Lumina PDF Logo

The Ultimate Privacy-First, Open-Core PDF Management and Automation Platform

GitHub stars License PRs Welcome

Lumina PDF is a premium, enterprise-grade, open-core PDF platform designed for complete data sovereignty and seamless document workflow orchestration. Run it as a highly responsive cross-platform desktop application, access it in the browser as a stateful modern React SPA, or deploy it on on-premise infrastructure as a highly scalable containerized service with an extensive private REST API.

Unlike cloud-based services, Lumina PDF performs document processing 100% locally or within your controlled infrastructure. There are no tracking scripts, no external analytics, and no remote document uploads.

• Desktop App (Tauri + Rust): Completely local, lightning-fast native experience with system JRE bundling, low memory footprints, and native PDF file association.
• Modern Web UI: A stunning, stateful React 18 Single Page Application styled with Mantine UI and TailwindCSS, boasting a rich dark mode, sleek micro-animations, and fluid transitions.
• Server Deployments: Optimized Spring Boot container images running on Java 21+ (Java 25 recommended), fully ready for deployment on GCP, AWS, Azure, or bare-metal Kubernetes.

• Chain Operations: Upload a PDF once and seamlessly pass it between tools (e.g., Split ➔ OCR ➔ Compress ➔ Sign) without losing context or reloading documents.
• 100GB+ Document Handling: Engineered with advanced memory management, client-side IndexedDB caching, Web Worker thumbnails, and explicit memory resource release to process massive PDF workloads in the browser without crashes.

• AI reasoning service: Built with FastAPI, Pydantic, and pydantic-ai to plan, reason, and interpret complex document tasks securely and locally.
• Structured Outputs: Native support for LLM-driven structured extraction, document understanding, and delegate tool orchestration.

• Security Mode: Enabled via DOCKER_ENABLE_SECURITY=true. Features multi-tenant logins, role-based access controls (RBAC), secure auditing logs, and single sign-on (SSO).
• Data Sovereignty: Built from the ground up for strict data protection compliance, fully air-gapped environment compatible.

Lumina PDF uses a clean, modern three-tier architecture separating presentation, heavy processing, and logical reasoning:

graph TD
 subgraph Client Layer
 Desktop["Tauri Desktop App (Rust + JRE)"]
 Browser["React 18 SPA (Mantine + Tailwind)"]
 end
 subgraph Service Proxy
 Backend["Spring Boot API Gateway (Java 21/25)"]
 end
 subgraph Core Processing Engines
 PDFBox["Apache PDFBox (Core Operations)"]
 LibreOffice["LibreOffice (Conversions)"]
 qpdf["qpdf (PDF Optimization)"]
 AIEngine["Python AI Engine (FastAPI + Pydantic-AI)"]
 end
 Client Layer -->|REST APIs / JSON| Backend
 Backend --> PDFBox
 Backend --> LibreOffice
 Backend --> qpdf
 Backend -->|Typed Contracts| AIEngine

Lumina PDF houses a comprehensive suite of PDF tools, organized into intuitive categories:

Launch the full-featured, local Lumina PDF server with a single command:

docker run -d -p 8080:8080 --name lumina-pdf ghcr.io/bennerdoo/luminapdf:latest

Once running, navigate to http://localhost:8080 in your web browser.

To enable enterprise user logins, role-based security, and access logs, deploy using the following Docker Compose configuration:

version: '3.8'
services:
 lumina-pdf:
 container_name: lumina-pdf-secure
 image: ghcr.io/bennerdoo/luminapdf:latest
 ports:
 - "8080:8080"
 environment:
 - DISABLE_ADDITIONAL_FEATURES=false
 - SECURITY_ENABLELOGIN=true
 - SYSTEM_DEFAULTLOCALE=en-GB
 - UI_APPNAME=Lumina PDF
 - SYSTEM_MAXFILESIZE=100
 volumes:
 - ./lumina/config:/configs:rw
 - ./lumina/logs:/logs:rw
 - ./lumina/tessdata:/usr/share/tessdata:rw
 restart: always

We provide three distinct Docker builds tailored for different environment requirements:

• Ultra-Lite (Dockerfile.ultra-lite): A highly stripped-down image containing only core PDFBox operations. Ideal for resource-constrained setups and microservice architectures.
• Standard (Dockerfile): The recommended image. Contains the Spring Boot core, Mantine frontend, qpdf, and standard system tools.
• Fat (Dockerfile.fat): Includes all of the above, plus pre-downloaded offline models (e.g., Tesseract OCR languages), heavy fonts, LibreOffice, and full offline dependencies for total air-gapped capability.

Lumina PDF uses Task as a unified command runner. Install Task, then use the following commands to speed up your local development.

1. Install All Dependencies:

   task install

2. Launch the Full Stack (Frontend + Backend + AI Engine):

   task dev:all

   • Frontend Dev Server: http://localhost:5173
   • Backend API: http://localhost:8080
   • AI Engine: http://localhost:5001

• Spring Boot Backend:
  • Start in dev mode: task backend:dev
  • Format code: task backend:format (Google Java Format)
• React Frontend:
  • Start Vite: task frontend:dev
  • Install frontend packages: task frontend:install
• Python AI Engine:
  • Start FastAPI engine: task engine:dev
  • Run static analysis & checks: task engine:check
• Tauri Desktop App:
  • Launch desktop developer window: task desktop:dev
  • Build production desktop package: task desktop:build

Ensure all components are strictly validated before making Pull Requests:

# Run backend, frontend, and engine tests in parallel
task test
# Full quality gate (Linter + TypeScript validation + Test suite)
task check
# Comprehensive integration test (Builds all Docker variants & executes Cucumber)
./test.sh

Lumina PDF is fully localized into over 40+ languages.

• Rule: All translation updates must be made in en-GB first (located in frontend/public/locales/en-GB/translation.toml).
• Translations are converted to JSON and loaded dynamically by the React SPA.
• Refer to the Language Guide for more details.

We love contributions! Please make sure to read our Developer Guide and Contributing Guidelines before writing code.

• Developer Guide — Comprehensive technical overview.
• Adding React Tools Guide — Step-by-step instructions on implementing new PDF tools.
• AI Engine Guidelines — Architecture and style rules for python services.

Lumina PDF is open-core. See the LICENSE file for details.