How debasys/sharepoint-qna-system works
Project type and purpose
• This is a full-stack, microservices-style Q&A system built for SharePoint Online.
• Frontend: an SPFx (SharePoint Framework) React web part that provides the UI.
• Backend: a Node.js/TypeScript REST API that processes documents, generates semantic embeddings, enforces ACLs, and performs vector search.
• External services: Azure AD for authentication, SharePoint Online (Graph/REST APIs) for document retrieval and ACLs, OpenAI for embeddings, and Pinecone for vector storage.
File structure
• Top-level:
• README.md and config.example
• /backend – Express API in TS (index.ts, middleware/, routes/, services/, utils/)
• /spfx-webpart – React SPFx solution (WebPart.tsx, SCSS, manifest, types)
• /backend:
• middleware handles errors and SharePoint auth
• routes split between document processing and search
• services implement logic: document.service.ts (fetches & embeds), search.service.ts (queries Pinecone)
• /spfx-webpart:
• standard SPFx layout under src/webparts/qnaWebPart
Extract README insights
• High-level flow described under “Architecture”
• Key integrations: Azure AD, Pinecone, OpenAI, SharePoint ACL enforcement
• Microservice style with clearly separated responsibilities
Detail
• Middleware layer intercepting incoming requests for auth and error handling.