mulmocast-preprocessor 0.1.1 → 0.2.0

package/lib/core/summarize/index.js

@@ -1,65 +1,12 @@
-import dotenv from "dotenv";
-import { GraphAI, GraphAILogger } from "graphai";
-import * as vanillaAgents from "@graphai/vanilla";
-import { openAIAgent } from "@graphai/openai_agent";
-import { anthropicAgent } from "@graphai/anthropic_agent";
-import { groqAgent } from "@graphai/groq_agent";
-import { geminiAgent } from "@graphai/gemini_agent";
 import { summarizeOptionsSchema } from "../../types/summarize.js";
-import { getProviderConfig, getProviderApiKey } from "./provider.js";
+import { executeLLM, filterScript } from "../llm/index.js";
 import { buildUserPrompt, getSystemPrompt } from "./prompts.js";
-import { filterBySection, filterByTags } from "../filter.js";
-dotenv.config({ quiet: true });
-const agents = vanillaAgents.default ?? vanillaAgents;
-/**
- * Create GraphAI graph for summarizing script
- */
-const createSummarizeGraph = (agentName) => ({
-    version: 0.5,
-    nodes: {
-        systemPrompt: {},
-        userPrompt: {},
-        model: {},
-        temperature: {},
-        maxTokens: {},
-        llmCall: {
-            agent: agentName,
-            inputs: {
-                system: ":systemPrompt",
-                prompt: ":userPrompt",
-                model: ":model",
-                temperature: ":temperature",
-                max_tokens: ":maxTokens",
-            },
-        },
-        result: {
-            isResult: true,
-            agent: "copyAgent",
-            inputs: {
-                summary: ":llmCall.text",
-            },
-        },
-    },
-});
-/**
- * Filter script based on options (section, tags)
- */
-const filterScript = (script, options) => {
-    const afterSection = options.section ? filterBySection(script, options.section) : script;
-    const afterTags = options.tags && options.tags.length > 0 ? filterByTags(afterSection, options.tags) : afterSection;
-    return afterTags;
-};
 /**
  * Main summarize function - generates a summary of the entire script
  */
 export const summarizeScript = async (script, options = {}) => {
     // Validate and apply defaults
     const validatedOptions = summarizeOptionsSchema.parse(options);
-    const providerConfig = getProviderConfig(validatedOptions.provider);
-    const apiKey = getProviderApiKey(validatedOptions.provider);
-    if (!apiKey) {
-        throw new Error(`API key not found for provider "${validatedOptions.provider}". ` + `Please set the ${providerConfig.keyName} environment variable.`);
-    }
     // Filter script if section/tags specified
     const filteredScript = filterScript(script, validatedOptions);
     const scriptTitle = script.title || "Untitled";
@@ -71,39 +18,11 @@ export const summarizeScript = async (script, options = {}) => {
             beatCount: 0,
         };
     }
-    // Build GraphAI config
-    const config = {
-        openAIAgent: { apiKey: process.env.OPENAI_API_KEY },
-        anthropicAgent: { apiKey: process.env.ANTHROPIC_API_KEY },
-        groqAgent: { apiKey: process.env.GROQ_API_KEY },
-        geminiAgent: { apiKey: process.env.GEMINI_API_KEY },
-    };
-    // Create GraphAI instance
-    const graph = new GraphAI(createSummarizeGraph(providerConfig.agentName), {
-        ...agents,
-        openAIAgent,
-        anthropicAgent,
-        groqAgent,
-        geminiAgent,
-    }, { config });
     // Build prompts
     const systemPrompt = getSystemPrompt(validatedOptions);
     const userPrompt = buildUserPrompt(filteredScript, validatedOptions);
-    if (validatedOptions.verbose) {
-        GraphAILogger.info(`Summarizing script "${script.title}" with ${validatedOptions.provider}...`);
-        GraphAILogger.info(`Beats: ${filteredScript.beats.length}, Format: ${validatedOptions.format}`);
-    }
-    // Inject values
-    graph.injectValue("systemPrompt", systemPrompt);
-    graph.injectValue("userPrompt", userPrompt);
-    graph.injectValue("model", validatedOptions.model ?? providerConfig.defaultModel);
-    graph.injectValue("temperature", validatedOptions.temperature ?? 0.7);
-    graph.injectValue("maxTokens", validatedOptions.maxTokens ?? providerConfig.maxTokens ?? 2048);
-    // Run graph
-    const graphResult = await graph.run();
-    // Extract summary from result node
-    const resultNode = graphResult.result;
-    const summary = resultNode?.summary || "";
+    // Execute LLM
+    const summary = await executeLLM(systemPrompt, userPrompt, validatedOptions, `Summarizing script "${script.title}" with ${validatedOptions.provider}... Beats: ${filteredScript.beats.length}, Format: ${validatedOptions.format}`);
     return {
         summary,
         format: validatedOptions.format,

