OCR Service

A web-based OCR (Optical Character Recognition) service that extracts text from images and saves results to multiple destinations: local Markdown files, Notion, and Apple Notes.

Features

• Upload up to 20 images at once (JPEG, PNG, WebP, GIF)
• Extract text using Tesseract OCR with multi-language support (default: Italian + English)
• Save extracted text to:
  • Local files — timestamped Markdown files
  • Notion — creates pages in a Notion database
  • Apple Notes — macOS only, via AppleScript
• Clean web UI with drag-and-drop, image previews, and copy-to-clipboard

Requirements

• Node.js 18+ and Yarn, or Docker

Note: Apple Notes integration is macOS-only and not available inside Docker.

Setup

1. Install dependencies

yarn install

2. Configure environment

cp .env.example .env

Edit .env with your settings (see Configuration).

3. Start the server

# Development (watch mode)
yarn dev

# Production
yarn build && yarn start

The server runs on http://localhost:3089 by default.

Docker

1. Configure environment

cp .env.example .env
# edit .env with your NOTION_TOKEN etc.

2. Build and run

docker compose up --build

The app will be available at http://localhost:3089. The output/ directory is mounted as a volume so extracted Markdown files persist on the host. All variables from .env are passed into the container automatically via env_file.

Without Compose:

docker build -t ocr-service .
docker run -p 3089:3089 --env-file .env -v $(pwd)/output:/app/output ocr-service

Stopping:

docker compose down

Apple Notes output requires macOS + AppleScript and is not supported inside Docker.

Configuration

All options are set via environment variables in .env:

Variable	Default	Description
PORT	3089	Server port
TESSERACT_LANG	ita+eng	OCR language(s). Combine with + (e.g. ita+eng+fra). See available languages.
OUTPUT_DIR	./output	Directory for local Markdown output files
NOTION_TOKEN	—	Notion integration token (from notion.so/my-integrations)
NOTION_DATABASE_ID	—	ID of the target Notion database

Adding Tesseract language models

The repo includes eng.traineddata and ita.traineddata. To add more languages, download the corresponding .traineddata file from the Tesseract tessdata repository and place it in the project root.

Notion setup

1. Create a Notion integration at notion.so/my-integrations and copy the token.
2. Share the target database with your integration.
3. Copy the database ID from its URL: notion.so/{workspace}/{DATABASE_ID}?v=...
4. Set NOTION_TOKEN and NOTION_DATABASE_ID in .env.

The database must have a property named Name (title type).

API

POST /extract

Extracts text from uploaded images and saves to the selected destinations.

Request (multipart/form-data):

Field	Type	Description
images	File[]	Images to process (max 20, max 20 MB each)
title	string (optional)	Note title. Defaults to OCR – {date}
outputs	string | string[]	Destinations: file, notion, notes

Response (JSON):

{
  "title": "My note",
  "text": "Extracted text content...",
  "results": {
    "filePath": "/path/to/output/2024-01-01T12-00-00-my-note.md",
    "notion": "ok",
    "notes": "ok"
  },
  "errors": {
    "file": null,
    "notion": null,
    "notes": null
  }
}

Project Structure

ocr-service/
├── src/
│   ├── server.ts          # Express server and API routes
│   ├── ocr.ts             # Tesseract OCR wrapper
│   ├── public/
│   │   ├── index.html     # Markup
│   │   ├── style.css      # Styles
│   │   └── app.js         # UI logic
│   └── outputs/
│       ├── file.ts        # Local Markdown output
│       ├── notion.ts      # Notion API output
│       └── notes.ts       # Apple Notes output (macOS)
├── output/                # Default local output directory
├── tmp/                   # Temporary upload directory
├── eng.traineddata        # Tesseract English language model
├── ita.traineddata        # Tesseract Italian language model
├── .env.example           # Environment variable template
├── Dockerfile
├── docker-compose.yml
└── tsconfig.json

Scripts

Command	Description
yarn dev	Start in watch mode (TypeScript, no build step)
yarn build	Compile TypeScript to dist/
yarn start	Run compiled production build