👽 RAG PDFBot - V3 (FastAPI + Streamlit)

This is the production-ready refactor of rag-bot-chroma, introducing a real separation between frontend (UI) and backend (logic) using Streamlit and FastAPI respectively. This modular architecture helps in scaling, extending, and deploying the bot in real-world environments.

---

🔗 Helpful Links

• 🧑‍💻 Version 1 - Basic RAG PDFBot (FAISS)

• ✍️ V1 Blog Walkthrough

• 🧑‍💻 Version 2 - Modular Streamlit + Chroma

• ✍️ V2 Blog Walkthrough

• 🧑‍💻 Version 3 - Streamlit + FastAPI

• ✍️ V3 Blog Walkthrough

---

🔄 What Changed from rag-bot-chroma

Feature	Version 2	Version 3
Codebase	One Streamlit app	Split into client/ + server/
PDF Upload	In Streamlit	Async FastAPI API
Chat	In Streamlit	Calls /chat API
Vectorstore	In UI	Controlled by backend
Model Options	Static	Dynamically fetched
Inspector	In sidebar	Main panel toggle
Splitting	RecursiveTextSplitter	TokenTextSplitter
UX	Crude	Responsive, clear, downloadable
Extendability	Hard	Easy to plug new LLMs, tools

---

🧪 How It Looks

Demo

---

🏗️ Architecture

---

🚀 Features

• 📁 Upload multiple PDFs and chat with them
• 🔌 Choose from Groq or Gemini as LLM providers
• 🔎 Query inspector for vectorstore transparency
• 🧠 RAG with LangChain + ChromaDB
• 📦 Streamlit frontend, FastAPI backend
• 🧪 Token-based chunking for LLM precision
• 💬 Downloadable chat history
• 🧰 Tools for reset, undo, clear
• 🌐 Fully API-driven interaction

---

🛠️ Tech Stack

• Frontend: Streamlit
• Backend: FastAPI
• LLMs: Groq & Gemini via LangChain
• Vector DB: ChromaDB
• Embeddings: HuggingFace & Google GenAI
• Chunking: TokenTextSplitter (was RecursiveCharacterTextSplitter)
• Parsing: PyPDF
• Orchestration: LangChain Retrieval Chain

---

📦 Installation

git clone https://github.com/Zlash65/rag-bot-fastapi.git
cd rag-bot-fastapi

Setup Virtual Environment:

python3 -m venv venv
source venv/bin/activate

Install frontend:

cd client
pip3 install -r requirements.txt

Install backend:

cd ../server
pip3 install -r requirements.txt

---

🔐 API Keys Required

• Groq API key from console.groq.com
• Google Gemini API key from ai.google.dev

Create a .env file:

GROQ_API_KEY=your-groq-key
GOOGLE_API_KEY=your-google-key

---

▶️ Run the Bot

Start FastAPI backend:

# Terminal 1
cd server
uvicorn main:app --reload

Start Streamlit frontend:

# Terminal 2
cd client
streamlit run app.py

---

📁 Project Structure

rag-bot-v3/
├── client/                         # Streamlit Frontend
│   ├── app.py                      # Main Streamlit entrypoint
│   ├── components/                 # UI modules
│   │   ├── chat.py
│   │   ├── inspector.py
│   │   └── sidebar.py
│   ├── state/
│   │   └── session.py              # Session state manager
│   ├── utils/
│   │   ├── api.py                  # Talks to backend
│   │   ├── config.py               # API_URL and config values
│   │   └── helpers.py              # API wrappers for frontend
│   ├── requirements.txt
│   └── README.md

├── server/                         # FastAPI Backend
│   ├── api/
│   │   ├── routes.py               # API endpoints
│   │   └── schemas.py              # Pydantic schemas for I/O
│   ├── core/
│   │   ├── document_processor.py   # Handles PDF validation and chunking
│   │   ├── llm_chain_factory.py    # Builds LLM chains and prompts
│   │   └── vector_database.py      # Embeddings + ChromaDB ops
│   ├── config/
│   │   └── settings.py             # App config, model provider setup
│   ├── utils/
│   │   └── logger.py               # Logging setup
│   ├── main.py                     # FastAPI app entrypoint
│   ├── requirements.txt
│   └── README.md

├── README.md                       # Root README (overview + instructions)
├── .gitignore

---

👓 Different Views

View	Description
💬 Chat	Renders chat bubbles, input box, and chat history download
🔬 Inspector	Renders inspector to test vectorstore responses

---

🧼 Tools Panel

Button	Function
🔄 Reset	Clears session state and reruns app
🧹 Clear Chat	Clears chat + PDF submission
↩️ Undo	Removes last question/response

---

📦 Download Chat History

Chat history is saved in the session state and can be exported as a CSV with the following columns:

Question	Answer	Model Provider	Model Name	PDF File	Timestamp
What is this PDF about?	This PDF explains...	Groq	llama3-70b-8192	file1.pdf, file2.pdf	2025-07-03 21:00:00

---

🙏 Acknowledgements

• LangChain
• Streamlit
• Groq
• Google Gemini
• Chroma

---

🧠 New to this Project?

Start from the basics: 👉 Version 1 - rag-bot-basic

Understand modular design: 👉 Version 2 - rag-bot-chroma

Then return here for real-world patterns.

---

Happy building! 🛠️