package/lib/core/summarize/prompts.d.ts

@@ -13,6 +13,6 @@ export declare const DEFAULT_SYSTEM_PROMPT_MARKDOWN = "You are creating a summar
  */
 export declare const buildUserPrompt: (script: ExtendedScript, options: SummarizeOptions) => string;
 /**
- * Get system prompt based on format
+ * Get system prompt based on format and language
  */
 export declare const getSystemPrompt: (options: SummarizeOptions) => string;

package/lib/core/summarize/prompts.js

@@ -1,3 +1,4 @@
+import { getLanguageName } from "../llm/index.js";
 /**
  * Default system prompt for text summary
  */
@@ -53,11 +54,17 @@ export const buildUserPrompt = (script, options) => {
     return parts.join("\n");
 };
 /**
- * Get system prompt based on format
+ * Get system prompt based on format and language
  */
 export const getSystemPrompt = (options) => {
     if (options.systemPrompt) {
         return options.systemPrompt;
     }
-    return options.format === "markdown" ? DEFAULT_SYSTEM_PROMPT_MARKDOWN : DEFAULT_SYSTEM_PROMPT_TEXT;
+    const basePrompt = options.format === "markdown" ? DEFAULT_SYSTEM_PROMPT_MARKDOWN : DEFAULT_SYSTEM_PROMPT_TEXT;
+    // Add language instruction if specified
+    if (options.lang) {
+        const langName = getLanguageName(options.lang);
+        return `${basePrompt}\n- IMPORTANT: Write the output in ${langName}`;
+    }
+    return basePrompt;
 };

package/lib/index.d.ts

@@ -1,9 +1,12 @@
-export { processScript } from "./core/process.js";
-export { applyProfile } from "./core/variant.js";
-export { filterBySection, filterByTags, stripExtendedFields } from "./core/filter.js";
-export { listProfiles } from "./core/profiles.js";
-export { summarizeScript } from "./core/summarize/index.js";
+export { processScript } from "./core/preprocessing/process.js";
+export { applyProfile } from "./core/preprocessing/variant.js";
+export { filterBySection, filterByTags, stripExtendedFields } from "./core/preprocessing/filter.js";
+export { listProfiles } from "./core/preprocessing/profiles.js";
+export { summarizeScript } from "./core/ai/command/summarize/index.js";
+export { queryScript } from "./core/ai/command/query/index.js";
 export type { BeatVariant, BeatMeta, ExtendedBeat, ExtendedScript, OutputProfile, ProcessOptions, ProfileInfo } from "./types/index.js";
 export type { SummarizeOptions, SummarizeResult, LLMProvider, SummarizeFormat, ProviderConfig } from "./types/summarize.js";
+export type { QueryOptions, QueryResult } from "./types/query.js";
 export { beatVariantSchema, beatMetaSchema, extendedBeatSchema, extendedScriptSchema, outputProfileSchema } from "./types/index.js";
 export { summarizeOptionsSchema, llmProviderSchema, summarizeFormatSchema } from "./types/summarize.js";
+export { queryOptionsSchema } from "./types/query.js";

