     return {
+        containerWidth: expectWidth,
         drawWidth,
         drawHeight,
     };
@@ -37,10 +39,10 @@ export const wrapText = (text, font, fontSize, maxWidth) => {
             const currentIsFull = isFullwidth(char);
             const nextIsFull = nextChar && isFullwidth(nextChar);
             const isBreakable = currentIsFull || (!currentIsFull && (char === " " || nextChar === " " || nextIsFull));
-            if (width > maxWidth && buffer) {
+            if (width > maxWidth && line) {
                 lines.push(line);
-                line = "";
-                buffer = char;
+                line = buffer;
+                buffer = "";
             }
             if (isBreakable || i === rawLine.length - 1) {
                 line += buffer;

package/lib/utils/preprocess.d.ts

@@ -101,21 +101,6 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
                     path: string;
                     kind: "path";
                 };
-            } | {
-                type: "movie";
-                source: {
-                    url: string;
-                    kind: "url";
-                } | {
-                    kind: "base64";
-                    data: string;
-                } | {
-                    text: string;
-                    kind: "text";
-                } | {
-                    path: string;
-                    kind: "path";
-                };
             } | {
                 type: "textSlide";
                 slide: {
@@ -147,6 +132,22 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
             } | {
                 type: "html_tailwind";
                 html: string | string[];
+            } | {
+                type: "movie";
+                source: {
+                    url: string;
+                    kind: "url";
+                } | {
+                    kind: "base64";
+                    data: string;
+                } | {
+                    text: string;
+                    kind: "text";
+                } | {
+                    path: string;
+                    kind: "path";
+                };
+                mixAudio: number;
             } | undefined;
             audio?: {
                 type: "audio";
@@ -172,6 +173,22 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
                 size?: string | undefined;
                 style?: string | undefined;
                 moderation?: string | undefined;
+                images?: Record<string, {
+                    type: "image";
+                    source: {
+                        url: string;
+                        kind: "url";
+                    } | {
+                        kind: "base64";
+                        data: string;
+                    } | {
+                        text: string;
+                        kind: "text";
+                    } | {
+                        path: string;
+                        kind: "path";
+                    };
+                }> | undefined;
             } | undefined;
             audioParams?: {
                 padding?: number | undefined;
@@ -179,6 +196,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
             textSlideParams?: {
                 cssStyles: string | string[];
             } | undefined;
+            imageNames?: string[] | undefined;
             imagePrompt?: string | undefined;
         }[];
         lang?: string | undefined;
@@ -189,6 +207,22 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
             size?: string | undefined;
             style?: string | undefined;
             moderation?: string | undefined;
+            images?: Record<string, {
+                type: "image";
+                source: {
+                    url: string;
+                    kind: "url";
+                } | {
+                    kind: "base64";
+                    data: string;
+                } | {
+                    text: string;
+                    kind: "text";
+                } | {
+                    path: string;
+                    kind: "path";
+                };
+            }> | undefined;
         } | undefined;
         textSlideParams?: {
             cssStyles: string | string[];
@@ -196,7 +230,7 @@ export declare const createOrUpdateStudioData: (_mulmoScript: MulmoScript, curre
         omitCaptions?: boolean | undefined;
         description?: string | undefined;
         references?: {
-            type: "image" | "audio" | "article" | "video";
+            type: "image" | "audio" | "article" | "paper" | "video";
             url: string;
             title?: string | undefined;
             description?: string | undefined;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mulmocast",
-  "version": "0.0.8",
+  "version": "0.0.9",
   "description": "",
   "type": "module",
   "main": "lib/cli/bin.js",
@@ -52,22 +52,22 @@
   "homepage": "https://github.com/receptron/mulmocast-cli#readme",
   "dependencies": {
     "@google-cloud/text-to-speech": "^6.1.0",
-    "@graphai/anthropic_agent": "^2.0.0",
+    "@graphai/anthropic_agent": "^2.0.2",
     "@graphai/browserless_agent": "^2.0.0",
-    "@graphai/gemini_agent": "^1.0.1",
-    "@graphai/groq_agent": "^1.0.1",
+    "@graphai/gemini_agent": "^2.0.0",
+    "@graphai/groq_agent": "^2.0.0",
     "@graphai/input_agents": "^1.0.1",
-    "@graphai/openai_agent": "^2.0.0",
-    "@graphai/stream_agent_filter": "^2.0.1",
-    "@graphai/vanilla": "^2.0.2",
-    "@graphai/vanilla_node_agents": "^2.0.0",
+    "@graphai/openai_agent": "^2.0.3",
+    "@graphai/stream_agent_filter": "^2.0.2",
+    "@graphai/vanilla": "^2.0.4",
+    "@graphai/vanilla_node_agents": "^2.0.1",
     "@pdf-lib/fontkit": "^1.1.1",
     "canvas": "^3.1.0",
     "clipboardy": "^4.0.0",
     "dotenv": "^16.4.7",
     "fluent-ffmpeg": "^2.1.3",
     "google-auth-library": "^9.15.1",
-    "graphai": "^2.0.3",
+    "graphai": "^2.0.5",
     "inquirer": "^12.6.1",
     "marked": "^15.0.11",
     "ora": "^8.2.0",

