How premchand-eidiko/llamaindex works
The repository’s main product is an Enterprise RAG Assistant: users upload documents, which are loaded, parsed, enriched with metadata, and indexed; they then query or chat against that knowledge base and receive answers with sources. Its API also documents hybrid retrieval, streaming, and GraphRAG capabilities. The tree contains a separate, simpler Day-5 backend/frontend and earlier Day-1–4 learning examples; those are not merged into the main runtime flow. Edges reflect shown calls or explicit module references; unsampled subsystems are included where documented, but uncertain wiring is omitted.
Main entry points: main.py API application · query_routes.py Query API · llm.py LLM provider factory · storage.py Index persistence
Components 30
API and sessions
- API application
Day-5/rag_project/backend/app/main.py - Upload API
Day-5/rag_project/backend/app/api/ingestion_routes.py - Query API
Day-5/rag_project/backend/app/api/query_routes.py - Chat API
Day-5/rag_project/backend/app/api/chat_routes.py - Graph API
Day-5/rag_project/backend/app/api/graph_routes.py - Admin API
Day-5/rag_project/backend/app/api/admin_routes.py - Request models
Day-5/rag_project/backend/app/schemas/models.py
Document ingestion
- Ingestion pipeline
Day-5/rag_project/backend/app/ingestion/pipeline.py - Document loaders
Day-5/rag_project/backend/app/ingestion/loaders.py - Document parsers
Day-5/rag_project/backend/app/ingestion/parsers.py - Metadata extraction
Day-5/rag_project/backend/app/ingestion/metadata.py
Retrieval and query
- Retriever manager
Day-5/rag_project/backend/app/retrieval/retriever.py - Metadata filters
Day-5/rag_project/backend/app/retrieval/filters.py - Result reranker
Day-5/rag_project/backend/app/retrieval/rerank.py - Hybrid search
Day-5/rag_project/backend/app/retrieval/hybrid.py - Query streaming
Day-5/rag_project/backend/app/query/streaming.py - Multi-document query
Day-5/rag_project/backend/app/query/multi_doc.py
Indexes and graph
- Vector index
Day-5/rag_project/backend/app/indexes/vector.py - Index persistence
Day-5/rag_project/backend/app/indexes/storage.py - Knowledge graph
Day-5/rag_project/backend/app/graph/knowledge_graph.py
LLM and configuration
- LLM provider factory
Day-5/rag_project/backend/app/core/llm.py - Runtime settings
Day-5/rag_project/backend/app/core/config.py
Alternate Day 5 app
- Alternate backend API
Day-5/project/backend/app/api/documents.py - Alternate ingestion
Day-5/project/backend/app/services/ingestion.py - Alternate RAG service
Day-5/project/backend/app/services/rag.py - Conversation storage
Day-5/project/backend/app/services/chat_history.py - Alternate chat API
Day-5/project/backend/app/api/chat.py
Other components
- User
- Groq service
- OpenAI service
Connections 36
- User → API application: sends requests
- API application → Upload API: registers
Day-5/rag_project/backend/app/main.py - API application → Query API: registers
Day-5/rag_project/backend/app/main.py - API application → Chat API: registers
Day-5/rag_project/backend/app/main.py - API application → Graph API: registers
Day-5/rag_project/backend/app/main.py - API application → Admin API: registers
Day-5/rag_project/backend/app/main.py - Upload API → Ingestion pipeline: ingests files
Day-5/rag_project/backend/app/api/ingestion_routes.py - Upload API → Vector index: creates index
Day-5/rag_project/backend/app/api/ingestion_routes.py - Upload API → Index persistence: saves index
Day-5/rag_project/backend/app/api/ingestion_routes.py - Ingestion pipeline → Document loaders: loads documents
Day-5/rag_project/backend/app/ingestion/pipeline.py - Ingestion pipeline → Document parsers: parses documents
Day-5/rag_project/backend/app/ingestion/pipeline.py - Ingestion pipeline → Metadata extraction: extracts metadata
Day-5/rag_project/backend/app/ingestion/pipeline.py - Query API → Index persistence: loads index
Day-5/rag_project/backend/app/api/query_routes.py - Query API → Metadata filters: builds filters
Day-5/rag_project/backend/app/api/query_routes.py - Query API → Retriever manager: creates retriever
Day-5/rag_project/backend/app/api/query_routes.py - Query API → Result reranker: reranks results
Day-5/rag_project/backend/app/api/query_routes.py - Query API → Hybrid search: handles hybrid search
Day-5/rag_project/backend/app/api/query_routes.py - Query API → Query streaming: streams tokens
Day-5/rag_project/backend/app/api/query_routes.py - Query API → LLM provider factory: gets LLM
Day-5/rag_project/backend/app/api/query_routes.py - Index persistence → Vector index: coordinates storage
Day-5/rag_project/backend/app/indexes/storage.py - Chat API → Request models: validates requests
Day-5/rag_project/backend/app/api/chat_routes.py - Chat API → Index persistence: loads index
Day-5/rag_project/backend/app/api/chat_routes.py - Chat API → LLM provider factory: logs via
- Chat API → Conversation storage: persists history
- Graph API → Knowledge graph: uses graph
Day-5/rag_project/backend/app/api/graph_routes.py - Admin API → Index persistence: manages index
Day-5/rag_project/backend/app/api/admin_routes.py - LLM provider factory → Runtime settings: reads settings
Day-5/rag_project/backend/app/core/llm.py - LLM provider factory → Groq service: selects provider
Day-5/rag_project/backend/app/core/llm.py - LLM provider factory → OpenAI service: falls back
Day-5/rag_project/backend/app/core/llm.py - API application → Index persistence: checks index
Day-5/rag_project/backend/app/main.py - API application → LLM provider factory: initializes
Day-5/rag_project/backend/app/main.py - Alternate backend API → Alternate ingestion: ingests documents
Day-5/project/backend/app/api/documents.py - Alternate backend API → Alternate RAG service: uses RAG service
Day-5/project/backend/app/api/documents.py - Alternate backend API → Conversation storage: manages history
Day-5/project/backend/app/api/documents.py - Alternate chat API → Conversation storage: uses history
Day-5/project/backend/app/api/chat.py - Alternate chat API → Alternate RAG service: queries RAG
Day-5/project/backend/app/api/chat.py