feat: solution for 'Повторный экзамен: FAQ-бот — ChromaDB + один MCP-tool'

2026-06-30 00:14:56 +03:00
parent f7c159346a
commit 1b9342d225
5 changed files with 225 additions and 68 deletions
@@ -1,21 +1,27 @@
-# FAQ Bot with ChromaDB and LangChain
+# FAQ Bot with ChromaDB and moderate-censor

-This project implements a simple FAQ bot that uses **ChromaDB** for vector storage and **LangChain** for building an intelligent agent. The bot can answer questions based on a small knowledge base stored in ChromaDB.
+This project implements an FAQ bot that uses **ChromaDB** for vector storage and retrieval, and **moderate-censor** as the single MCP-tool for content moderation.

 ## Features

- Stores documents in ChromaDB with embeddings from OpenAI.
- Uses the latest LangChain agent creation method (`initializeAgentExecutorWithOptions`).
- Simple CLI interface for interacting with the bot.
- Easy to extend with more documents or tools.
+- Vector-based FAQ retrieval using OpenAI embeddings and ChromaDB.
+- User input moderation with moderate-censor.
+- Simple HTTP API (`/ask`) to query the bot.
+
+## Prerequisites
+
+- Node.js v18+ (or any LTS version)
+- npm
+- OpenAI API key (set in `.env`)
+- ChromaDB server running locally (default path: `chromadb`)

 ## Setup

 1. **Clone the repository**

   ```bash
-   git clone https://github.com/your-username/faq-bot-chromadb.git
-   cd faq-bot-chromadb
+   git clone <repo-url>
+   cd <repo-directory>
   ```

