mulmocast-preprocessor 0.5.2 → 0.5.3

package/lib/context.d.ts

@@ -0,0 +1 @@
+export { buildScriptContent, buildBeatContent, buildScriptMetaContent, scriptToViewerData, DEFAULT_INTERACTIVE_SYSTEM_PROMPT, } from "./core/ai/context-builder.js";

package/lib/context.js

@@ -0,0 +1,2 @@
+// Lightweight entry point for context-building utilities (no GraphAI dependency)
+export { buildScriptContent, buildBeatContent, buildScriptMetaContent, scriptToViewerData, DEFAULT_INTERACTIVE_SYSTEM_PROMPT, } from "./core/ai/context-builder.js";

package/lib/core/ai/command/query/prompts.d.ts

@@ -12,10 +12,8 @@ export declare const getSystemPrompt: (options: QueryOptions) => string;
  * Build user prompt from script and question
  */
 export declare const buildUserPrompt: (script: ExtendedMulmoScript, question: string) => string;
-/**
- * Default system prompt for interactive query
- */
-export declare const DEFAULT_INTERACTIVE_SYSTEM_PROMPT = "You are answering questions based on the content provided.\n- Answer based ONLY on the information in the provided content\n- If the answer cannot be found in the content, say so clearly\n- Be concise and direct in your answers\n- Do not make up information that is not in the content\n- You may reference previous conversation when answering follow-up questions\n- If references are available and the user asks for more details, mention which reference could provide more information\n- When you suggest fetching a reference for more details, include [SUGGEST_FETCH: <url>] in your response";
+import { DEFAULT_INTERACTIVE_SYSTEM_PROMPT } from "../../context-builder.js";
+export { DEFAULT_INTERACTIVE_SYSTEM_PROMPT };
 /**
  * Default system prompt for interactive query with fetched content
  */

package/lib/core/ai/command/query/prompts.js

@@ -1,4 +1,4 @@
-import { getLanguageName, buildScriptContent } from "../../llm.js";
+import { getLanguageName, buildScriptContent, scriptToViewerData } from "../../llm.js";
 /**
  * Default system prompt for query
  */
@@ -28,7 +28,7 @@ export const getSystemPrompt = (options) => {
 export const buildUserPrompt = (script, question) => {
     const parts = [];
     // Add common script content (title, language, sections with beats)
-    parts.push(buildScriptContent(script));
+    parts.push(buildScriptContent(scriptToViewerData(script)));
     parts.push("---");
     parts.push("");
     parts.push(`Question: ${question}`);
@@ -36,17 +36,8 @@ export const buildUserPrompt = (script, question) => {
     parts.push("Answer:");
     return parts.join("\n");
 };
-/**
- * Default system prompt for interactive query
- */
-export const DEFAULT_INTERACTIVE_SYSTEM_PROMPT = `You are answering questions based on the content provided.
-- Answer based ONLY on the information in the provided content
-- If the answer cannot be found in the content, say so clearly
-- Be concise and direct in your answers
-- Do not make up information that is not in the content
-- You may reference previous conversation when answering follow-up questions
-- If references are available and the user asks for more details, mention which reference could provide more information
-- When you suggest fetching a reference for more details, include [SUGGEST_FETCH: <url>] in your response`;
+import { DEFAULT_INTERACTIVE_SYSTEM_PROMPT } from "../../context-builder.js";
+export { DEFAULT_INTERACTIVE_SYSTEM_PROMPT };
 /**
  * Default system prompt for interactive query with fetched content
  */
@@ -77,7 +68,7 @@ export const getInteractiveSystemPrompt = (options) => {
 export const buildInteractiveUserPrompt = (script, question, history) => {
     const parts = [];
     // Add common script content (title, language, sections with beats)
-    parts.push(buildScriptContent(script));
+    parts.push(buildScriptContent(scriptToViewerData(script)));
     parts.push("---");
     parts.push("");
     // Add conversation history if exists

package/lib/core/ai/command/summarize/prompts.js

@@ -1,4 +1,4 @@
-import { getLanguageName, buildScriptContent } from "../../llm.js";
+import { getLanguageName, buildScriptContent, scriptToViewerData } from "../../llm.js";
 /**
  * Default system prompt for text summary
  */
@@ -24,7 +24,7 @@ export const DEFAULT_SYSTEM_PROMPT_MARKDOWN = `You are creating a summary based
 export const buildUserPrompt = (script, options) => {
     const parts = [];
     // Add common script content (title, language, sections with beats)
-    parts.push(buildScriptContent(script));
+    parts.push(buildScriptContent(scriptToViewerData(script)));
     // Add target length if specified
     if (options.targetLengthChars) {
         parts.push(`Target summary length: approximately ${options.targetLengthChars} characters`);

