README.md

Codebase Embedding and Querying API

This project provides a RESTful API for embedding and querying codebases. It enables users to download a GitHub repository, process the codebase, store it in a vector database, and query it for relevant information.

---

Features

• Embed Codebase:

  • Downloads a GitHub repository using a URL.
  • Processes and chunks the codebase.
  • Stores the processed data in a vector database.

• Query Codebase:

  • Accepts a natural language query.
  • Searches the vector database for the most relevant code chunks.
  • Returns the results for further processing.

---

API Endpoints

1. Embed Codebase

• POST /api/embed-codebase
• Description: Downloads and processes a GitHub repository, embedding its code into a vector database.
• Request Body:

  {
    "repoUrl": "https://github.com/user/repository.git"
  }

• Response:

  {
    "message": "Repository downloaded and code collected."
  }

2. Query Codebase

• POST /api/query
• Description: Queries the vector database with a natural language query.
• Request Body:

  {
    "query": "What does the codebase do?",
    "folderName": "user_repository_timestamp"
  }

• Response:

  {
    "response": [
      /* relevant code chunks */
    ]
  }

---

Installation

1. Clone the Repository:

   git clone https://github.com/your-repo/codebase-embedding-api.git
   cd codebase-embedding-api

2. Install Dependencies:

   npm install

3. Build the Project:

   npm run build

4. Start Server (TypeScript):

   node dist/server.js

---

Usage

1. Start the server:

   npm run start

2. Use an API client (like Postman or cURL) to interact with the endpoints.

---

Requirements

• Node.js (v18+)
• NPM
• TypeScript

---

Folder Structure

src/
├── server.ts             # Main server file
├── utils.ts              # Utility functions
├── code_service.ts       # Code processing logic
├── chroma_db.ts          # Vector database functions
cloned_codebases/         # Directory for downloaded repositories

---