DocuMind API

English | 简体中文

DocuMind is a lightweight, high-performance RAG (Retrieval-Augmented Generation) engine built with FastAPI and LangChain. It enables users to upload documents (PDF/TXT/Word/Excel/MD) and ask questions based on their content, similar to ChatPDF.

🚀 Features

• Document Ingestion: Supports PDF, TXT, Word (.docx), Excel (.xlsx), and Markdown (.md) file uploads.
• Modern UI: Built-in Next.js frontend with file management and chat interface.
• RAG Pipeline: Automated text chunking, embedding, and vector storage.
• Smart Retrieval: Uses semantic search to find relevant context for answers.
• LLM Integration: Powered by OpenAI (or compatible) models for high-quality responses.
• Docker Ready: Fully containerized for easy deployment.
• WAF Bypass: Built-in mechanisms to handle API proxy restrictions (e.g., User-Agent spoofing).

🛠️ Tech Stack

Backend

• Framework: FastAPI
• RAG & Orchestration: LangChain
• Vector Database: ChromaDB (Persistent)
• Embeddings: OpenAI / Gemini (via compatible API)

Frontend

• Framework: Next.js (React)
• Styling: Tailwind CSS
• Components: Shadcn UI

📂 Project Structure

documind-api/
├── app/                     # Backend Source Code
│   ├── core/
│   │   ├── rag_engine.py    # Core RAG logic
│   │   └── __init__.py
│   ├── schemas/
│   │   ├── chat.py          # API Pydantic models
│   │   └── __init__.py
│   ├── main.py              # FastAPI entry point
│   └── ...
├── web/                     # Frontend Source Code (Next.js)
│   ├── app/
│   │   └── page.tsx         # Main chat interface
│   └── ...
├── chroma_db/               # Persisted vector database storage
├── temp_uploads/            # Temporary storage for file processing
├── .dockerignore
├── .env                     # Environment variables (GitIgnored)
├── Dockerfile
├── requirements.txt
└── README.md

⚡ Quick Start

Prerequisites

• Docker Desktop installed and running.
• An OpenAI API Key (or compatible provider key).
• Node.js 18+ (if running frontend locally without Docker).

🐳 Running with Docker Compose (Recommended)

This will start both the backend (API) and frontend (UI) in containers.

1. Configure Environment: Create a .env file in the root directory:

   # Global Config
   OPENAI_API_KEY=sk-your-api-key
   OPENAI_API_BASE=https://api.openai.com/v1

2. Start Services:

   docker-compose up -d --build

3. Access Application:

   • Frontend: http://localhost:3000
   • Backend API: http://localhost:8000/docs

🔧 Manual Installation (Dev Mode)

If you prefer running without Docker:

📚 API Usage

1. Upload Document

• Endpoint: POST /upload
• Supported Formats: PDF, TXT, DOCX, XLSX, MD
• Description: Upload a file to index its content.

2. Chat (Full Context)

• Endpoint: POST /chat
• Body:

  {
    "query": "What is the main topic of the document?"
  }

3. Query (Pure API)

• Endpoint: POST /api/query
• Description: Lightweight endpoint for 3rd-party integration (returns only answer and source paths).
• Body:

  {
    "query": "Summarize the contract.",
    "top_k": 3
  }

📄 License

This project is open-source and available under the MIT License.