package/lib/core/ai/context-builder.d.ts

@@ -0,0 +1,21 @@
+import type { ExtendedMulmoScript, ExtendedMulmoViewerData, ExtendedMulmoViewerBeat } from "@mulmocast/extended-types";
+/**
+ * Default system prompt for interactive query
+ */
+export declare const DEFAULT_INTERACTIVE_SYSTEM_PROMPT = "You are answering questions based on the content provided.\n- Answer based ONLY on the information in the provided content\n- If the answer cannot be found in the content, say so clearly\n- Be concise and direct in your answers\n- Do not make up information that is not in the content\n- You may reference previous conversation when answering follow-up questions\n- If references are available and the user asks for more details, mention which reference could provide more information\n- When you suggest fetching a reference for more details, include [SUGGEST_FETCH: <url>] in your response";
+/**
+ * Build beat content including metadata
+ */
+export declare const buildBeatContent: (beat: ExtendedMulmoViewerBeat, index: number) => string;
+/**
+ * Build script-level metadata section
+ */
+export declare const buildScriptMetaContent: (data: ExtendedMulmoViewerData) => string;
+/**
+ * Convert ExtendedMulmoScript to ExtendedMulmoViewerData (extract only needed fields)
+ */
+export declare const scriptToViewerData: (script: ExtendedMulmoScript) => ExtendedMulmoViewerData;
+/**
+ * Build script content for user prompt (common part)
+ */
+export declare const buildScriptContent: (data: ExtendedMulmoViewerData) => string;

