Mohamed-El/morocco-traffic-law-rag

--- title: Morocco Traffic Law RAG emoji: 🇲🇦 colorFrom: red colorTo: yellow sdk: streamlit sdk_version: 1.55.0 app_file: src/app.py pinned: false --- # Morocco Traffic Law RAG This project implements a Retrieval-Augmented Generation (RAG) system for the Moroccan traffic law. The system processes the official traffic law PDF, extracts structured legal articles, stores them in a SQLite database, builds a FAISS vector index, and answers user questions with citations to the relevant legal provisions. The project demonstrates how to build a production-style AI application combining document processing, semantic search, large language models, and containerized deployment. System Architecture Document ingestion (PDF) ↓ Text extraction and preprocessing ↓ Structured storage in SQLite (chapters → articles) ↓ Vector embeddings generation ↓ FAISS vector index construction ↓ Hybrid retrieval (vector search + keyword search) ↓ LLM answer generation (Gemini) ↓ FastAPI backend ↓ Streamlit web interface The entire system can be deployed using Docker. Key Features Structured extraction of Moroccan traffic law articles Hybrid retrieval using FAISS vector search and SQLite FTS keyword search Multilingual query support (Arabic and French) LLM-generated answers grounded in legal text Citation of relevant articles in responses FastAPI backend API Streamlit interactive interface Dockerized deployment Project Structure morocco-traffic-law-rag/ artifacts/ traffic_law.db law.index law_meta.json data/ 52_05_traffic_law.pdf docker/ Dockerfile.api Dockerfile.ui src/ api.py app.py build_db.py build_index.py chat.py ingest_pdf_ocr.py docker-compose.yml Makefile requirements.txt README.md .env.example Requirements The recommended way to run the project is using Docker. You will also need a Gemini API key. Create the environment file: cp .env.example .env Then edit .env and add your key: GEMINI_API_KEY=your_api_key_here Running with Docker Build and start the system: d …