Zebra668/Bano_qabil_Chatbot

bq_chatbot/
├── main.py                  # Flask app — entry point
├── requirements.txt         # Dependencies
├── .env                     # Your secrets (create from .env.example)
├── .env.example             # Template
├── bq_chatbot.db            # SQLite DB (auto-created)
├── knowledge/               # Put your RAG PDFs here
│   └── RAG.pdf
├── frontend/                # Optional: HTML frontend
│   └── index.html
└── backend/
    ├── config.py            # Environment variables
    ├── database.py          # SQLite setup + helpers
    ├── memory.py            # Conversation history
    ├── llm.py               # Gemini 2.5 Flash calls
    ├── rag.py               # PDF knowledge retrieval
    ├── tools.py             # Web search (DuckDuckGo)
    ├── relevance.py         # BQ/tech topic detection
    ├── multimodal.py        # Image analysis
    └── personalization.py   # User profile management

git clone <your-repo>
cd bq_chatbot
pip install -r requirements.txt

cp .env.example .env
# Edit .env and add your Gemini API key

GEMINI_API_KEY=your_actual_gemini_api_key_here
SECRET_KEY=banoqaabil2024
DB_PATH=bq_chatbot.db

# Put your Bano Qabil knowledge PDF in knowledge/ folder
cp your_bq_knowledge.pdf knowledge/RAG.pdf

python main.py
# Server starts at http://localhost:7860

{
  "message": "bano qabil mein registration kaise karein?",
  "session_id": 1,
  "user_id": 1
}

{
  "response": "Bano Qabil mein registration ke liye...",
  "session_id": 1,
  "sources_used": ["pdf_rag", "llm_knowledge"]
}

multipart/form-data
- file: image file
- question: "is mein kya course hai?"
- session_id: 1
- user_id: 1

{
  "status": "ok",
  "db": true,
  "rag": {"loaded": true, "chunks": 245}
}

{"session_id": 5}

# Procfile
web: python main.py

User Message
     ↓
Relevance Check (BQ/tech related?)
     ↓ Yes
[Source 1] PDF RAG → search knowledge base
[Source 2] Web Search → DuckDuckGo (time-sensitive queries)
     ↓
Combine context + user message
     ↓
Gemini 2.5 Flash (with system prompt + history)
     ↓
Response → Save to SQLite → Return to user