LangChain RAG Document Q&A Pack

Ask questions about your own documents — locally, with no API key needed by default. Point this pack at a folder of txt, markdown, or text-based PDF files, build an index in seconds, then ask questions and get an answer together with the exact source chunks it came from. The pipeline mirrors LangChain concepts (loaders, splitter, embeddings, vector store, chain) and ships a LangChain adapter, but the core runs without LangChain — numpy is the only hard dependency.

Exact contents: ask.py (index + ask CLI), rag_qa/loaders.py (.txt/.md/.pdf loaders), rag_qa/chunker.py (configurable size/overlap splitter), rag_qa/embeddings.py (built-in local hashing embedder plus an optional OpenAI hook), rag_qa/vectorstore.py (numpy cosine-similarity store, persists to .npz, uses FAISS automatically if installed), rag_qa/chain.py (extractive answers by default, composed LLM answers with your own key), rag_qa/langchain_adapter.py, sample_docs/ (3 short docs: photosynthesis, warehouse safety, coffee brewing), tests/smoke_test.py, README.md, requirements.txt and LICENSE.txt.

Instant download — your file is available right after checkout from your account. No shipping, no waiting.
What's included

• ask.py CLI — index a docs folder, then ask questions with source chunks printed
• Loaders for .txt, .md and text-based .pdf files (pypdf when installed, built-in minimal extractor otherwise)
• Configurable chunker (chunk size and overlap)
• Built-in local hashing embedder — deterministic, no API key, no downloads — plus an optional OpenAI embeddings hook
• Numpy cosine-similarity vector store persisted to .npz, with automatic FAISS use when installed
• LangChain adapter wiring the same components through LangChain when installed
• 3 sample documents on different topics and a smoke test that verifies retrieval (passes: 3/3 queries return the expected source doc)
• README with setup, usage, and how to swap embedders
• Requires: Python 3.10+ and numpy (only hard dependency); optional pypdf, openai, faiss-cpu and langchain for the extra modes
• Instant download after checkout — no shipping
• Quality-checked by MarketplacePlanet before listing
• Free re-downloads for 30 days from your account

Free sample: Download the free sample before you buy.