A dictionary app built on wiktionary and opensource dictionaries.

Consists of a frontend in React and a simple python running an in memory duckdb database backend. Contains 1.5 million words and derivations.

Scripts

Sources of data:

• Raw data download in json format from wiktionary
• Opensource dictionary wordset

The raw data files are over 20GB and after converting to parquet and duckdb the same data turns into a 222 mb file.

Run parse-wiki.py to convert to parquest format. Run parse-wordset.py to convert to parquest format. Runread-db.py to read parquet files and insert into duckdb database.

Dev

cd frontent
pnpm/yarn/npm run dev

cd backend
python ./backend/server.py

cd backend
python ./backend/server.py

Production

It will download the database file upon running to keep the image small: 345MB

docker build -t dictionary:latest .
docker run -p 8000:8000 dictionary