package/lib/index.js

@@ -1,10 +1,12 @@
-// Core API
-export { processScript } from "./core/process.js";
-export { applyProfile } from "./core/variant.js";
-export { filterBySection, filterByTags, stripExtendedFields } from "./core/filter.js";
-export { listProfiles } from "./core/profiles.js";
-// Summarize API
-export { summarizeScript } from "./core/summarize/index.js";
+// Preprocessing API
+export { processScript } from "./core/preprocessing/process.js";
+export { applyProfile } from "./core/preprocessing/variant.js";
+export { filterBySection, filterByTags, stripExtendedFields } from "./core/preprocessing/filter.js";
+export { listProfiles } from "./core/preprocessing/profiles.js";
+// AI API
+export { summarizeScript } from "./core/ai/command/summarize/index.js";
+export { queryScript } from "./core/ai/command/query/index.js";
 // Schemas (for validation)
 export { beatVariantSchema, beatMetaSchema, extendedBeatSchema, extendedScriptSchema, outputProfileSchema } from "./types/index.js";
 export { summarizeOptionsSchema, llmProviderSchema, summarizeFormatSchema } from "./types/summarize.js";
+export { queryOptionsSchema } from "./types/query.js";

package/lib/types/query.d.ts

@@ -0,0 +1,30 @@
+import { z } from "zod";
+/**
+ * Query Options - configuration for querying script content
+ */
+export declare const queryOptionsSchema: z.ZodObject<{
+    provider: z.ZodDefault<z.ZodEnum<{
+        openai: "openai";
+        anthropic: "anthropic";
+        groq: "groq";
+        gemini: "gemini";
+    }>>;
+    model: z.ZodOptional<z.ZodString>;
+    temperature: z.ZodOptional<z.ZodNumber>;
+    maxTokens: z.ZodOptional<z.ZodNumber>;
+    lang: z.ZodOptional<z.ZodString>;
+    systemPrompt: z.ZodOptional<z.ZodString>;
+    verbose: z.ZodDefault<z.ZodBoolean>;
+    section: z.ZodOptional<z.ZodString>;
+    tags: z.ZodOptional<z.ZodArray<z.ZodString>>;
+}, z.core.$strip>;
+export type QueryOptions = z.infer<typeof queryOptionsSchema>;
+/**
+ * Query Result - the generated answer
+ */
+export interface QueryResult {
+    answer: string;
+    question: string;
+    scriptTitle: string;
+    beatCount: number;
+}

package/lib/types/query.js

@@ -0,0 +1,21 @@
+import { z } from "zod";
+import { llmProviderSchema } from "./summarize.js";
+/**
+ * Query Options - configuration for querying script content
+ */
+export const queryOptionsSchema = z.object({
+    // LLM settings
+    provider: llmProviderSchema.default("openai"),
+    model: z.string().optional(),
+    temperature: z.number().min(0).max(2).optional(),
+    maxTokens: z.number().positive().optional(),
+    // Output language (e.g., "ja", "en", "fr")
+    lang: z.string().optional(),
+    // Custom prompt
+    systemPrompt: z.string().optional(),
+    // Processing options
+    verbose: z.boolean().default(false),
+    // Beat filtering
+    section: z.string().optional(),
+    tags: z.array(z.string()).optional(),
+});

package/lib/types/summarize.d.ts

@@ -34,6 +34,7 @@ export declare const summarizeOptionsSchema: z.ZodObject<{
         text: "text";
         markdown: "markdown";
     }>>;
+    lang: z.ZodOptional<z.ZodString>;
     targetLengthChars: z.ZodOptional<z.ZodNumber>;
     systemPrompt: z.ZodOptional<z.ZodString>;
     verbose: z.ZodDefault<z.ZodBoolean>;

package/lib/types/summarize.js

@@ -18,6 +18,8 @@ export const summarizeOptionsSchema = z.object({
     maxTokens: z.number().positive().optional(),
     // Output format
     format: summarizeFormatSchema.default("text"),
+    // Output language (e.g., "ja", "en", "fr")
+    lang: z.string().optional(),
     // Target length (optional)
     targetLengthChars: z.number().positive().optional(),
     // Custom prompt

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mulmocast-preprocessor",
-  "version": "0.1.1",
+  "version": "0.2.0",
   "description": "Preprocessor for MulmoScript",
   "type": "module",
   "main": "lib/index.js",