How mawandu/assistant-web-educatif works
This is a full-stack educational assistant: students ask questions about uploaded course PDFs and receive answers with document/page citations; teachers and administrators manage access and documents. The main path runs from the React chat through the questions API, retrieval, and answer generation. A separate teacher upload path extracts and indexes PDF text. Authentication, feedback, and document metadata support these workflows. The README describes local Ollama, while the sampled answer-generation service uses Groq/OpenAI/Google; the exact runtime wiring for Ollama is not established by the excerpts.
Main entry points: documents.py Documents API · question_handler.py Question handler · users.py Users API · main.jsx Web routing
Components 27
Web interface
- Chat interface
frontend/src/App.jsx - Admin panel
frontend/src/Admin.jsx - Login page
frontend/src/pages/LoginPage.jsx - Auth context
frontend/src/context/AuthContext.jsx - Web routing
frontend/src/main.jsx
API and access
- API application
backend/main.py - Questions API
backend/api/routes/questions.py - Documents API
backend/api/routes/documents.py - Users API
backend/api/routes/users.py - Feedback API
backend/api/routes/feedback.py - Auth dependencies
backend/api/dependencies.py - Token security
backend/utils/security.py
Document knowledge
- PDF processing
backend/services/document_processor.py - Vector retrieval
backend/services/vector_store.py
Answer generation
- Question handler
backend/services/question_handler.py - Answer cache
backend/services/cache_manager.py - Answer generation
backend/services/llm_service.py - Ollama client
backend/services/ollama_client.py
Persistent state
- Document metadata
backend/models/document.py - User records
backend/models/user.py - Feedback records
backend/models/feedback.py - Question records
backend/models/question.py - PostgreSQL metadata
backend/config/database.py
Other components
- Student
- Teacher
- ChromaDB vectors
- LLM providers
Connections 36
- Student → Chat interface: asks questions
frontend/src/App.jsx - Teacher → Admin panel: manages content
README.md - Web routing → Chat interface: routes to
frontend/src/main.jsx - Web routing → Admin panel: routes to
frontend/src/main.jsx - Web routing → Login page: routes to
frontend/src/main.jsx - Web routing → Auth context: provides auth
frontend/src/main.jsx - Login page → Auth context: logs in through
frontend/src/pages/LoginPage.jsx - Chat interface → Questions API: posts questions
frontend/src/App.jsx - Chat interface → Feedback API: posts ratings
frontend/src/App.jsx - Auth context → Users API: requests token
frontend/src/context/AuthContext.jsx - API application → Questions API: registers route
backend/main.py - API application → Documents API: registers route
backend/main.py - API application → Users API: registers route
backend/main.py - API application → Feedback API: registers route
backend/main.py - Questions API → Question handler: delegates questions
backend/api/routes/questions.py - Documents API → PDF processing: extracts and splits
backend/api/routes/documents.py - Documents API → Vector retrieval: indexes chunks
backend/api/routes/documents.py - Documents API → Document metadata: creates metadata
backend/api/routes/documents.py - Documents API → Auth dependencies: checks access
backend/api/routes/documents.py - Users API → User records: manages accounts
backend/api/routes/users.py - Users API → Token security: uses password security
backend/api/routes/users.py - Feedback API → Feedback records: stores feedback
backend/api/routes/feedback.py - Questions API → Question records: uses question model
backend/api/routes/questions.py - Auth dependencies → User records: loads user
backend/api/dependencies.py - Auth dependencies → Token security: decodes token
backend/api/dependencies.py - Question handler → Answer cache: checks and saves
backend/services/question_handler.py - Question handler → Vector retrieval: retrieves chunks
backend/services/question_handler.py - Question handler → Answer generation: generates answer
backend/services/question_handler.py - Question handler → Document metadata: resolves citations
backend/services/question_handler.py - Answer generation → LLM providers: calls provider
backend/services/llm_service.py - Vector retrieval → ChromaDB vectors: reads and writes vectors
backend/services/vector_store.py - Documents API → PostgreSQL metadata: uses database session
- Users API → PostgreSQL metadata: uses database session
- Feedback API → PostgreSQL metadata: uses database session
- Question handler → PostgreSQL metadata: queries metadata
backend/services/question_handler.py - Ollama client → LLM providers: provider integration