How sanju1098/rag-chatbot works
AstraMind is a shared-library PDF question-answering app. Users upload PDFs for extraction, chunking, embedding, and storage; they then ask questions in chat, where retrieval selects relevant passages before Gemini streams an answer. The upload and retrieval paths are supported by the README and source-index references; unsampled internals are represented only at documented stages. The graph also includes document listing/deletion and the landing-page entry point, but omits presentation-only and legal pages.
Main entry points: actions.ts Upload actions · search.ts Vector retrieval · route.ts Chat API · embeddings.ts Embedding generation
Components 14
Web experience
- Landing page
src/app/page.tsx - Upload and library
src/app/upload/page.tsx - Chat interface
src/app/chat/page.tsx
PDF ingestion
- Upload actions
src/app/upload/actions.ts - PDF extraction
src/lib/pdf-utils.ts - Text chunking
src/lib/chunking.ts
Chat and retrieval
- Chat API
src/app/api/chat/route.ts - Vector retrieval
src/lib/search.ts
Persistence and AI
- Embedding generation
src/lib/embeddings.ts - Database connection
src/lib/db-config.ts - Document schema
src/lib/db-schema.ts - Neon vector store
Other components
- User
- Google Gemini
Connections 17
- User → Landing page: opens
src/app/page.tsx - User → Upload and library: uploads PDF
README.md - User → Chat interface: asks questions
README.md - Upload and library → Upload actions: submits and manages
src/app/upload/page.tsx - Upload actions → PDF extraction: extracts text
src/app/upload/actions.ts - Upload actions → Text chunking: chunks text
src/app/upload/actions.ts - Upload actions → Embedding generation: embeds chunks
src/app/upload/actions.ts - Upload actions → Neon vector store: stores documents
README.md - Upload actions → Document schema: uses schema
src/app/upload/actions.ts - Embedding generation → Google Gemini: requests embeddings
README.md - Chat interface → Chat API: sends chat
README.md - Chat API → Vector retrieval: retrieves passages
src/app/api/chat/route.ts - Vector retrieval → Embedding generation: embeds query
src/lib/search.ts - Vector retrieval → Neon vector store: matches vectors
README.md - Vector retrieval → Document schema: queries documents
src/lib/search.ts - Chat API → Google Gemini: streams answer
README.md - Database connection → Neon vector store: connects
README.md