 2. **Install dependencies**
@@ -24,26 +30,83 @@ This project implements a simple FAQ bot that uses **ChromaDB** for vector stora
   npm install
   ```

-3. **Configure environment**
-
-   Create a `.env` file in the project root:
+3. **Create a `.env` file**

   ```env
   OPENAI_API_KEY=your_openai_api_key
-   CHROMA_DB_PATH=./chromadb
+   PORT=3000
   ```

-4. **Run the bot**
+4. **Prepare FAQ data**
+
+   Create a `faq.json` file in the project root with the following format:
+
+   ```json
+   [
+     {
+       "question": "What is ChromaDB?",
+       "answer": "ChromaDB is a vector database for storing and retrieving embeddings."
+     },
+     {
+       "question": "How do I use the bot?",
+       "answer": "Send a POST request to /ask with a JSON body containing the 'question' field."
+     }
+   ]
+   ```
+
+5. **Ingest FAQ data**
+
+   ```bash
+   npm run ingest
+   ```
+
+   This will read `faq.json`, generate embeddings, and store them in ChromaDB.
+
+6. **Start the bot**

   ```bash
   npm start
   ```

-   You can also use `npm run dev` for automatic restarts with nodemon.
+   The server will listen on the port specified in `.env` (default 3000).

-## Adding Documents
+## Usage

-The bot comes with two sample FAQ entries. To add more, edit `src/index.js` or use the `addDocument` function from `src/vectorstore.js`.
+Send a POST request to `/ask`:
+
+```bash
+curl -X POST http://localhost:3000/ask \
+     -H "Content-Type: application/json" \
+     -d '{"question":"What is ChromaDB?"}'
+```
+
+Response:
+
+```json
+{
+  "answer": "ChromaDB is a vector database for storing and retrieving embeddings."
+}
+```
+
+If the question contains disallowed content, the bot will respond with a 403 status and reasons.
+
+## Project Structure
+
+```
+├── package.json
+├── src
+│   ├── index.js          # HTTP server and bot logic
+│   ├── ingest.js         # FAQ ingestion script
+│   └── middleware.js     # Moderation middleware
+├── faq.json              # FAQ data file
+└── README.md
+```
+
+## Notes
+
+- The bot uses the `text-embedding-ada-002` model for embeddings.
+- Only one MCP-tool (`moderate-censor`) is used as required.
+- Ensure the ChromaDB server is running before ingesting data or starting the bot.

 ## License

@@ -1,21 +1,17 @@
 {
-  "name": "faq-bot-chromadb",
+  "name": "faq-bot-chromadb-mcp",
  "version": "1.0.0",
-  "description": "FAQ bot using ChromaDB and LangChain",
+  "description": "FAQ bot using ChromaDB for vector storage and moderate-censor as the MCP-tool",
  "main": "src/index.js",
-  "type": "module",
  "scripts": {
    "start": "node src/index.js",
-    "dev": "nodemon src/index.js"
+    "ingest": "node src/ingest.js"
  },
  "dependencies": {
    "chromadb": "^0.3.0",
-    "langchain": "^0.2.0",
-    "langchain-community": "^0.2.0",
    "dotenv": "^16.4.5",
-    "openai": "^4.27.0"
-  },
-  "devDependencies": {
-    "nodemon": "^3.0.1"
+    "express": "^4.18.2",
+    "moderate-censor": "^1.0.0",
+    "openai": "^4.18.0"
  }
 }
@@ -1,50 +1,70 @@
-import dotenv from "dotenv";
-import readline from "readline";
-import { createAgent } from "./agent.js";
-import { addDocument } from "./vectorstore.js";
+require('dotenv').config();
+const express = require('express');
+const { OpenAI } = require('openai');
+const { ChromaClient } = require('chromadb');
+const { moderateInput } = require('./middleware');

-dotenv.config();
+const app = express();
+app.use(express.json());

-const COLLECTION = "faq_collection";
+const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });
+const chroma = new ChromaClient({ path: 'chromadb' });

-async function main() {
-  // Optional: add some sample documents
-  await addDocument(
-    COLLECTION,
-    "What is the return policy?",
-    { source: "FAQ" }
-  );
-  await addDocument(
-    COLLECTION,
-    "How can I track my order?",
-    { source: "FAQ" }
-  );
+const COLLECTION_NAME = 'faq_collection';
+const TOP_K = 3;

-  const agent = await createAgent(COLLECTION);
+// Initialize collection
+let collectionPromise = chroma.getOrCreateCollection({
+  name: COLLECTION_NAME,
+  metadata: { description: 'FAQ embeddings' }
+});

-  const rl = readline.createInterface({
-    input: process.stdin,
-    output: process.stdout,
-    prompt: "You: ",
-  });
-
-  console.log("FAQ Bot is ready. Type your question and press Enter.");
-  rl.prompt();
-
-  rl.on("line", async (line) => {
-    const question = line.trim();
+app.post('/ask', async (req, res) => {
+  try {
+    const { question } = req.body;
    if (!question) {
-      rl.prompt();
-      return;
+      return res.status(400).json({ error: 'Question is required' });
    }
-    try {
-      const result = await agent.call({ input: question });
-      console.log(`Bot: ${result.output}`);
-    } catch (err) {
-      console.error("Error:", err);
-    }
-    rl.prompt();
-  });
-}

-main().catch((err) => console.error(err));
+    // Moderate user input
+    const moderationResult = await moderateInput(question);
+    if (!moderationResult.allowed) {
+      return res.status(403).json({
+        error: 'Question contains disallowed content',
+        reasons: moderationResult.reasons
+      });
+    }
+
+    // Embed the question
+    const embeddingResponse = await openai.embeddings.create({
+      model: 'text-embedding-ada-002',
+      input: question
+    });
+    const embedding = embeddingResponse.data[0].embedding;
+
+    // Query ChromaDB
+    const collection = await collectionPromise;
+    const queryResult = await collection.query({
+      queryEmbeddings: [embedding],
+      nResults: TOP_K,
+      includeMetadata: true
+    });
+
+    if (!queryResult.ids || queryResult.ids.length === 0) {
+      return res.json({ answer: "I don't have an answer for that." });
+    }
+
+    // Pick the top result
+    const topAnswer = queryResult.metadatas[0]?.answer || "I don't have an answer for that.";
+
+    res.json({ answer: topAnswer });
+  } catch (err) {
+    console.error(err);
+    res.status(500).json({ error: 'Internal server error' });
+  }
+});
+
+const PORT = process.env.PORT || 3000;
+app.listen(PORT, () => {
+  console.log(`FAQ bot listening on port ${PORT}`);
+});
@@ -0,0 +1,55 @@
+require('dotenv').config();
+const fs = require('fs');
+const path = require('path');
+const { OpenAI } = require('openai');
+const { ChromaClient } = require('chromadb');
+
+const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });
+const chroma = new ChromaClient({ path: 'chromadb' });
+
+const COLLECTION_NAME = 'faq_collection';
+const FAQ_FILE = path.join(__dirname, '..', 'faq.json');
+
+async function ingest() {
+  try {
+    const rawData = fs.readFileSync(FAQ_FILE, 'utf-8');
+    const faqEntries = JSON.parse(rawData);
+
+    const collection = await chroma.getOrCreateCollection({
+      name: COLLECTION_NAME,
+      metadata: { description: 'FAQ embeddings' }
+    });
+
+    const documents = [];
+    const embeddings = [];
+    const ids = [];
+    const metadatas = [];
+
+    for (let i = 0; i < faqEntries.length; i++) {
+      const { question, answer } = faqEntries[i];
+      const embeddingResponse = await openai.embeddings.create({
+        model: 'text-embedding-ada-002',
+        input: question
+      });
+      const embedding = embeddingResponse.data[0].embedding;
+
+      documents.push(question);
+      embeddings.push(embedding);
+      ids.push(`faq-${i}`);
+      metadatas.push({ answer });
+    }
+
+    await collection.add({
+      documents,
+      embeddings,
+      ids,
+      metadatas
+    });
+
+    console.log(`Ingested ${faqEntries.length} FAQ entries into collection '${COLLECTION_NAME}'.`);
+  } catch (err) {
+    console.error('Error during ingestion:', err);
+  }
+}
+
+ingest();
@@ -0,0 +1,23 @@
+const moderate = require('moderate-censor');
+
+/**
+ * Moderates user input using moderate-censor.
+ * @param {string} text
+ * @returns {Promise<{allowed: boolean, reasons: string[]}>}
+ */
+async function moderateInput(text) {
+  try {
+    const result = await moderate.moderate(text);
+    if (result.isAllowed) {
+      return { allowed: true, reasons: [] };
+    } else {
+      return { allowed: false, reasons: result.reasons || [] };
+    }
+  } catch (err) {
+    console.error('Moderation error:', err);
+    // If moderation fails, default to allowing to avoid blocking legitimate queries
+    return { allowed: true, reasons: [] };
+  }
+}
+
+module.exports = { moderateInput };