package/scripts/templates/children_book.json

@@ -4,13 +4,6 @@
     "credit": "closing"
   },
   "title": "桃太郎",
-  "canvasSize": {
-    "width": 1536,
-    "height": 1024
-  },
-  "imageParams": {
-    "style": "A hand-drawn style illustration with a warm, nostalgic atmosphere. The background is rich with natural scenery—lush forests, cloudy skies, and traditional Japanese architecture. Characters have expressive eyes, soft facial features, and are portrayed with gentle lighting and subtle shading. The color palette is muted yet vivid, using earthy tones and watercolor-like textures. The overall scene feels magical and peaceful, with a sense of quiet wonder and emotional depth, reminiscent of classic 1980s and 1990s Japanese animation."
-  },
   "lang": "ja",
   "beats": [
     {

package/scripts/templates/image_prompts_template.json

@@ -0,0 +1,41 @@
+{
+  "$mulmocast": {
+    "version": "1.0",
+    "credit": "closing"
+  },
+  "title": "[TITLE: Brief, engaging title for the topic]",
+  "lang": "en",
+  "references": [
+    {
+      "url": "[SOURCE_URL: URL of the source material]",
+      "title": "[SOURCE_TITLE: Title of the referenced article, or paper]",
+      "type": "[SOURCE_TYPE: article, paper]"
+    }
+  ],
+  "beats": [
+    {
+      "text": "[OPENING_BEAT: Introduce the topic with a hook. Reference the source material and set up why this topic matters. Usually 2-3 sentences that grab attention and provide context.]",
+      "imagePrompt": "[IMAGE_PROMPT: A prompt for the image to be generated for this beat.]"
+    },
+    {
+      "text": "[MAIN_CONCEPT: Define or explain the core concept/idea. This should be the central focus of your narrative. Keep it clear and accessible.]",
+      "imagePrompt": "[IMAGE_PROMPT: A prompt for the image to be generated for this beat.]"
+    },
+    {
+      "text": "[SUPPORTING_DETAIL_1: Additional context, examples, or elaboration that helps illustrate the main concept. This could include how it works, why it's important, or real-world applications.]",
+      "imagePrompt": "[IMAGE_PROMPT: A prompt for the image to be generated for this beat.]"
+    },
+    {
+      "text": "[SUPPORTING_DETAIL_2: Continue with more examples, deeper explanation, or different aspects of the topic if needed.]",
+      "imagePrompt": "[IMAGE_PROMPT: A prompt for the image to be generated for this beat.]"
+    },
+    {
+      "text": "[ADDITIONAL_BEATS: Add more beats as necessary to fully explore the topic. Complex topics may require 6-10+ beats to cover adequately. Each beat should advance the narrative or provide valuable information.]",
+      "imagePrompt": "[IMAGE_PROMPT: A prompt for the image to be generated for this beat.]"
+    },
+    {
+      "text": "[CONCLUSION/IMPACT: Wrap up with the significance, implications, or key takeaway. Help the audience understand why this matters to them.]",
+      "imagePrompt": "[IMAGE_PROMPT: A prompt for the image to be generated for this beat.]"
+    }
+  ]
+}

package/scripts/templates/sensei_and_taro.json

@@ -6,17 +6,6 @@
   "title": "韓国の戒厳令とその日本への影響",
   "description": "韓国で最近発令された戒厳令とその可能性のある影響について、また日本の憲法に関する考慮事項との類似点を含めた洞察に満ちた議論。",
   "lang": "ja",
-  "imageParams": {
-    "style": "<style>Ghibli style. Student (Taro) is a young teenager with a dark short hair with glasses. Teacher is a middle-aged man with grey hair and moustache.</style>"
-  },
-  "speechParams": {
-    "provider": "nijivoice",
-    "speakers": {
-      "Announcer": { "displayName": { "ja": "アナウンサー" }, "voiceId": "afd7df65-0fdc-4d31-ae8b-a29f0f5eed62" },
-      "Student": { "displayName": { "ja": "太郎" }, "voiceId": "a7619e48-bf6a-4f9f-843f-40485651257f" },
-      "Teacher": { "displayName": { "ja": "先生" }, "voiceId": "bc06c63f-fef6-43b6-92f7-67f919bd5dae" }
-    }
-  },
   "beats": [
     {
       "speaker": "Announcer",

