feat: solution for 'Агент с RAG-памятью'

src/index.ts

@@ -1,27 +1,80 @@
-import express from 'express';
-import dotenv from 'dotenv';
-import bodyParser from 'body-parser';
-import { VirtualFileSystem } from './vfs';
-import { LLM } from './llm';
-import { Agent } from './agent';
-import { createRoutes } from './routes';
-import { errorHandler } from './middleware';
+import { OpenAI } from "langchain/llms/openai";
+import { OpenAIEmbeddings } from "langchain/embeddings/openai";
+import { FAISS } from "langchain/vectorstores/faiss";
+import { RetrievalQA } from "langchain/chains/retrieval-qa";
+import { Tool } from "langchain/tools/base";
+import { initializeAgentExecutorWithOptions } from "langchain/agents";
+import { RecursiveCharacterTextSplitter } from "langchain/text_splitter";
+import * as dotenv from "dotenv";
 
 dotenv.config();
 
-const app = express();
-const port = process.env.PORT ? parseInt(process.env.PORT, 10) : 3000;
+async function main() {
+  // Ensure API key is set
+  const apiKey = process.env.OPENAI_API_KEY;
+  if (!apiKey) {
+    console.error("Error: OPENAI_API_KEY environment variable is not set.");
+    process.exit(1);
+  }
 
-app.use(bodyParser.json());
+  // Sample document
+  const sampleText = `
+  The quick brown fox jumps over the lazy dog. This sentence is often used to test typing and fonts.
+  The capital of France is Paris. Paris is known for its art, gastronomy, and culture.
+  The Earth revolves around the Sun every 365.25 days. The Moon orbits the Earth approximately every 27.3 days.
+  `;
 
-const vfs = new VirtualFileSystem();
-const llm = new LLM();
-const agent = new Agent(vfs, llm);
+  // Text splitter configuration (chunk size 1000, overlap 200)
+  const splitter = new RecursiveCharacterTextSplitter({
+    chunkSize: 1000,
+    chunkOverlap: 200,
+  });
 
-app.use('/', createRoutes(agent));
+  // Split the document into chunks
+  const docs = await splitter.splitText(sampleText);
 
-app.use(errorHandler);
+  // Initialize embeddings and vector store
+  const embeddings = new OpenAIEmbeddings({ openAIApiKey: apiKey });
+  const vectorStore = await FAISS.fromTexts(docs, [], embeddings);
 
-app.listen(port, () => {
-  console.log(`RAG Memory Agent listening on port ${port}`);
+  // Initialize LLM
+  const llm = new OpenAI({
+    openAIApiKey: apiKey,
+    temperature: 0,
+  });
+
+  // Create RetrievalQA chain
+  const qaChain = RetrievalQA.fromLLM(llm, vectorStore);
+
+  // Define a tool that uses the QA chain
+  const ragTool = new Tool({
+    name: "RAG",
+    description: "Answer questions based on the provided documents using Retrieval-Augmented Generation.",
+    func: async (input: string) => {
+      const result = await qaChain.invoke({ query: input });
+      return result.output as string;
+    },
+  });
+
+  // Initialize the agent with the updated method
+  const agent = await initializeAgentExecutorWithOptions(
+    [ragTool],
+    llm,
+    {
+      agentType: "zero-shot-react-description",
+      verbose: true,
+    }
+  );
+
+  // Run a sample query
+  const query = "What is the capital of France?";
+  const response = await agent.invoke({ input: query });
+
+  console.log("\n=== Agent Response ===");
+  console.log(response.output);
+}
+
+main().catch((err) => {
+  console.error("Error in main execution:", err);
+  process.exit(1);
 });