How amirmahdimtz/multirag-clean works
MultiRAG Clean is a FastAPI service for dataset management, vector search, RAG systems, access control, and streamed chat. The principal workflow accepts a dataset through an API controller, delegates business operations, and uses configured persistence and processing providers; RAG endpoints search or answer questions against configured systems. The graph retains the documented chat, user-dataset, and provider capabilities. Some internals are represented by their exact source directories because their implementations were not sampled; those relationships are limited to documented layer flow.
Main entry points: dataset Dataset operations · dataset_controller.py Dataset APIs · repositories Repositories · rag_system RAG operations
Components 24
API and Host
- Web host
src/host/app.py - Web service
src/application/web.py - API-key security
src/application/security/api_key_dependencies.py - Dataset APIs
src/application/dataset/dataset_controller.py - User dataset APIs
src/application/dataset/user_dataset_controller.py - RAG system APIs
src/application/rag_system/rag_system_controller.py - RAG access APIs
src/application/rag_access/rag_access_controller.py - Chat APIs
src/application/chat/chat_controller.py - User APIs
src/application/user/user_controller.py
Core Use Cases
- Dataset operations
src/core/dataset - RAG operations
src/core/rag_system - Access operations
src/core/rag_access - Chat orchestration
src/core/chat/chat_business.py - User operations
src/core/user - LLM selection
src/core/llm
Domain Data
- Domain entities
src/domain/models
Infrastructure Providers
- SQL database
src/infrastructure/database/db_context.py - Repositories
src/infrastructure/repositories - Document processing
src/infrastructure/document_processing - File storage
src/infrastructure/storage/file_storage_service.py - Embedding providers
src/infrastructure/embedding - Vector stores
src/infrastructure/vector_store - Chat-model providers
src/infrastructure/llm
Other components
- API clients
Connections 28
- API clients → Web host: sends request
- Web host → Web service: starts
src/host/app.py - Web service → Dataset APIs: registers routes
- Web service → RAG system APIs: registers routes
- API clients → Dataset APIs: invokes dataset API
- API clients → RAG system APIs: invokes RAG API
- Dataset APIs → API-key security: requires admin key
src/application/dataset/dataset_controller.py - RAG system APIs → API-key security: requires admin key
src/application/rag_system/rag_system_controller.py - Dataset APIs → Dataset operations: delegates operations
src/application/dataset/dataset_controller.py - User dataset APIs → Dataset operations: delegates operations
src/application/dataset/user_dataset_controller.py - RAG system APIs → RAG operations: delegates queries
src/application/rag_system/rag_system_controller.py - RAG access APIs → Access operations: delegates access
src/application/rag_access/rag_access_controller.py - Chat APIs → Chat orchestration: delegates chat
src/application/chat/chat_controller.py - User APIs → User operations: delegates user tasks
src/application/user/user_controller.py - Dataset APIs → Document processing: processes documents
- Dataset operations → File storage: stores files
- Dataset operations → Embedding providers: requests embeddings
- Dataset operations → Vector stores: vectorizes and searches
- RAG operations → Vector stores: searches vectors
- RAG operations → LLM selection: generates answers
- Chat orchestration → LLM selection: selects chat model
- LLM selection → Chat-model providers: uses provider
- Dataset operations → Repositories: uses contracts
- RAG operations → Repositories: uses contracts
- Access operations → Repositories: uses contracts
- Repositories → SQL database: opens sessions
src/infrastructure/repositories/rag_access_repository.py - Repositories → Domain entities: reads and writes
src/infrastructure/repositories/rag_access_repository.py - Access operations → Domain entities: manages access records