package/scripts/templates/text_only_template.json

@@ -0,0 +1,35 @@
+{
+  "$mulmocast": {
+    "version": "1.0",
+    "credit": "closing"
+  },
+  "title": "[TITLE: Brief, engaging title for the topic]",
+  "lang": "en",
+  "references": [
+    {
+      "url": "[SOURCE_URL: URL of the source material]",
+      "title": "[SOURCE_TITLE: Title of the referenced article, or paper]",
+      "type": "[SOURCE_TYPE: article, paper]"
+    }
+  ],
+  "beats": [
+    {
+      "text": "[OPENING_BEAT: Introduce the topic with a hook. Reference the source material and set up why this topic matters. Usually 2-3 sentences that grab attention and provide context.]"
+    },
+    {
+      "text": "[MAIN_CONCEPT: Define or explain the core concept/idea. This should be the central focus of your narrative. Keep it clear and accessible.]"
+    },
+    {
+      "text": "[SUPPORTING_DETAIL_1: Additional context, examples, or elaboration that helps illustrate the main concept. This could include how it works, why it's important, or real-world applications.]"
+    },
+    {
+      "text": "[SUPPORTING_DETAIL_2: Continue with more examples, deeper explanation, or different aspects of the topic if needed.]"
+    },
+    {
+      "text": "[ADDITIONAL_BEATS: Add more beats as necessary to fully explore the topic. Complex topics may require 6-10+ beats to cover adequately. Each beat should advance the narrative or provide valuable information.]"
+    },
+    {
+      "text": "[CONCLUSION/IMPACT: Wrap up with the significance, implications, or key takeaway. Help the audience understand why this matters to them.]"
+    }
+  ]
+}

package/assets/templates/ghibli_strips.json

@@ -1,6 +0,0 @@
-{
-  "title": "American Comic Strips",
-  "description": "Template for Dilbert-style comic strips.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate comic strips for each beat based on the text description of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
-  "scriptName": "ghibli_strips.json"
-}

package/scripts/templates/comic_strips.json

@@ -1,30 +0,0 @@
-{
-  "$mulmocast": {
-    "version": "1.0",
-    "credit": "closing"
-  },
-  "title": "Sample Title",
-  "lang": "en",
-  "canvasSize": {
-    "width": 1536,
-    "height": 1024
-  },
-  "imageParams": {
-    "style": "<style>A multi panel comic strips. 1990s American workplace humor. Clean, minimalist line art with muted colors. One character is a nerdy office worker with glasses</style>"
-  },
-  "references": [
-    {
-      "url": "https://www.somegreatwebsite.com/article/123",
-      "title": "Title of the article we are referencing",
-      "type": "article"
-    }
-  ],
-  "beats": [
-    {
-      "text": "Today we're exploring a fascinating concept that has shaped some of the most innovative companies and leaders of our time. This story is based on an Awesome News articled, titled 'The reality distortion field'. This term has become synonymous with visionary leadership and the ability to accomplish seemingly impossible goals."
-    },
-    {
-      "text": "The Reality Distortion Field is a concept that was first introduced by Steve Jobs, the co-founder of Apple. It refers to the ability of a leader to convince people that something is possible, even if it seems impossible."
-    }
-  ]
-}

package/scripts/templates/ghibli_strips.json

@@ -1,30 +0,0 @@
-{
-  "$mulmocast": {
-    "version": "1.0",
-    "credit": "closing"
-  },
-  "title": "Sample Title",
-  "lang": "en",
-  "canvasSize": {
-    "width": 1536,
-    "height": 1024
-  },
-  "imageParams": {
-    "style": "<style>A multi panel comic strips. Ghibli style. The presenter is a young woman with a dark short hair with glasses.</style>"
-  },
-  "references": [
-    {
-      "url": "https://www.somegreatwebsite.com/article/123",
-      "title": "The Awesome News: The reality distortion field",
-      "type": "article"
-    }
-  ],
-  "beats": [
-    {
-      "text": "Today we're exploring a fascinating concept that has shaped some of the most innovative companies and leaders of our time. This story is based on an Awesome News articled, titled 'The reality distortion field'. This term has become synonymous with visionary leadership and the ability to accomplish seemingly impossible goals."
-    },
-    {
-      "text": "The Reality Distortion Field is a concept that was first introduced by Steve Jobs, the co-founder of Apple. It refers to the ability of a leader to convince people that something is possible, even if it seems impossible."
-    }
-  ]
-}