How alexzxu/local-rag works
This is a small document-ingestion and retrieval-augmented generation (RAG) application. Its ingestion workflow loads local files or webpages, filters already-known sources, splits new documents into chunks, and stores them in Chroma; its query workflow retrieves matching documents and prints their content and metadata. The README is empty, so the map reflects the sampled code. Web loading is explicit; embedding-provider calls during storage or retrieval are delegated to LangChain and are not shown in these excerpts.
Main entry points: main.py Workflow orchestrator · rag_vector_store.py Chroma setup · rag_loaders.py Source loaders · rag_document_tracker.py Source tracking
Components 12
Document ingestion
- Workflow orchestrator
main.py - Source loaders
rag_loaders.py - Document chunking
rag_parsing.py
Index and storage
- Source tracking
rag_document_tracker.py - Embedding setup
rag_embedding.py - Chroma setup
rag_vector_store.py - Chroma database
rag_vector_store.py
Query runtime
- Query retrieval
main.py - Printed matches
main.py
Other components
- User
- Documents and URLs
- Websites
Connections 16
- User → Workflow orchestrator: starts workflow
main.py - Documents and URLs → Source loaders: provides sources
rag_loaders.py - Workflow orchestrator → Source loaders: loads documents
main.py - Source loaders → Websites: fetches pages
rag_loaders.py - Workflow orchestrator → Source tracking: filters sources
main.py - Workflow orchestrator → Document chunking: splits new documents
main.py - Workflow orchestrator → Embedding setup: creates embeddings
main.py - Workflow orchestrator → Chroma setup: creates store
main.py - Workflow orchestrator → Source tracking: adds chunks
main.py - Source tracking → Chroma setup: checks and writes
rag_document_tracker.py - Chroma setup → Chroma database: persists documents
rag_vector_store.py - Workflow orchestrator → Query retrieval: invokes query
main.py - Query retrieval → Chroma database: retrieves matches
main.py - Query retrieval → Printed matches: prints matches
main.py - User → Query retrieval: submits query
main.py - Embedding setup → Chroma setup: configures embeddings
main.py