Docling Webapp

A web application for converting documents (PDF, DOCX, PPTX, XLSX, HTML, Markdown) to Markdown using Docling.

Features real-time conversion progress via SSE streaming, batch file support, configurable OCR engines, and a dark/light theme.

Stack

Layer	Technology	Version
Document conversion	Docling	2.76.0
API backend	FastAPI + Uvicorn	0.135.1 / 0.41.0
Python runtime	Python	3.12.13
Frontend framework	React	19.2.0
Build tool	Vite	7.3.1
TypeScript	TypeScript	5.9.3
CSS framework	Tailwind CSS	4.2.1
Reverse proxy	nginx	1.27-alpine
Node build image	Node.js	22-alpine
Backend base image	python	3.12-slim-bookworm

OCR Engines

Engine	Version	Notes
RapidOCR (ONNX)	1.4.4	Default — CPU-only, multilingual, fast
EasyOCR	1.7.2	GPU-capable, supports 80+ languages
Tesseract (via tesserocr)	2.10.0	Classic engine, requires language packs

OCR is auto-selected at runtime (RapidOCR by default). All three engines are pre-installed in the Docker image.

Supported Input Formats

PDF DOCX PPTX XLSX HTML MD — max 50 MB per file.

---

Quick Start (pre-built images)

Pull pre-built multi-arch images from DockerHub — no source code or build tools required.

Requirements

• Docker Engine 24+ with Compose v2
• ~4 GB disk space on first pull (Docling ML models are pre-baked in the image)
• No GPU required (CPU inference)
• Supported architectures: linux/amd64, linux/arm64

Deploy

docker compose up -d

The app is available at http://localhost:3000.

To use a different port:

PORT=8080 docker compose up -d

Note: The backend image is ~1.5 GB (Docling ML models pre-baked). First pull takes a few minutes depending on network speed. Subsequent starts are instant.

Update (existing users)

docker compose pull
docker compose up -d

Images: ricky75/docling-webapp-backend · ricky75/docling-webapp-frontend

---

Quick Start (build from source)

Requirements

• Docker Engine 24+ with Compose v2
• ~4 GB disk space (models are pre-downloaded into the image)
• No GPU required (CPU inference)

Build and run

docker compose -f docker-compose.build.yml up --build -d

The app is available at http://localhost:3000.

To use a different port:

PORT=8080 docker compose -f docker-compose.build.yml up --build -d

Development (hot reload)

docker compose -f docker-compose.build.yml -f docker-compose.dev.yml up --build

This mounts ./backend into the container and enables --reload on Uvicorn. The frontend is still served via nginx (no Vite dev server in this mode). Edit frontend/ files and run npm run build locally to see changes, or use the local dev setup below.

---

Local Development (no Docker)

Backend

cd backend
python -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate
pip install -r requirements.txt
uvicorn main:app --reload --port 8000

Note: First run will download Docling models (~1.5 GB) to ~/.cache/docling/models/.

Frontend

cd frontend
npm install
npm run dev

The Vite dev server runs at http://localhost:5173 and proxies /api/ to http://localhost:8000.

---

Platform Notes

macOS (Apple Silicon / Intel)

Works out of the box with Docker Desktop or OrbStack.

docker compose up -d

For local development without Docker, install Tesseract if you plan to use that OCR engine:

brew install tesseract tesseract-lang

Linux (x86_64 / arm64)

Works with Docker Engine and Docker Compose v2 (no Desktop needed).

# Install Docker Engine if not already present
curl -fsSL https://get.docker.com | sh

docker compose up -d

For Tesseract OCR in local dev (no Docker):

# Debian/Ubuntu
sudo apt-get install tesseract-ocr tesseract-ocr-eng tesseract-ocr-ita

# Fedora/RHEL
sudo dnf install tesseract tesseract-langpack-eng tesseract-langpack-ita

Windows

Use Docker Desktop with WSL 2 backend.

1. Install Docker Desktop
2. Enable WSL 2 integration in Docker Desktop settings
3. Open a WSL terminal and run:

docker compose up -d

For local development without Docker, use WSL 2 (Ubuntu recommended) and follow the Linux instructions above.

---

Configuration

Environment variables (apply to both docker-compose.yml and docker-compose.build.yml):

Variable	Default	Description
PORT	3000	Host port exposed by nginx
MAX_UPLOAD_SIZE_MB	50	Max upload size per file
MAX_CONCURRENT_JOBS	2	Max parallel conversion jobs

Pass them inline or in a .env file:

PORT=8080 MAX_CONCURRENT_JOBS=4 docker compose up -d

---

Project Structure

.
├── backend/              # FastAPI application
│   ├── main.py           # API endpoints (/convert, /jobs/:id/stream, /health)
│   ├── adapter.py        # Docling wrapper with async support
│   ├── config.py         # Allowed extensions, upload limits
│   └── requirements.txt
├── frontend/             # React + Vite application
│   └── src/
│       ├── App.tsx
│       ├── components/   # UploadZone, OptionsPanel, BatchList, ResultViewer…
│       └── hooks/        # useBatchUpload, useJobStream
├── Dockerfile            # Backend image (Python 3.12 + Docling models)
├── Dockerfile.frontend   # Frontend image (Node 22 build + nginx serve)
├── nginx.conf            # Reverse proxy with SSE-safe configuration
├── docker-compose.yml       # Default stack (pre-built DockerHub images)
├── docker-compose.build.yml # Build from source
└── docker-compose.dev.yml   # Development overrides (hot reload)