📄 PDF Q&A ChatBot

Upload a PDF and ask questions about its content using Gemini AI!
Built with Streamlit, this app uses LangChain, FAISS, and Google Generative AI to provide accurate answers based on your document.

---

🚀 Features

• 📚 Upload any PDF and ask context-aware questions
• 🧠 Powered by Google Gemini-Pro (via LangChain)
• 🧾 Custom prompt template support
• ⚡ Fast text extraction and vector search with FAISS
• 🐞 Debug Mode to view internal operations
• 🌐 Clean and responsive Streamlit UI

---

🛠️ Tech Stack

• Frontend: Streamlit
• AI Model: Gemini-Pro (Google Generative AI)
• Text Extraction: PyPDF2
• Vector Store: FAISS
• RAG Chain: LangChain
• Environment: Python, dotenv

---

🔧 Setup

1. Clone the Repository

git clone https://github.com/your-username/pdf-chatbot.git
cd pdf-chatbot

2. Create a Virtual Environment

# Linux/macOS
python -m venv venv
source venv/bin/activate

# Windows
python -m venv venv
venv\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

---

🧪 Run the App

Make sure your .env is configured (see below), then run:

streamlit run app.py

🔐 .env Configuration

Create a .env file in the root directory:

GOOGLE_API_KEY=your-google-api-key-here

✅ To get your API key, enable the Google Generative AI API and generate one in Google Cloud Console.

📁 Project Structure

pdf-chatbot/
├── app.py                 # Main Streamlit app
├── requirements.txt       # Python dependencies
├── .env.example           # Sample environment file
├── modules/
│   ├── pdf_loader.py      # Extracts text from PDF
│   ├── vector_store.py    # Embedding + FAISS index
│   └── rag_chain.py       # LangChain QA logic

📦 requirements.txt

streamlit 
langchain 
faiss-cpu
PyMuPDF
langchain-google-genai
langchain-huggingface
langchain-community 
langchain-text-splitters
tf-keras

💡 Usage

1. Upload a PDF via the interface.
2. Ask questions in natural language.
3. Get context-aware answers based on your document.
4. Use Debug Mode (sidebar) to view extracted text, vector store status, and response time.

---

🐞 Troubleshooting

• GOOGLE_API_KEY not set
  Make sure .env exists and contains your key.

• No response or timeouts
  Gemini may take a few seconds. Use Debug Mode to trace errors.

• FAISS import issues
  Run pip install faiss-cpu (Windows/Linux/macOS supported).

---

🧠 Example Custom Prompt

Customize your prompt in the sidebar:

You are a helpful assistant. Answer the question based on the following document context.

Context:
{context}

Question: {question}

---

🖼️ Screenshot

---

📜 License

MIT License. Use freely and contribute back if you like it! 🤝

---

💬 Feedback or Contributions

Open issues, suggest ideas, or submit pull requests! Let's build better AI tools together.

---