package/lib/core/ai/context-builder.js

@@ -0,0 +1,146 @@
+/**
+ * Default system prompt for interactive query
+ */
+export const DEFAULT_INTERACTIVE_SYSTEM_PROMPT = `You are answering questions based on the content provided.
+- Answer based ONLY on the information in the provided content
+- If the answer cannot be found in the content, say so clearly
+- Be concise and direct in your answers
+- Do not make up information that is not in the content
+- You may reference previous conversation when answering follow-up questions
+- If references are available and the user asks for more details, mention which reference could provide more information
+- When you suggest fetching a reference for more details, include [SUGGEST_FETCH: <url>] in your response`;
+/**
+ * Build beat content including metadata
+ */
+export const buildBeatContent = (beat, index) => {
+    const lines = [];
+    // Main text
+    const text = beat.text || "";
+    if (!text.trim())
+        return "";
+    lines.push(`[${index}] ${text}`);
+    // Add metadata if available
+    const meta = beat.meta;
+    if (meta) {
+        // Tags for categorization
+        if (meta.tags && meta.tags.length > 0) {
+            lines.push(`  Tags: ${meta.tags.join(", ")}`);
+        }
+        // Context provides additional information not in the text
+        if (meta.context) {
+            lines.push(`  Context: ${meta.context}`);
+        }
+        // Keywords highlight important terms
+        if (meta.keywords && meta.keywords.length > 0) {
+            lines.push(`  Keywords: ${meta.keywords.join(", ")}`);
+        }
+        // Expected questions this beat can answer
+        if (meta.expectedQuestions && meta.expectedQuestions.length > 0) {
+            lines.push(`  Can answer: ${meta.expectedQuestions.join("; ")}`);
+        }
+    }
+    return lines.join("\n");
+};
+/**
+ * Build script-level metadata section
+ */
+export const buildScriptMetaContent = (data) => {
+    const meta = data.scriptMeta;
+    if (!meta)
+        return "";
+    const lines = [];
+    // Background info
+    if (meta.background) {
+        lines.push(`Background: ${meta.background}`);
+    }
+    // Audience and prerequisites
+    if (meta.audience) {
+        lines.push(`Target audience: ${meta.audience}`);
+    }
+    if (meta.prerequisites && meta.prerequisites.length > 0) {
+        lines.push(`Prerequisites: ${meta.prerequisites.join(", ")}`);
+    }
+    // Goals
+    if (meta.goals && meta.goals.length > 0) {
+        lines.push(`Goals: ${meta.goals.join("; ")}`);
+    }
+    // Keywords
+    if (meta.keywords && meta.keywords.length > 0) {
+        lines.push(`Keywords: ${meta.keywords.join(", ")}`);
+    }
+    // References
+    if (meta.references && meta.references.length > 0) {
+        lines.push("References:");
+        meta.references.forEach((ref) => {
+            const title = ref.title || ref.url;
+            const desc = ref.description ? ` - ${ref.description}` : "";
+            lines.push(`  - [${ref.type || "web"}] ${title}: ${ref.url}${desc}`);
+        });
+    }
+    // FAQ
+    if (meta.faq && meta.faq.length > 0) {
+        lines.push("FAQ:");
+        meta.faq.forEach((faq) => {
+            lines.push(`  Q: ${faq.question}`);
+            lines.push(`  A: ${faq.answer}`);
+        });
+    }
+    // Author info
+    if (meta.author) {
+        lines.push(`Author: ${meta.author}`);
+    }
+    return lines.length > 0 ? lines.join("\n") : "";
+};
+/**
+ * Convert ExtendedMulmoScript to ExtendedMulmoViewerData (extract only needed fields)
+ */
+export const scriptToViewerData = (script) => {
+    return {
+        beats: script.beats.map((b) => ({
+            text: b.text,
+            meta: b.meta,
+            variants: b.variants,
+            id: b.id,
+        })),
+        title: script.title,
+        lang: script.lang,
+        scriptMeta: script.scriptMeta,
+        outputProfiles: script.outputProfiles,
+    };
+};
+/**
+ * Build script content for user prompt (common part)
+ */
+export const buildScriptContent = (data) => {
+    const parts = [];
+    // Add script title and language
+    parts.push(`# Script: ${data.title}`);
+    parts.push(`Language: ${data.lang}`);
+    parts.push("");
+    // Add script-level metadata
+    const scriptMetaContent = buildScriptMetaContent(data);
+    if (scriptMetaContent) {
+        parts.push("## About this content");
+        parts.push(scriptMetaContent);
+        parts.push("");
+    }
+    // Collect all content from beats grouped by section
+    const sections = new Map();
+    data.beats.forEach((beat, index) => {
+        const content = buildBeatContent(beat, index);
+        if (!content)
+            return;
+        const section = beat.meta?.section || "main";
+        if (!sections.has(section)) {
+            sections.set(section, []);
+        }
+        sections.get(section).push(content);
+    });
+    // Output by section
+    sections.forEach((contents, section) => {
+        parts.push(`## Section: ${section}`);
+        contents.forEach((c) => parts.push(c));
+        parts.push("");
+    });
+    return parts.join("\n");
+};

package/lib/core/ai/llm.d.ts

@@ -1,5 +1,6 @@
 import type { ExtendedMulmoScript } from "@mulmocast/extended-types";
 import type { LLMProvider } from "../../types/summarize.js";
+export { buildBeatContent, buildScriptMetaContent, buildScriptContent, scriptToViewerData } from "./context-builder.js";
 /**
  * Base options for LLM operations
  */
@@ -39,10 +40,6 @@ export declare const filterScript: (script: ExtendedMulmoScript, options: BaseLL
  * Get language name from code
  */
 export declare const getLanguageName: (langCode: string) => string;
-/**
- * Build script content for user prompt (common part)
- */
-export declare const buildScriptContent: (script: ExtendedMulmoScript) => string;
 /**
  * Command execution result
  */

