Building a Document Management System

Every archive ends up the same way: files in a dozen places, three copies of "contract_final_v2_DEFINITIVE.pdf", and no way to find what you know you have.

This book shows you how to build the software that fixes that, from the first line of code to a production server. The project is DjangoVault, a real system the author uses to keep his own documents in order. It is not a toy example. The choices you will meet, including the wrong ones that were later corrected, are the ones actually made, and every chapter explains why.

What you will build:

• Upload, deduplication and a full version history: every change is a new version, nothing is overwritten
• Background processing with Celery and Redis, including OCR for scanned documents
• Keyword search and semantic search by meaning, powered by PostgreSQL and pgvector
• An assistant that answers questions about your archive and cites the documents behind each answer
• Automatic category suggestions, applied only when the system is very sure
• Sharing at three levels (read, edit, administer) and review with approvals
• Real-time updates, a watched folder, mail ingestion, and a phone app with document scanning
• Production deployment with Docker Compose, Nginx, private HTTPS and automatic backups

The approach is deliberate: pages rendered by the server and small touches of HTMX, no second project to maintain, and a system a single person can run. Confidential documents never reach an external AI service unless you consent, document by document.

Who it is for: Python developers who have already built at least one small Django application. You do not need to know Celery, Docker, Redis, Nginx, pgvector or HTMX: each tool is introduced at the point where the project needs it, and the book explains the problem it solves before showing how to use it.

How it is organized: 17 chapters in five parts (foundations, reading, ordering, ingestion, production) and five appendices: an operating guide, a glossary, a command reference, a chapter-to-project map, and a bonus guide to building a file server with remote access. Every chapter ends with a summary and exercises, and the code for each chapter is on GitHub, so you can always compare your work and restart from the right point.