wagari-endalew/doc-chat

Rag_Chat with doc/pdf that support amharic and english language # DocChat — Bilingual RAG Web App 🇪🇹 ### አማርኛ / English Document Intelligence A modern, minimal, production-ready document QA platform. **Stack:** FastAPI · Gemini 1.5 Flash · ChromaDB · Vanilla JS (no framework) --- ## Features | Feature | Detail | |---------|--------| | 📄 Multi-format | PDF, DOCX, TXT, PPTX | | 🌍 Bilingual | Amharic (አማርኛ) + English auto-detection | | ⚡ Streaming | Token-by-token streaming like ChatGPT | | 🧠 RAG Pipeline | Chunk → Embed → Vector Store → Retrieve → Generate | | 🔍 Semantic Search | Gemini text-embedding-004 | | 🗃 Vector DB | ChromaDB (in-memory / persistent) | | 🐳 Docker | One-command deployment | --- ## Architecture ``` Browser (HTML/CSS/JS) │ Upload files │ SSE stream ▼ ▼ ┌─────────────────────────────────────────┐ │ FastAPI Backend │ │ │ │ /upload /chat/stream │ │ │ │ │ │ Extract text Embed query │ │ Chunk (800 chars) ChromaDB search │ │ Embed (Gemini) Top-5 chunks │ │ ChromaDB.add() Gemini generate │ │ SSE yield │ └─────────────────────────────────────────┘ ``` --- ## Quick Start ### 1. Clone & Configure ```bash git clone github.com cd doc-chat ``` Set your Gemini API key: ```bash export GEMINI_API_KEY="your-key-here" ``` Or create `.env`: ``` GEMINI_API_KEY=your-key-here ``` ### 2a. Docker (Recommended) ```bash docker-compose up --build ``` - Frontend: localhost - Backend API: localhost - API Docs: localhost ### 2b. Local Dev **Backend:** ```bash cd backend pip install -r requirements.txt uvicorn main:app --reload --port 8000 ``` **Frontend:** ```bash # Any static server works cd frontend npx serve . -p 3000 # or: python -m http.server 3000 ``` --- ## API Endpoints | Method | Path | Description | |--------|------|-- …