package/lib/core/ai/llm.js

@@ -6,6 +6,8 @@ import { anthropicAgent } from "@graphai/anthropic_agent";
 import { groqAgent } from "@graphai/groq_agent";
 import { geminiAgent } from "@graphai/gemini_agent";
 import { filterBySection, filterByTags } from "../preprocessing/filter.js";
+// Re-export pure context-building functions (no GraphAI dependency)
+export { buildBeatContent, buildScriptMetaContent, buildScriptContent, scriptToViewerData } from "./context-builder.js";
 dotenv.config({ quiet: true });
 const agents = vanillaAgents.default ?? vanillaAgents;
 const provider2Agent = {
@@ -103,124 +105,6 @@ export const getLanguageName = (langCode) => {
     };
     return langMap[langCode] || langCode;
 };
-/**
- * Build beat content including metadata
- */
-const buildBeatContent = (beat, index) => {
-    const lines = [];
-    // Main text
-    const text = beat.text || "";
-    if (!text.trim())
-        return "";
-    lines.push(`[${index}] ${text}`);
-    // Add metadata if available
-    const meta = beat.meta;
-    if (meta) {
-        // Tags for categorization
-        if (meta.tags && meta.tags.length > 0) {
-            lines.push(`  Tags: ${meta.tags.join(", ")}`);
-        }
-        // Context provides additional information not in the text
-        if (meta.context) {
-            lines.push(`  Context: ${meta.context}`);
-        }
-        // Keywords highlight important terms
-        if (meta.keywords && meta.keywords.length > 0) {
-            lines.push(`  Keywords: ${meta.keywords.join(", ")}`);
-        }
-        // Expected questions this beat can answer
-        if (meta.expectedQuestions && meta.expectedQuestions.length > 0) {
-            lines.push(`  Can answer: ${meta.expectedQuestions.join("; ")}`);
-        }
-    }
-    return lines.join("\n");
-};
-/**
- * Build script-level metadata section
- */
-const buildScriptMetaContent = (script) => {
-    const meta = script.scriptMeta;
-    if (!meta)
-        return "";
-    const lines = [];
-    // Background info
-    if (meta.background) {
-        lines.push(`Background: ${meta.background}`);
-    }
-    // Audience and prerequisites
-    if (meta.audience) {
-        lines.push(`Target audience: ${meta.audience}`);
-    }
-    if (meta.prerequisites && meta.prerequisites.length > 0) {
-        lines.push(`Prerequisites: ${meta.prerequisites.join(", ")}`);
-    }
-    // Goals
-    if (meta.goals && meta.goals.length > 0) {
-        lines.push(`Goals: ${meta.goals.join("; ")}`);
-    }
-    // Keywords
-    if (meta.keywords && meta.keywords.length > 0) {
-        lines.push(`Keywords: ${meta.keywords.join(", ")}`);
-    }
-    // References
-    if (meta.references && meta.references.length > 0) {
-        lines.push("References:");
-        meta.references.forEach((ref) => {
-            const title = ref.title || ref.url;
-            const desc = ref.description ? ` - ${ref.description}` : "";
-            lines.push(`  - [${ref.type || "web"}] ${title}: ${ref.url}${desc}`);
-        });
-    }
-    // FAQ
-    if (meta.faq && meta.faq.length > 0) {
-        lines.push("FAQ:");
-        meta.faq.forEach((faq) => {
-            lines.push(`  Q: ${faq.question}`);
-            lines.push(`  A: ${faq.answer}`);
-        });
-    }
-    // Author info
-    if (meta.author) {
-        lines.push(`Author: ${meta.author}`);
-    }
-    return lines.length > 0 ? lines.join("\n") : "";
-};
-/**
- * Build script content for user prompt (common part)
- */
-export const buildScriptContent = (script) => {
-    const parts = [];
-    // Add script title and language
-    parts.push(`# Script: ${script.title}`);
-    parts.push(`Language: ${script.lang}`);
-    parts.push("");
-    // Add script-level metadata
-    const scriptMetaContent = buildScriptMetaContent(script);
-    if (scriptMetaContent) {
-        parts.push("## About this content");
-        parts.push(scriptMetaContent);
-        parts.push("");
-    }
-    // Collect all content from beats grouped by section
-    const sections = new Map();
-    script.beats.forEach((beat, index) => {
-        const content = buildBeatContent(beat, index);
-        if (!content)
-            return;
-        const section = beat.meta?.section || "main";
-        if (!sections.has(section)) {
-            sections.set(section, []);
-        }
-        sections.get(section).push(content);
-    });
-    // Output by section
-    sections.forEach((contents, section) => {
-        parts.push(`## Section: ${section}`);
-        contents.forEach((c) => parts.push(c));
-        parts.push("");
-    });
-    return parts.join("\n");
-};
 /**
  * Execute a command (summarize, query, etc.) with common logic
  */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mulmocast-preprocessor",
-  "version": "0.5.2",
+  "version": "0.5.3",
   "description": "Preprocessor for MulmoScript",
   "type": "module",
   "main": "lib/index.js",
@@ -12,6 +12,10 @@
     ".": {
       "types": "./lib/index.d.ts",
       "default": "./lib/index.js"
+    },
+    "./context": {
+      "types": "./lib/context.d.ts",
+      "default": "./lib/context.js"
     }
   },
   "files": [